reproductions supplied by edrs are the best that can be made · christopher e. smith, phd kaarin...

DOCUMENT RESUME

ED 439 580 EC 307 812

AUTHOR Carr, Edward G.; Horner, Robert H.; Turnbull, Ann P.;Marquis, Janet G.; McLaughlin, Darlene Magito; McAtee,Michelle L.; Smith, Christopher E.; Ryan, Kaarin Anderson;Ruef, Michael B.; Doolabh, Ajit; Braddock, David, Ed.

TITLE Positive Behavior Support for People with DevelopmentalDisabilities: A Research Synthesis.

INSTITUTION American Association on Mental Retardation, Washington, DC.SPONS AGENCY Special Education Programs (ED/OSERS), Washington, DC.ISBN ISBN-0-940898-60-8PUB DATE 1999-00-00NOTE 119p.; Portions of'this paper were presented at the annual

meeting of the American Association on Mental Retardation,New York, May 1997.

CONTRACT H023E50001AVAILABLE FROM American Association on Mental Retardation, 444 North

Capitol St., NW, Suite 846, Washington, DC 20001-1512.PUB TYPE Books (010) Information Analyses (070) Reports

Research (143)EDRS PRICE MF01/PC05 Plus Postage.DESCRIPTORS Behavior Change; *Behavior Disorders; *Behavior

Modification; Classroom Techniques; Demography; ElementarySecondary Education; *Environmental Influences;Intervention; Outcom,3s of Treatment; *PositiveReinforcement; Program Evaluation; *Research Utilization;Theory Practice Relation hip

IDENTIFIERS *Functional Behavioral Assessment; *Positive BehavioralSupport

ABSTRACTThis book, prepared in response to a request from the United

States Department of Education, Office of Special Education Programs, reviewsthe published literature on positive behavior interventions and uses thisdatabase to provide four main content areas for research. Positive behaviorsupport (PBS) is defined as an approach for dealing with problem behaviorthat focuses on the remediation of deficient contexts (such as environmentalconditions and/or behavioral repertoires) that by functional assessment aredocumented to be the source of the problem. The research published on PBSbetween 1985 and 1996 (n =107 articles) was reviewed with respect to fourcategories of variables: demographics, assessment practices, interventionstrategies, and outcomes. Results indicated that: (1) PBS is widelyapplicable to people with serious problem behavior; (2) the field is growingrapidly overall, but especially in the use of assessment and in interventionsthat focus on correcting environmental deficiencies; (3) using stringentcriteria of success, PBS is effective in reducing problem behavior inone-half to two-thirds of cases; (4) success rates nearly double whenintervention is based on a prior functional assessment; and (5) consumerneeds that emphasize comprehensive lifestyle support, long-term change,practicality and relevance, and direct support for consumers themselves areinadequately addressed by the research base. Recommendations are made forbridging the research-to-practice gap. (Contains more than 300 references.)(Author/CR)

Reproductions supplied by EDRS are the best that can be madefrom the original document.

r

POSITIVE

BE VIOR PORT

FOR PEOP ITH

DEVE P

DIS ILITIES

A EARCH S HESIS

by Edward G. CarrRobert H. HomerAnn P. Turnbulland colleagues

David BraddockEditor, Research Monographs and Books

2A A

A

U.S. DEPARTMENT OF EDUCATIONNATIONAL INSTITUTE OF EDUCATION

EDUCATIONAL RESOURCES INFORMATIONCENTER (ERIC)

This document has been reproduced asreceived from the person or organizationoriginating it

ID Minor changes have been made to improvereproduction quality

Points of view or opinions stated in this docu-ment do not necessanly represent official NIEposition or policy

Positive Behavior Supportfor People With

Developmental Disabilities:

A Research SynthesisBy

Edward G. Carr, PhDState University of New York at Stony Brook and

Developmental Disabilities Institute

Robert H. Homer, PhDUniversity of Oregon

Ann P. Turnbull, EdDJanet G. Marquis, PhD

University of Kansas

Darlene Magito McLaughlin, MAMichelle L. McAtee, MA

Christopher E. Smith, PhDKaarin Anderson Ryan, MA

State University of New York at Stony Brook

Michael B. Ruef, PhDUniversity of Kansas

Ajit Doolabh, MAState University of New York at Stony Brook

David Braddock, PhDEditor, Research Monographs and Books

ANVIRAmerican Association on Mental Retardation

3

©1999 American Association on Mental Retardation

Published byAmerican Association on Mental Retardation444 North Capitol Street, NW, Suite 846Washington, DC 20001-1512

The points of view herein are those of the authors and do not necessarily representthe official policy or opinion of the American Association on Mental Retardation.Publication does not imply endorsement by the editor, the Association, or itsindividual members.

Printed in the United States of America.

Library of Congress Cataloging-in-Publication DataPositive behavior support for people with developmental disabilities:

a research synthesis/by Edward G. Carr... [et. al.].p. cm.

Includes bibliographical references.ISBN 0-940898-60-81. Mentally handicappedBehavior modification.

1. Carr, Edward G., 1947-RC451.4.M47P67 1999616.85'884dc21 99-22505

CIP

Preparation of this manuscript was supported in part by Grant #11023E50001 from theOffice of Special Education Programs,"Synthesizing and Communicatinga ProfessionalKnowledge Base on Positive Behavioral Support" The opinions expressed herein do notnecessarily reflect those of the U.S. Department of Education.

Portions of this paper were presented at the annual meeting of the American Associationon Mental Retardation, New York, May 1997.

411

TABLE OF CONTENTS

LIST OF TABLES vi

LIST OF FIGURES vii

ACKNOWLEDGMENTS ix

FOREWORD BY TODD R. RISLEY, PHD xi

FOREWORD BY MADELEINE WILL xv

OVERVIEW 1

CHAPTER 1 INTRODUCTION 3

Positive Behavioral Support: Overview and Definition 3

Need for a Review of Positive Behavior Support 6

Research Questions Posed: Contributions of the Review 7

Elaboration of Defining Characteristics of PBS 7

Research Questions 8

How Widely Applicable Is PBS? 8

In What Ways Is the Field Evolving? 8

How Effective Is PBS? 9

What Factors Modulate the Effectiveness of PBS? 9

How Responsive Is the PBS Literature to the Needs of Consumers(Nonresearchers)? 9

The Structure of the Research Synthesis 9

CHAPTER 2 METHODS 11

Operational Definitions 11

Demographics 11

Assessment Practices 11

Summary 12

Intervention Strategies 12

Intervention Categories 12

Systems Change 15

Ecological Validity 16

Summary 17

Outcome Measures 18

Positive Behavior 18

Problem Behavior 18

Stimulus Generalization 18

Response Generalization 18

Maintenance 19

Lifestyle Change 19

Social Validity 19

5 iii

Summary 20Literature Search and Eligibility Criteria 20

Initial Selection Criteria 20Literature Search 21

Exclusion Criteria 21

Data Collection Methods 24How Intervention Effects Were Measured 24How Reliability Was Measured 26

CHAPTER 3 RESULTS 27How Widely Applicable Is PBS? 27In What Ways Is the Field Evolving? 27

Size of the Database 29Demographics 29Assessment 35Intervention 38Outcomes 41

How Effective Is PBS? 43Changes in Positive Behavior 43Distribution of Outcome Effectiveness for PBS Interventions 44Success Rates for PBS Interventions Pooled Across Outcomes 45Stimulus and Response Generalization 47Maintenance 48Lifestyle Change 48Social Validity 48

What Factors Modulate the Effectiveness of PBS? 50Influence of Demographic Variables 50Influence of Assessment Variables 53Systems Change 53Ecological Validity 54Medication 54

CHAPTER 4 DISCUSSION 57Potential Biases in the Retrieved Literature 57

Rigor Versus Relevance? 57Rigor Over Reliance? 58Is Rigor Over Relevance the Only Choice? 59

External Validity of Excluded Studies 59Drawing Inferences From the Results 61

iv 6



General Status of PBS 61

Demographics 62

Assessment 62

Intervention 64

Outcomes 66


Changes in Positive Behavior 66

Distribution of Outcome Effectiveness: What Constitutes a Success? 67

Success Rates Pooled Across Outcomes 67

Stimulus Generalization, Response Generalization, and Maintenance 70

Lifestyle Change and Social Validity 70


Demographic Variables 71

Influence of Assessment Variables 72

Systems Change 73

Ecological Validity 74

Implications for Future Research 74

How Responsive Is the PBS Literature to the Needs of Consumers(Nonresearchers)? 75

Comprehensive Lifestyle Support 75

Long-Term Change 76

Practicality and Relevance 76

Consumers Want Support, Too 78

CHAPTER 5 SUMMARY 81





How Responsive Is the PBS Literature to the Needsof Consumers (Nonresearchers)? 82

CHAPTER 6 RECOMMENDATIONS 85

For Researchers 85

For Service Providers 85

For Social Policy Advocates 86

For the Government 86

REFERENCES 87

v

TABLES

Table 1. Journals examined and number of articles included and excluded 22Table 2. Criteria used to exclude articles from the database 23Table 3. Relationship between outcomes and participants 25Table 4. Characteristics of participants, intervention agents, and settings 28Table 5. Distribution of outcome effectiveness 44Table 6. Success rates for PBS interventions pooled across outcomes 45Table 7. Types of non-PBS interventions used 46Table 8. Generalization measures of outcome effectiveness 47Table 9. Maintenance measures of outcome effectiveness 48Table 10. Relationship between demographic variables and

outcome effectiveness 51

Table 11. Relationship between assessment variables andoutcome effectiveness 52

Table 12. Relationship between systems-change variables andoutcome effectiveness 54

Table 13. Relationship between ecological validity and outcome effectiveness 54

8

vi

FIGURES

Figure 1. Percentage of total articles, participants, and outcomesin each 4-year block of time 29

Figure 2. Percentage of total participants by genderin each 4-year block of time 30

Figure 3. Percentage of total participants by age in each 4-year block of time 30

Figure 4. Percentage of total participants by diagnosisin each 4-year block of time 31

Figure 5. Percentage of total participants by level of retardationin each 4-year block of time 32

Figure 6. Percentage of total participants by type of problem behavior

in each 4-year block of time 32

Figure 7. Percentage of total participants by type of intervention agentin each 4-year block of time 33

Figure 8. Percentage of total participants by type ofintervention setting in each 4-year block of time 34

Figure 9. Percentage of total outcomes by prior assessmentin each 4-year block of time 34

Figure 10. Percentage of total outcomes by type of assessmentin each 4-year block of time 35

Figure 11. Percentage of total outcomes by type of motivationin each 4-year block of time 37

Figure 12. Percentage of total outcomes by type of interventionin each 4-year block of time 38

Figure 13. Percentage of total outcomes by type of systems change


Figure 14. Percentage of total outcomes by intervention in all relevant contexts


Figure 15. Percentage of total outcomes by type of generalizationin each 4-year block of time 42

Figure 16. Percentage of total outcomes by type of socialvalidation


Figure 17. Number of positive behavior outcomes by changes in frequency and

changes in percentages 43

9 vii

ACKNOWLEDGMENTS

We thank Doug Anderson, Mike Cataldo, Harris Cooper, Esther Lerner,

Denise Poston, Wayne Sailor, Martha Snell,and Mark Wolery for their

helpful comments. We especially thank Ellen Schiller of the Office of

Special Education Programs for her ongoing support and feedback.

Address all correspondence to Edward Carr, Dept. of Psychology,

State University of New York, Stony Brook, NY 11794-2500.

ix

FOREWORD

Positive Behavioral Support and Applied Behavior Analysis

Todd R. Risley, PhD

University of Alaska

The dimensions of Applied Behavior Analysis were developed by Montrose

Wolf and his colleagues in a remarkable series of exploratory studies, across a

variety of children and problems, at the University of Washington in 1962-64.

Those studies modeled how to arrange interventions in clinics, playgrounds,

classrooms, and homes. They demonstrated the speed and power to change

behavior of procedures based on the principles of behavior as delineated by

B. F. Skinner. And they developed measurement tactics and experimental

design variations to fit each new real-world condition and problem behavior

encountered (see Risley, 1997 for a listing of these studies). It is important to

note the three unprecedented and audacious features of those pioneering

studies: 1) deliberate interventions in the daily lives of people; 2) fast and large

behavior change; and 3) scientific documentation of field research. With the

institution of the Journal of Applied BehaviorAnalysis in 1968, each of these

features has been narrowed and codified across succeeding generations of

researchers.

Thirty years later this monograph"Positive Behavior Support for People

With Developmental Disabilities: A Research Synthesis"has reviewed the last

10 years of a major branch of Applied Behavior Analysis and has recommended

some changes in direction. Their review is carefully designed and conducted

this is not biased assertion, but an objective meta-analysis with well defined,

reliable variables. (Its clarity and sophistication are such that I regard it as a

model of logic and methodology for Graduate research courses.) Their

findings are presented with cautionthis is not polemic or salesmanship, but

reasoned consideration. (In fact, I think it understates the evidence favoring

Positive Behavioral Support strategies.) Their recommendations are wise and

practicalthis is not pie-in-the-sky wishful thinking but a useful guide to the

next generation of research, service, and policy. (I am gratified by this evidence

of wisdom, leadership, and cooperation in the third and fourth generations of

applied behavioral researchers.)

In their recommendations they state that we need a new applied science.

Thirty-five years ago Montrose Wolf and a few others said the sameandproceeded to found Applied Behavior Analysis with only a little experience

from which to deduce its characteristics. In contrast, these 10 authors repre-

sent scores of other researchers and have examined hundreds of studies. Their

basis for recommending the dimensions of future Positive Behavioral Support

is simply a conservative extrapolation of information.

11xi

xii

Some of their recommendations to researchers suggest only a moresystematic approach to things currently done. "Consumer needs" and "Partici-patory Action Research" is an (admittedly large) extension in the effort andtiming of "Social Validity." If functional assessment is to have ecologicalvalidity it must be done throughout life, and therefore,"hypotheses" of antece-dent and consequence functions can often be tested by close monitoring duringinterventions. Goals of life improvement and long-term benefits are nowfrequently (if poorly) addressed in anecdotal reports supplementing primarydata.

However, there are fundamental attributes of Applied Behavior Analysiswith which their recommendations conflict: procedural specificity, anddemonstration of causality. Are these attributes of Applied Behavior Analysisfundamental to the advancement of knowledge? Perhaps not as much as weassume.

The recommendations for multicomponent interventions, individuallyadapted to circumstances and revised over time, is contrary to the goals ofspecifiable treatment "packages" or "manualized" treatments or "model"treatment programs common in Applied Behavior Analysis. Some specificity ofintervention is, of course, desirable but to what use has it been put in 30 yearsof behavioral research?

This monograph, itself, is a most rare example of careful comparison ofprocedures across studies. More commonlywe have quite similar procedureswith different "proprietary" labels, and common labels masking large differ-ences between intervention programs. Other areas of science and technologythat have different customs for analyzing procedures across researchers shouldbe examined for guidance; as should field trial protocols which specifytreatments by their adherence to a set of decision rules rather than a set offixed procedures.

The recommendations for larger interventions, goals, and measures cannotbe met while adhering to high requirements of experimental manipulation.Despite occasional articles demonstrating or advocating the usefulness ofgroup designs, correlations, naturalistic observations, and case studies in theJournal of Applied Behavior Analysis, demonstrations of causality over timehave remained the standard for participation in Applied Behavior Analysis. It isthis doctrine that most biases applied behavior analysis research toward moremanageable contexts and problems. The demonstration of causality betweenindependent and dependent variables through repeated manipulation ofindependent variables across time defined the Experimental Analysis ofBehavior. The founders of Applied Behavior Analysis were active participantsin the Experimental Analysis of Behavior journal and meetings. They neverconsidered any other research logic. Some of the founders went so far as todefine Applied Behavior Analysis as the use of single-subject time-seriesdesigns. Others disagreed and considered any research design to be simply atool, to be chosen and used when needed.

12

It is clear that Positive Behavioral Support researchers need to use other

tools in addition to experimental manipulations. Will Applied Behavior

Analysis accommodate this, or will there become two distinct groupsone,allied to the Experimental Analysis of Behavior, and the other, dedicated toempirical problem-solving? It should be noted that empirical problem solving

was the original impetus for the studies on which Applied Behavior Analysis

was founded: problems were addressed as they presented themselves; measure-

ment and research design considerations were overlaid to achieve the best

information the circumstances allowed; the resulting publications were

carefully considered reports of the problem-solving process, with due attention

to threats to internal and external validitynot experiments proving a point.

A healthy dose of Donald Campbell's consideration of problem-solving

research (1957) and reforms as experiments (1969) would help the enterprise

of Positive Behavioral Support pursue the recommendations proposed in this

monograph and to return to the roots of Applied Behavior Analysis.

Campbell, D. T. (1957). Factors relevant to the validity of experiments in social settings.

Psychological Bulletin, 54, 297-312.

Campbell, D. T. (1969). Reforms as experiments. American Psychologist, 24,409-29.

Risley, T. R. (1997). Montrose M. Wolf: The origin of the dimensions of Applied Behavior

Analysis, Journal of Applied Behavior Analysis, 30, 377-381.

13

FOREWORD

Madeleine Will, Former Assistant Secretary, OSERS

U.S. Department of Education

There are occasions when science confirms aspirations, when research justifies

conviction. The analysis by Carr et al. is one of those.

It is a matter of record that during the time when I served as Assistant

Secretary of Education, Office of Special Education and Rehabilitative Services(1983-1989), professionals, parents of children and adults with disabilities, self-

advocates, and policy makers were locked in fiercedebate concerning four

matters on which I had chosen to take a leadership role: (a) inclusion of

students with disabilities in regular education, (b) transition of students with

disabilities from high school to work, (c) withholding or withdrawal of

efficacious medical treatment from newborns who had obvious disabilities,

and (d) the use of aversive behavioral interventions with children and adults

with disabilities. In each of these issues policy decisions were made based on

both technology and hope, on science and aspiration. Ineach case there was a

capacity to do the right thing. More than that, there was a hope shared by

myself and many others, that policy eventually would dignify the lives of people

with disabilities by granting them new rights. This was especially true in the

area of behavior support.

As I read the present monograph on Positive Behavior Support I could not

help but recall the focus of our policy struggle in the late 1980s. Our efforts to

match our aspirations with existing science were hindered, as is often the case,

by the absence of science about the use of positive behavioral interventions.

The monograph you hold before you is in many ways the document we wanted

a decade ago. Here is a careful analysis documenting that positive behavioral

procedures can produce important change in the behavior and lives of people

with disabilities. Positive interventions can be effective.

Simply put: the monograph objectively reviews the published research on

positive behavioral interventions and draws conclusions from this database.

The organization and presentation of the published research is of special

importance because it allows each reader to consider the conclusions in light of

their personal interpretation of the data.

What I myself take from this review is that positive behavioral interven-

tions are indeed beneficial; that they complement other practices that are now

codified into policy (inclusion, transition to real work, and access to medical

treatment); and that positive interventions are justified now on scientific

grounds, just as some of us thought, long ago and still,that they are com-

manded on humane, moral, and constitutional grounds. Here, at last, is the

14 xv

proof positive of their efficacy. My son, Jonathan Will, and many other citizenswith disabilities now have available to them (should they need it) a promisingscience that it was my privilege to stimulate and support. In a very practicalway this monograph defines the science that confirms our aspiration; theresearch that justifies our convictions.

xvi

OVERVIEW

OVERVIEW

This monograph was prepared inresponse to a request from the UnitedStates Department of Education, Office ofSpecial Education Programs, for a reviewof the literature on positive behaviorsupport that provided (a) a definition ofthe approach, (b) an analysis of thedatabase, (c) a delineation of gaps in ourknowledge, and (d) suggestions for futuredirections. Because of the scope andcomplexity of the positive behaviorsupport approach, it will be useful tobegin with an overview of the maincontent areas of our monograph.

Positive behavior support (PBS) is anapproach for dealing with problembehavior that focuses on the remediationof deficient contexts (i.e., environmentalconditions and/or behavioral repertoires)that by functional assessment aredocumented to be the source of theproblem. The research published on PBSbetween 1985 and 1996 was reviewedwith respect to four categories of vari-ables: demographics, assessmentpractices, intervention strategies, andoutcomes.

The data derived from examiningthese four categories of variables were

analyzed to answer five questions:(a) How widely applicable is PBS? (b) Inwhat ways is the field evolving? (c) Howeffective is PBS? (d) What factorsmodulate the effectiveness of PBS?(e) How responsive is the PBS literature tothe needs of consumers?

Results indicated that (a) PBS iswidely applicable to people with seriousproblem behavior; (b) the field is growingrapidly overall, but especially in the use ofassessment and in interventions thatfocus on correcting environmentaldeficiencies; (c) using stringent criteria ofsuccess, PBS is effective in reducingproblem behavior in one half to twothirds of the cases; (d) success ratesnearly double when intervention is basedon a prior functional assessment; and(e) consumer needs that emphasizecomprehensive lifestyle support, long-term change, practicality and relevance,and direct support for the consumersthemselves are inadequately addressed bythe research base.

Recommendations are made forbridging the research-to-practice gap inthe areas of research priorities, serviceprovision, social policy, and governmentalaction.

16

INTRODUCTION

CHAPTER 1

INTRODUCTIONA new era of behavioral support isemerging. It is building from the careful,rigorous science of the past three decadesand the practical demands of families,teachers, and clinicians. The newapproach, called positive behavioralsupport (Koegel, Koegel, & Dunlap, 1996),

is evolving rapidly as new informationand challenges arise. The volume ofrecent empirical work suggests the needfor a synthesis of the current researchliterature. The purpose of this researchsynthesis is to (a) define positive behav-ioral support, (b) provide a systematicanalysis of the existing research databasewith respect to positive behavioralsupport, (c) compare that database withcurrent clinical needs in the field, and(d) suggest future directions for research,practice, and policy. Our focus is on theapplication of behavioral support forpeople with developmental disabilitiesand autism.

We provide an introduction to theresearch synthesis by presenting (a) abrief overview and definition of positivebehavior support, (b) the rationale for thepresent research synthesis of positivebehavior support, (c) the researchquestions posed, and (d) the structure ofthe research synthesis.

Positive BehavioralSupport: Overview and

DefinitionProblem behaviors such as aggression,self-injury, tantrums, and propertydestruction have long been barriers tosuccessful education, socialization,

employment, and community adaptation(Meyer, Peck, & Brown, 1991;Scheerenberger, 1990; White, Lakin,Bruininks, & Li, 1991). The goal ofpositive behavior support (PBS) is toapply behavioral principles in thecommunity in order to reduce problembehaviors and build appropriate behav-iors that result in durable change and arich lifestyle. The foundation of PBS liesin early efforts to apply principles ofbehavior to improve the lives of childrenwith severe problem behaviors (Bijou &Baer, 1961; Bijou, Peterson, & Ault, 1968;Browning & Stover, 1971). The lawfulrelationships between behavior andenvironment were applied to people withreal problems. The results were encourag-ing and led to the development in the1960s and 1970s of an array of interven-tion procedures (Barrett, 1986; Foxx,1982; Kazdin, 1980). Each of theseprocedures proved successful at reducingproblem behaviors in some situations andunsuccessful in others.

The need to improve the applicationof intervention procedures led to a renewedappreciation for the earlier call to organizeinterventions based on a careful functionalassessment of the problem behavior(Baer, Wolf, & Risley, 1968; Bijou et al.,1968). Beginning in the late 1970s (Carr,1977) and continuing through today, theconcept of functional assessment hasbeen transformed into a practicaltechnology for guiding the developmentof behavioral interventions (Bailey &Pyles, 1989; Gardner & Sovner, 1994;O'Neill, Horner, Albin, Storey, & Sprague,1997a & b; Reichle &Wacker, 1993).

3

CHAPTER 1

A central result has been an expan-sion of interventions beyond those basedon consequence manipulations to include(a) altering the environment beforeproblem behaviors occur, and(b) teaching appropriate behaviors as aneffective strategy for reducing unwantedbehaviors. Behavioral support is becom-ing less a process of selecting an interven-tion, and more the construction ofacomprehensive set of procedures thatinclude change of the environment tomake problem behaviors irrelevant,instruction on appropriate behaviors thatmakes the problem behavior inefficient,and manipulation of consequences toensure that appropriate behaviors aremore consistently and powerfullyreinforced than are problem behaviors. Asthe structure of behavioral supportexpands, so does recognition that acomplete technology will also requireattention to those interacting variables ina setting/system that affect the imple-mentation of effective procedures (Sailor,1996; Taylor-Greene et al., 1997).

As effective approaches to behavioralsupport emerged, expectations for theoutcomes and acceptability of thetechnology were redefined. The initialfocus of behavioral interventions was onsimple reduction of problem behaviors.As this proved possible, yet insufficient,expectations changed. Through reexami-nation of original assumptions (Baer eta1.,1968) and attention to the messagesprovided by real-world users of thetechnology, investigators expandedfunctional assessment technology and theexpectations for behavioral interventionswere redefined. Effective behavioralsupport needed not only to reduceproblem behaviors, but also to buildprosocial behavior, document durablechange, generalize across the full range of

4 18

situations an individual encountered, andproduce access to a rich lifestyle.

This is the key concept that definesPBS: To remediate problem behavior, it isnecessary first to remediate deficientcontexts. There are two kinds of deficien-cies: those relating to environmentalconditions, and those relating to behaviorrepertoires. Environmental conditions aredeficient to the extent that they involvelack of choice, inadequate teachingstrategies, minimal access to engagingmaterials and activities, poorly selecteddaily routines, and a host of otherproximal and distal antecedent stimulirelated to the previous factors. Behaviorrepertoires are deficient to the extent thatcommunication skills, self-management,social skills, and other constructivebehaviors are inadequately developed orabsent. (The constituent elements relatingto the two kinds of deficiencies justsummarized are defined at length inChapter 2, under the subhead Interven-tion Strategies.) Recently there has beenmuch discussion concerning the strongassociation between these two types ofdeficiencies and the display of problembehavior; many have concluded thatproblem behavior can be effectivelyaddressed by focusing on the assessmentand remediation of context (Emerson,McGill, & Mansell, 1994; Koegel et al.,1996; Luiselli & Cameron, 1998; Lutzker& Campbell, 1994; Reichle &Wacker,1993).

As noted, the primary focus of thefield has not always been on context perse. For many years, researchers andclinicians alike emphasized strategies thatconcentrated on the problem behavioritself (e.g., aggression, self-injury, andproperty destruction), rather than ondeficiencies such as poor environmentalconditions or a lack of functional skills.

INTRODUCTION

This emphasis led to the development of awide variety of reactive (often punitive)interventions designed to suppress oreliminate the problem behavior directly(Bucher & Lovaas, 1968; Cataldo, 1991;Harris & Ersner-Hershfield, 1978; Matson& DiLorenzo, 1984). The relative merits ofreactive versus contextually basedinterventions have been hotly debated inthe literature (Guess, Helmstetter,Turnbull, & Knowlton, 1987; Repp &Singh, 1990). Yet people on all sides of thisdebate have always acknowledged that theultimate goal is not simply the reductionof problem behavior, but rather improv-ing people's lives. There is now wide-spread appreciation of the fact that thenegative sequelae of problem behaviorinclude not only physical danger to selfand others, but also educational segrega-tion, limited employment opportunities,rejection by members of the community,separation from home and family, and,finally, social ostracism and a life withoutfriendship (National Institutes of Health,1991). Clearly, behavioral technologymust now become more comprehensivein scope and more cognizant of the role ofthe larger systems (e.g., family, school,employment, and funding) that influencethe practicality of effective behavioralsupport (Sailor, 1996). Consideration ofthese larger issues has prompted interestin the kinds of environmental andbehavioral context variables that are theessential characteristics of PBS.

It may be worthwhile at this point tofurther distinguish PBS from othercontemporary approaches. The two mostsubstantive and frequently employedalternatives to PBS are the use of pharma-cotherapy (medication) and aversiveprocedures. As noted previously, the mainfocus of aversive procedures is theelimination, through punishment, of

problem behavior. Such procedures arefundamentally and by definition reactivein nature. They are not employed until theproblem behavior occurs. In contrast, PBSis proactive in nature. It is an attempt toremediate environmental and behavioraldeficiencies so as to prevent futureoccurrences of the problem behavior. Insum, using aversive procedures conformsbest to a crisis management paradigm;using PBS conforms best to a preventionparadigm. A detailed analysis of theaversives literature is ably presentedelsewhere (Cataldo, 1991).

The second non-PBS strategy,medication, involves the use of one ormore drugs to suppress problem behavior.Often the drugs are administered over along period of time, to address hypoth-esized or identified biochemical aberra-tions thought to underlie problembehavior. In this case, the use of medica-tion conforms to a preventive paradigm,because the successful use of medicationwould block the occurrence of futureepisodes of problem behavior. Sophisti-cated analyses of the voluminous andcomplex pharmacotherapy literature arealso available elsewhere (Reiss & Amu',1998; Schaal & Hackenberg, 1994;Schroeder & Tessel, 1994; Thompson,Hackenberg, & Schaal, 1991). There is anobvious distinction between PBS andpharmacotherapy: Whereas the formerfocuses on the role of environmentalfactors in assessing and remediatingproblem behavior, the latter focuses onthe role of biochemical factors.

Some people question the advisabil-ity of comparing PBS with medicationand the use of aversives. Our position,articulated in detail later, is that there aremany good reviews of non-PBS strategiesbut a dearth of reviews concerning theunique contributions of PBS per se.

19 5

CHAPTER 1

Therefore, we will focus our review onPBS alone. More important, makingcomparisons among these approacheswould implicitly support what we believeis a false assumption: that one mustalways choose among the three ap-proaches because only one can beascendant. Clinical experience wouldseem to contradict this assumption. Forexample, for many years, it has beenconsidered a best practice to accompanythe use of aversives with a detailededucational and social support plan thatembodies the major features of PBS (Carr& Lovaas, 1983; Foxx, 1982, 1990; Foxx,Bittle, & Faw, 1989). Likewise, theliterature on dual diagnosis indicates apotentially important role for medication.Specifically, some people with develop-mental disabilities may receive a secon-dary diagnosis such as depression,anxiety, bipolar disorder, or obsessive-compulsive disorder (Bodfish & Madison,1993; Lowry & Sovner, 1992; Ratey,Sovner, Parks, & Rogentine, 1991; Reiss &Rojahn, 1993). An emerging literaturesuggests that medication given toalleviate the symptoms associated withthe secondary diagnosis may help reduceproblem behavior (e.g., Bodfish &Madison,1993; Sovner, 1989). As yet,there is no definitive research demon-strating a causal link between obsessive-compulsive disorder, for example, andproblem behavior. Nonetheless, theempirical work on dual diagnosis soundsa cautionary note in that practitionersmust consider the possibility that PBSalone might prove insufficient for dealingwith individuals carrying diagnosticlabels beyond developmental disabilitiesper se. Similarly, medication alone mightprove insufficient for improving anindividual's lifestyle, one of the statedgoals of PBS.

6 20

In sum, the three approaches for dealingwith problem behavior do not existindependently of one another inpractice. As yet, however, there is nowell-developed research literature thatexplores, systematically, how theapproaches interrelate.

Need for a Review ofPositive Behavior SupportTo date, there have been few synthesisreviews focused on PBS per se. Thisapproach did not gain momentum untilthe mid-to-late 1980s, after which therewas an explosive growth in the number ofresearch studies, conceptual papers, andintervention manuals related to PBS (Carret al., 1994; Donnellan, LaVigna, Negri-Shoultz, & Fassbender, 1988; Durand,1990; Evans & Meyer, 1985; Horner,Dunlap et al., 1990; LaVigna & Donnellan,1986; Meyer & Evans, 1989; Smith, 1990).There have been many excellent reviewsdealing with the general issue ofremediating problem behavior. Typically,however, reviewers combined the analysisof PBS with the analysis of other ap-proaches that differ from PBS, or theyreviewed only a subset of PBS procedures(Didden, Duker, & Korzilius, 1997;Lancioni & Hoogeveen, 1990; Lennox,Miltenberger, Spengler, & Erfanian, 1988;Matson & Taras, 1989; O'Brien & Repp,1990; Vollmer & lwata, 1992). Conse-quently, previous reviews did not analyze,in depth, the unique contributions madeby PBS per se.

The dearth of synthesis reviewsrelated to PBS was one reason for theNational Institutes of Health commission-ing a Consensus Development Conferencein 1989 to deal with the issue of destruc-tive behavior. One product of thatconference was a synthesis review of PBS

INTRODUCTION

based on the literature that existedthrough 1989 (Carr, Robinson, Taylor, &Carlson, 1990). Shortly thereafter asecond analysis appeared reviewing theliterature from 1976 to 1987 (Scotti,Evans, Meyer, & Walker, 1991). Thisanalysis considered a number of issuesrelevant to PBS but devoted considerablecoverage to non-PBS strategies as well.Finally, a recent review (Scotti, Ujcich,Weigle, Holland, & Kirk, 1996) analyzedintervention practices (but not outcomes)from 1988 to 1992 and extended theinitial Scotti et al. (1991) review. Again,however, the analysis included non-PBSstrategies. In sum, the absence, since1989, of a synthesis that focuses exclu-sively on PBS provided an importantjustification for undertaking the presentreview.

Traditionally reviews have empha-sized issues of special interest to re-searchers: population demographics,experimental design considerations,assessment strategies, and measures ofcomparative intervention effectiveness.But as the field has matured, there hasbeen greater appreciation for the perspec-tives of nonresearchers (consumers). Inthis review, we use the term consumers toinclude people with disabilities, theirteachers, their friends, members of theirfamilies, administrators, and policymakers. The literature suggests thatalthough PBS has made a major contribu-tion in dealing with the issue of problembehavior, a considerable gap existsbetween the needs and interests ofresearchers and nonresearcher consum-ers (Billingsley & Cross, 1991; Dunlap,Robbins, & Darrow, 1994; Haring, 1996;Horner, Diemer, & Brazeau, 1992; Sailor,1996; Turnbull & Turnbull, 1996).Considering this gap, we set out todetermine how far the literature has come

in addressing this gap and, by implica-tion, what must be done next. Thisobjective could be achieved only througha careful analysis of those parts of thePBS literature that, explicitly or implicitly,bear on the perspectives ofnonresearchers. The relative absence ofthis type of analysis from previousreviews was another reason to examinethe PBS literature as it has evolved to date.

Research Questions Posed:Contributions of

the ReviewIn this section, we outline the majorresearch questions posed in the reviewand the kinds of information derivedfrom answering them. First, however, it isnecessary to elaborate further on thedefining characteristics of PBS so theycan be systematically related to the keyresearch questions.

Elaboration of DefiningCharacteristics of PBS

From the standpoint of the independentvariable, the PBS approach refers to thoseinterventions that involve alteringdeficient environmental conditions (e.g.,activity patterns, choice options, prompt-ing procedures) and/or deficient behaviorrepertoires (e.g., communication, self-management, social skills). The alterationof environmental conditions can beachieved by modifying proximal stimuli(e.g., curriculum materials, prompts) ordistal stimuli (e.g., rearranging thesequence of daily life routines). Becauseall such strategies focus on assessing andmanipulating stimuli, we will refer tothese strategies as stimulus-basedinterventions. The alteration of behaviorrepertoires can be achieved by modifying

21 7

CHAPTER 1

socially appropriate, functional behaviorsthat are currently inadequately developedor absent (e.g., communication, job skills,social skills, independent living skills,self-management behavior). Because allstrategies that increase the probability ofsuch behaviors invariably involve asystematic and targeted application ofreinforcement, we refer to these strategiesas reinforcement-based intervention. Thevarious parameters of stimulus-basedand reinforcement-based interventionsdefine the core of PBS with respect to theindependent variables.

The remediation of deficient contextsalso helps to define the PBS approachwith respect to dependent variables:(a) increased positive behavior,(b) improved lifestyle, and (c) decreasedproblem behavior. Specifically, as noted,stimulus- and reinforcement-basedinterventions are both designed toincrease the probability that sociallyappropriate, functional behaviors (i.e.,positive behaviors) will occur. Forexample, improvements made in instruc-tional procedures (a stimulus-basedstrategy) may increase correct academicresponding (a positive behavior); that is,the revised instructional procedurescontain discriminative stimuli that evokecorrect academic responding at a higherrate. Likewise, strengthening communica-tive skills (a reinforcement-basedstrategy) may increase a variety ofpositive, constructive behaviors such asmaking requests, providing information,and protesting unwanted interactions.Increases in the probability of functionalpositive behaviors (whether produced bystimulus- or reinforcement-basedinterventions) can also potentiallyfacilitate widespread changes in anindividual's life situation, bringing aboutimprovements in social, vocational, and

8

22

educational status (i.e.,lifestyle change).Finally, improvements in environmentalconditions and repertoires of positivebehavior can produce, as a side effect,decreases in problem behavior. In sum,increases in positive behavior, lifestylechange, and subsequent decreases inproblem behavior define the core of PBSwith respect to the dependent variables(Homer, Dunlap et al., 1990).

Research QuestionsHaving defined PBS in terms of its coreindependent and dependent variables, wecan now pose the research questions thatprovide the structure for this review.

How Widely Applicable Is PBS?The answer to this question will contrib-ute information concerning whether PBSinterventions are applicable acrossgender, a broad age range, diagnosis, levelof retardation, and type of problembehavior. Also, the answer will make clearwho implements PBS (interventionagent) and where it takes place (interven-tion setting).

In What Ways Is the FieldEvolving?The answer to this question will contrib-ute information concerning the trendsthat have taken place over the 12-yearperiod covered by the synthesis. Thesetrends involve a consideration of changesacross time in the volume of literaturepublished as well as the types of interven-tions used, problem behaviors treated,assessments carried out, factors identifiedas maintaining the problem behavior,type of systems change, interventionagents and settings involved (ecologicalvalidity), breadth of intervention effects,and judgments of outcome and socialvalidity made by significant others (e.g.,

INTRODUCTION

consumers such as parents, teachers, jobcoaches). The analysis of trends alsoprovides information as to whether thefield as a whole is progressing toward amore widespread adoption of what areregarded as best practices.

How Effective Is PBS?Answers to this question provide criticalinformation used to compare theeffectiveness of the entire category ofstimulus-based intervention with theentire category of reinforcement-basedintervention. We also present informationconcerning changes in effectiveness whenthe categories are combined with oneanother and with non-PBS strategies thatare frequently a part of a multicomponentapproach to intervention.

What Factors Modulate theEffectiveness of PBS?Answers to this question provideinformation concerning how demo-graphic, assessment, systems change, andecological validity variables affectoutcomes, that is, how these variablesmodulate the effectiveness of PBS-basedintervention.

How Responsive Is the PBSLiterature to the Needs of Con-sumers (Nonresearchers)?Given that this question is typically notincluded in a research synthesis, we firstdiscuss the rationale for its inclusion. Thegap between the knowledge produced byresearch and the needs of consumers ofresearch has prompted increasingnational concern (Bruyere, 1993; Carnine,1997; Fuchs & Fuchs, 1990; Hess &Mullen, 1995; Hoshmand & Polkinghorne,1992; Huberman, 1990; Kaufman, Schiller,Birman, & Coutinho, 1993; Lather, 1986;Lloyd, Weintraub, & Safer, 1997). A

number of factors converge, related toPBS, to escalate this national concern:(a) trends of deinstitutionalization andcommunity inclusion for students withsevere disabilities (Braddock, Hemp,Fujiura, Bachelder, & Mitchell, 1990),(b) the mandate of the VocationalRehabilitation Act to conduct consumer-responsive research (S. Rep. No. 102-357,1992), and (c) the 1997 Amendments tothe Individuals With Disabilities Educa-tion Act that strongly emphasize require-ments for functional behavior assessmentand behavioral intervention in dealingwith issues pertaining to studentdiscipline (Individuals With DisabilitiesEducation Act, 1997).

Answers to the question related tothe responsiveness of the PBS literaturecontribute information concerning thedegree to which the research literature isusable and accessible from consumers'perspectives (Carnine, 1997). We identifyand examine in light of the databasepriority concerns in the consumer litera-ture. The gap between consumer concernsand the database contribute heuristicinformation concerning the formulationof a future research agenda and thedelineation of roles related to researchtranslation, dissemination, and use.

The Structure of theResearch Synthesis

In Chapter 2,"Methods:' we (a) opera-tionally define the demographic, assess-ment, intervention, and outcomevariables pertinent to the database,(b) explicate literature search strategiesand eligibility (inclusion/exclusion)criteria, and (c) describe data collectionand measurement methods.

Chapter 3,"Results',' is structuredaround the first four research questions. It

23 9

CHAPTER 1

begins with summary descriptivestatistics on the demographic variablesthat characterize the literature included,thereby clarifying whether PBS is widelyapplicable across various demographiccharacteristics, problem behaviors, andintervention agents and settings. Thissection is followed by a presentation ofdata dealing with trends across time.Then we present data on interventioneffectiveness, followed by data onvariables that modulate effectiveness.

In Chapter 4,"Discussion:' we firstdeal with the issue of potential biases inthe literature retrieved. Then we develop anumber of generalizations that can be

24

10

inferred from the database on demo-graphics, assessment, interventions, andoutcomes. Within this chapter, we discussthe impact of the results on assessmentand intervention practices. The lastsection outlines where the major gaps inknowledge are and offers a plan foraddressing these gaps. This section drawson the database to address the finalresearch question, namely, how respon-sive the PBS literature has been to theneeds of consumers (nonresearchers).

In the final two chapters of thereview, we summarize the major findingsand provide a list of recommendations foradvancing the field of PBS.

METHODS

CHAPTER 2

METHODSIn this chapter, we (a) provide operationaldefinitions of the four categories ofvariables, (b) explicate the literaturesearch and eligibility criteria, and(c) describe the data collection methods.

Operational DefinitionsEach article identified was scored withrespect to fourcategories of variables:(a) demographics, (b) assessmentpractices, (c) intervention strategies, and(d) outcome measures.

DemographicsThe following demographic variableswere scored: (a) the year in which thearticle was published (1985-1996); thosearticles listed as being in press at the timeof the review were assigned to the year1996, for reasons explained later in thesection on eligibility criteria; (b) genderof participants (male or female);(c) diagnosis (mental retardation;autism/pervasive developmental disabil-ity; mental retardation + autism/pervasive developmental disability;mental retardation and/or autism/pervasive developmental disability +other [e.g., anxiety disorder, motor skillsdisorder, tic disorder, etc.]); (d) age (agein years rounded to the nearest wholenumber); (e) level of mental retardation(profound, severe, moderate, mild);(f) type of problem behavior (aggression,self-injurious behavior, property destruc-tion, tantrums).

Assessment PracticesAssessment practices are the methodspractitioners and researchers use todetermine (a) classes of problem

behavior, (b) antecedents that occasionand do not occasion problem behavior,and (c) variables responsible for main-taining problem behavior. These main-taining variables are often referred to asthe function, purpose, goal, intent,reinforcers, or motivation of problembehavior, terms that are roughly synony-mous (Carr, 1993; Lee, 1988; Skinner,1974).

The objective of assessment is togenerate information that can be used toguide the selection and development ofintervention strategies.

There are three categories ofassessment: indirect observation(sometimes referred to as informalobservation), direct observation (some-times referred to as formal observation),and functional analysis (Lennox &Miltenberger, 1989; O'Neill et al., 1997a &b; Sturmey, 1994). Indirect or informalobservation involves assessment strate-gies in which information about problembehavior (B), its antecedents (A), andconsequences (C) are gathered indirectly(via informants) through the use ofinterviews (e.g., Carr et al., 1994; O'Neillet al., 1997 a & b), questionnaires (e.g,Durand & Crimmins, 1992), rating scales(Aman, Singh, Stewart, & Field, 1985),setting-event inventories that focus onbroad contextual variables such as dailyschedules and health status (e.g., Gardner& Sovner, 1994; O'Neill et al., 1997a & b),or anecdotal observations. Direct orformal observation involves directmeasurement through the use of A-B-Cdata sheets (e.g., Bijou et al., 1968),scatterplots that document temporalcorrelations between problem behaviorand specific situations (Touchette,

25 11

CHAPTER 2

MacDonald, & Langer, 1985), and timesampling and/or frequency counts (e.g.,Lalli, Browder, Mace, & Brown, 1993;O'Neill et al., 1997a & b). Functionalanalysis involves the systematic (experi-mental) manipulation of the variablesthought to control problem behavior andis carried out to test hypotheses aboutmotivation (Carr, 1994). Assessmentsmay be repeated over time if circum-stances change, warranting furtherinvestigation of motivational hypotheses(Mace, 1994; Vollmer, Marcus, & LeBlanc,1994).

The product of these assessments isa statement concerning the problembehavior, controlling antecedents, andmaintaining consequences (motivation).There are four commonly identifiedmotivational categories: attention, escape,tangibles/activities, and sensory rein-forcement (Carr, 1977; Iwata, Dorsey,Slifer, Bauman, & Richman, 1982; Wiesler,Hanson, Chamberlain, & Thompson,1985). Sometimes problem behavior(a) functions to secure attention,nurturance, and comfort from others(Carr & McDowell, 1980; Lovaas, Freitag,Gold, & Kassorla, 1965; Martin & Foxx,1973); (b) helps individuals escape oravoid difficult, boring, or arduous tasksand other aversive situations (Carr &Newsom, 1985; Carr, Newsom, & Binkoff,1976, 1980; Patterson, 1982); (c) helpsprovide the individual with access todesirable tangible items and preferredactivities (Derby et al., 1992; Durand &Crimmins, 1988); or (d) generates sensoryreinforcement in the form of visual,auditory, tactile, and even gustatorystimulation (Favell, McGimsey, & Schell,1982; Rincover, Cook, Peoples, & Packard,1979). A given problem behavior mayhave more than one function (Day,Horner, & O'Neill, 1994; Haring &Kennedy, 1990; Iwata et al., 1982). It

1226

should be noted that other sources ofmotivation, such as social avoidance(Taylor & Carr, 1992), have been identi-fied. Indeed, it has been hypothesized thatthere may be as many as 16 differentmotives for problem behavior (Reiss &Havercamp, 1997). However, the empiricalbase demonstrating these additionalsources of motivation is, as yet, too smallfor review purposes.

SummaryUsing the terms just delineated, we scoredthose aspects of assessment practices thatcorresponded to the following questions:(a) Was there an assessment of function(yes/no)? (b) What assessment strategywas used (informal observation, formaldirect observation, functional analysis)?(c) What functions were identified(attention, escape, tangibles/activities,sensory)? (d) Was the assessmentrepeated over time? (e) Was the assess-ment information subsequently used todesign an intervention?

Intervention StrategiesWithin the theme of interventionstrategies, we scored articles with respectto (a) intervention category,(b) systems change, and (c) ecologicalvalidity.

Intervention CategoriesAs noted previously, there are twocategories of PBS intervention: thosedesigned to make positive behavior moreprobable by (a) remediating deficientenvironmental conditions (stimulus-based intervention) and (b) remediatingdeficient behavior repertoires(reinforcement-based intervention).The literature reviewed demonstrated alarge number of variations associatedwith each category of intervention but,

METHODS

typically, only a small number of casesassociated with any particular variation.To further clarify the defining propertiesof the two generic categories that servedas the basis for our data coding, wedelineate here the characteristics of someof the many variations identified in theliterature.

Stimulus-Based InterventionFrom a functional standpoint, deficientenvironments provide too few stimulithat support positive behavior and toomany that support problem behavior.Thus, one functional theme unites allvariations in this category: Environmen-tal repair is attempted through themanipulation of stimuli that are proximaland distal to the behaviors of interest. Thestructural nuances that distinguish eachvariation are less important than thefunctional utility shared by all in modify-ing the environment to promote positivebehavior. To illustrate the breadth ofstimulus-based intervention, we heredescribe some of the more salientvariations noted in the literature.

Interspersal training. In the literaturethis strategy is variously referred to asinterspersal training (Homer, Day,Sprague, O'Brien, & Heathfield, 1991),behavioral momentum (Mace, Hock, etal., 1988), pretask requesting (Singer,Singer, & Homer, 1987), task variation(Dunlap & Koegel, 1980; Winter ling,Dunlap, & O'Neill, 1987), and embedding(Carr et al., 1976). The essence of theseprocedures is to present a stimulus (e.g., adifficult task demand such as "clean upyour toys") known to be discriminativefor problem behavior (e.g., aggression)within the context of stimuli (e.g., easydemands such as "give me a hug") knownto be discriminative for nonproblembehavior (e.g., complying with the requestto hug). These stimulus changes result in

the formerly problematic stimulus (i.e.,"clean up your toys") now evokingcooperation rather than aggression.

Expansion of choice. This strategyinvolves presenting the individual with anumber of choice stimuli (options)related to a wide variety of activities and/or tasks, and permitting the individual toexpress a preference (choose) among theoptions. Research suggests that expand-ing choices can be an effective way toreduce problem behavior (Bannerman,Sheldon, Sherman, & Harchik, 1990;Dunlap, dePerczel et al., 1994; Dyer,Dunlap, & Winterling, 1990; Koegel, Dyer,& Bell, 1987; Vaughn & Homer, 1997).

Curricular modification. The essence ofthis strategy is to identify the aversivefeatures of task stimuli that evoke escape-motivated problem behavior, and then tominimize or eliminate those features.Dunlap, Kern-Dunlap, Clarke, andRobbins (1991) demonstrated the efficacyof this procedure by altering features suchas task length, task outcomes, and clarityof instructions, the modification of whichwas correlated with decreases in problembehavior.

Manipulation of setting events.Another intervention from the genericstimulus-based category involves themanipulation of setting events. Settingevents are broad contextual variables(often involving distal stimuli) that alterthe relationship between discriminativestimuli and responses (Bijou & Baer,1961). Setting events include (a) physicalfactors such as environmental enrich-ment (Homer, 1980), (b) biologicalfactors such as drugs (Thompson et al.,1991) and illness (Carr & Smith, 1995),and (c) social factors such as the presenceversus absence of specific people(Touchette et al., 1985) and the sequenc-ing of interpersonal activities (Brown,

2713

CHAPTER 2

1991). Setting events often serve asestablishing operations (Michae1,1982),that is, as factors that momentarily changethe reinforcing or aversive properties ofresponse consequences, thereby influenc-ing the probability of constructivebehavior as well as problem behavior.

Reinforcement-Based InterventionA deficient positive-behavior repertoiremakes it difficult (or impossible) for anindividual to meet his or her needs (i.e.,access preferred reinforcers), which, inturn, increases the level of frustration(i.e., maximizes episodes of extinction),thereby leading to problem behavior; if anindividual's current repertoire ofnonproblem (positive) behavior isineffective in gaining reinforcers but hisor her problem behavior is effective, thenproblem behavior will become moreprobable. From a functional standpoint,the presence of positive behaviorscompete with and/or make problembehavior unnecessary because thepositive behaviors themselves providealternative avenues for accessing valuedreinforcers.

There are many variations ofreinforcement-based interventions. Onecommon theme unites all the variations:They target specific behaviors or classesof behaviors for consistent, systematicreinforcement.

By way of illustrating this broadgeneric category, we here describe severalof the many variations of reinforcement-based interventions noted in the litera-ture. Note that one procedure, differentialreinforcement of other behavior (DRO),despite its name, is not an example ofreinforcement-based intervention giventhe criteria we are using. DRO involvesdelivering reinforcement contingent onthe nonoccurrence of the target problembehavior for a prespecified period of time

2814

(Vollmer & Iwata, 1992). No positivebehavior(s) is (are) explicitly targeted forreinforcement, which is why someresearchers prefer to call the procedure"differential reinforcement of notresponding" or "zero responding" (Poling& Ryan, 1982; Zeiler, 1970). Indeed, somehave argued that the DRO procedure canbe viewed as a form of punishment,because frequent display of problembehavior results in repeated omission ofpositive reinforcers, an aversive event(Rolider &Van Houten, 1990).

Functional communication training.This intervention involves teaching anindividual a specific communicativeresponse that serves the same function(functional equivalence) as the problembehavior it is intended to replace (Carr &Durand, 1985). In illustration, a func-tional analysis indicates that a young girlwith autism becomes self-injurious in thepresence of negative academic feedbackbecause of a history of reinforcement forsuch behavior (i.e., self-injury results intermination of the putatively aversiveinstructional situation). Following thefunctional analysis, the teacher makeschanges. She prompts the girl to request"help" in response to negative feedback(e.g., the teacher says,"No, that's not theright answer"; then the teacher provides aprompt, "Say, `Help me, please"). If thenew communicative response is moreefficient (Horner & Day, 1991) at termi-nating the aversive events associated withthe task than self-injury (i.e., communi-cation requires less effort to escape fromthe negative feedback than self-injuriousbehavior), then the problem behavior islikely to decrease (Bird, Dores, Moniz, &Robinson, 1989; Carr & Durand, 1985;Day, Rea, Schussler, Larsen, & Johnson,1988; Durand & Carr, 1991, 1992; Wackeret al., 1990).

METHODS

Self-management. Self-managementinvolves any of three component skills:self-monitoring, self-evaluation, and self-reinforcement. Self-monitoring involvesteaching an individual to discriminateappropriate versus inappropriatebehaviors and to describe each of them(e.g., "I made my bed" versus "I bit myhand"). Self-evaluation consists oflabeling a behavior as desirable versusundesirable. For example, after makingthe bed, the individual might be taught tosay, "I did a good job:' In contrast, afterself-biting, the individual might be taughtto say,"I didn't do a good job:' Self-reinforcement consists of teaching theindividual to deliver reinforcers (e.g.,praise, tangibles) to him- or herselffollowing a positive self-evaluation butnot after a negative self-evaluation. Insum, the individual is taught to positivelyreinforce desirable response alternativesto the problem behavior. Studies suggestthat instruction in self-management canreduce or eliminate disruptive, aggressive,and self-injurious behaviors (Gardner,Cole, Berry, & Nowinski, 1983; Koegel,Koegel, Hurley, & Frea, 1992).

Differential reinforcement of alterna-tive behavior (DRA). This interventioninvolves reinforcing those behaviors thatare topographically different from thetargeted problem behavior. The newbehaviors thus serve as alternatives to theproblem behavior. In an early demonstra-tion of DRA, Hall, Lund, and Jackson(1968) reinforced (with teacher attention)the "study" behavior of elementary schoolstudents, while applying extinction totheir disruptive behavior. As the alterna-tive (study) behaviors were strengthened,there was a concomitant decrease inproblem behavior (disruption).

Non-PBS (Environmentally Based)InterventionThere were many instances in theliterature in which PBS interventionswere combined with non-PBS (environ-mentally based) interventions. Theselatter interventions were defined as thosefor which the primary goal was thereduction of problem behavior throughthe direct application of proceduresreactive to the display of problembehavior. Of the many variations of non-PBS procedures, three are described thatillustrate the reactive nature of non-PBS(in contrast to the proactive natureof PBS).

The first procedure, differentialreinforcement of other behavior (DRO),has already been discussed. In brief, DRO,applied during a bout of problembehavior, involves delivering reinforce-ment contingent on the nonoccurrence ofproblem behavior for a prespecifiedperiod of time. A second procedure,extinction, involves withdrawing thereinforcer that maintains the problembehavior each time that the problembehavior occurs. The status of extinctionas a non-PBS procedure is debatable. In alater section, we discuss the possibility,based on conceptual and pragmaticconsiderations, that extinction might alsobe viewed as a key aspect of PBS. A thirdprocedure, timeout, involves the with-drawal of all positive reinforcement for afixed time period following the occur-rence of problem behavior. This proce-dure is considered a form of punishment.

Systems ChangePBS involves systems change, not justchange in the individual who displaysproblem behavior. Both stimulus-basedintervention and reinforcement-basedintervention potentially result in changes

2 9 {- 15

CHAPTER 2

in how other people respond to theperson with disabilities and how theenvironment is organized. We examinedthese two aspects of systems change.

Behavior Change on the Part ofSignificant OthersWe determined whether people otherthan the person with disabilities wererequired to alter aspects of their behavioras part of the intervention. In illustration,consider functional communicationtraining. If an individual with disabilitieswas taught to request help in response toa difficult task, then the support person(e.g., a parent or teacher) would beexpected to respond to the request byproviding help; the support person wasrequired to change his or her behavior inresponse to change in the behavior of theperson with disabilities. If an investigatorexplicitly noted such a requirement, wescored this aspect of systems change asbeing present.

Broad Environmental Reorganizationand RestructuringWe also determined whether broadreorganization and/or restructuring ofthe environment (Emerson et al., 1994)was reported as part of the interventionapproach. These environmental variablesincluded any variation of the following:systematic personnel changes; alterationsin the scheduling of activities; provisionof supported employment; provision ofnew, enriching community activities;provision of respite services; friendshipfacilitation; provision of additional staff;physical alteration of the home and/orschool setting; addition (and/or removal)of individuals with disabilities to/fromspecific classrooms and/or group homes;and provision of choices all day long (i.e.,not just in selected circumstances). If aninvestigator systematically included one

16 30

or more of the preceding variables as partof an intervention, this aspect of systemschange was scored as being present.

Ecological ValidityPBS is not intended to be a laboratory-based demonstration or analog but,rather, a strategy for dealing withproblem behavior in all pertinent naturalcontexts. We documented this aspect ofPBS intervention by examining (a) whocarried out the intervention (the inter-vention agent), (b) where the interventiontook place (intervention setting), and(c) whether the intervention was imple-mented in all contexts in which problembehavior was noted to occur (all relevantcontexts).

Intervention AgentA distinction can be made betweenintervention agents who would normallybe expected to be the primary supportpeople/caregivers in a particular commu-nity setting (hereafter referred to astypical intervention agents) and thosewho would not normally be involved(hereafter referred to as atypical interven-tion agents). In the home setting, thetypical intervention agent would be aparent or other close relative of a child oradolescent; in the school setting, ateacher; in a group home setting, directcare staff; in supported living, ahousemate; and, in the workplace, a jobcoach or designated fellow employee. Wedefined all such individuals, in these andother relevant community settings, astypical intervention agents. In contrast,atypical intervention agents includedpsychologists, behavior specialists,researchers, and others who would not beexpected to provide support on a day-to-day basis under normal circumstances.

In our synthesis, the involvement of

METHODS

typical intervention agents reflects highecological validity; the involvement ofatypical agents reflects low validity.

Intervention SettingA distinction can be made between livingenvironments that are considerednormative for an individual of a given age(hereafter referred to as typical settings),and those not considered normative(hereafter referred to as atypical settings).According to this criterion, typicalsettings include the home, integratedschool, group home/own home, job site,neighborhood, and a variety of commu-nity settings (e.g., those related torecreational activities, shopping, eating,and entertainment). Using the samecriterion, atypical settings includesegregated schools, psychiatric wards/hospitals, medical clinics, state institu-tions, and sheltered workshops. Withrespect to the concept of PBS, interven-tions taking place in typical settingsreflect high ecological validity, whilethose taking place in atypical settingsreflect low validity.

Intervention in All Relevant ContextsContext has two dimensions: temporaland situational. For scoring purposes, therelevant context for carrying out anintervention for problem behaviorincluded all the time periods for whichthe problem was reported to occur, andall the situations for which it was reportedto occur. In illustration, a teacher mightreport that problem behavior occurredthroughout the school day and acrossmany different situations (e.g., duringgym, lunch time, reading, group circle,dismissal, boarding the bus). An investi-gator might respond to the teacher'sreferral by removing the child from theclassroom and conducting 20-minuteintervention sessions in a special tutorial

room. Because these sessions do not coverthe entire time period or array ofsituations for which the teacher reportedthe presence of problem behavior, wescored the implemented intervention asnot having occurred in all relevantcontexts. In contrast, a teacher mightreport that problem behavior occurredexclusively during the 20-minute dailygym session because the motor activitiesinvolved were singularly aversive to theindividual. To bring about improvedbehavior, an investigator might instructthe teacher, in the gym setting, to alter thecurriculum during the 20-minute session.In this case, the intervention sessioncovers the entire time period in thesituation identified by the teacher. Thisintervention would be scored as havingoccurred in all relevant contexts.

Because PBS interventions areintended to deal with problem behaviorwhenever and wherever it naturallyoccurs, the first intervention describedwould reflect low ecological validity; thesecond would reflect high validity.

SummaryUsing the terms just delineated, we scoredthose aspects of an intervention thatcorresponded to the following questions:(a) Was the intervention stimulus-based,reinforcement-based, non-PBS based(yes/no for each)? (b) Systems change:Did the intervention involve change onthe part of significant others and/orbroad environmental reorganization andrestructuring (yes/no for each)?(c) Ecological validity: What type ofintervention agent was involved (typical/atypical)? What type of interventionsetting was involved (typical/atypical)?Did intervention occur in all relevantcontexts (yes/no)?

31 17

CHAPTER 2

Outcome MeasuresBecause PBS involves the multidimen-sional remediation of deficient context,the outcomes of the approach are likewisemultidimensional. We examined sevenoutcome measures: (a) positive behavior,(b) problem behavior, (c) stimulusgeneralization, (d) response generaliza-tion, (e) maintenance, (f) lifestyle change,and (g) social validity.

. .

Positive BehaviorPBS intervention involves the use ofstrategies designed to make sociallydesirable responses (positive behaviors)more probable. Therefore, we examinedwhether positive behaviors did indeedincrease following intervention. For eacharticle we scored changes in positivebehaviors from baseline to intervention.In illustration, consider a reinforcement-based procedure such as functionalcommunication training. Because theprocedure teaches specific communica-tive alternatives to problem behavior, onewould expect to see increases in the level(i.e., frequency, percentage) of thisalternative behavior following interven-tion. When data on communicativeresponses were reported in an article,we scored them. In principle, eachreinforcement-based procedure should beassociated with an increase in one ormore types of positive behavior.

Likewise, stimulus-based proceduresshould promote positive behavior, byaltering features of the environment. Inillustration, consider a procedure such asinterspersal training. When a difficulttask demand (known to be discriminativefor problem behavior) is interspersedamong stimuli known to be discrimina-tive for cooperation, one would expect anincrease in the level of the latter positive

18 32

behavior. Again, in our synthesis wescored all data reflecting changes inpositive behavior.

Problem BehaviorWe measured another key outcome: .

whether an intervention produced adecrease in problem behavior. Interven-tion effects were always measured interms of percentage reduction in problembehavior relative to baseline (using thecalculation method described later in thisreview).

Stimulus GeneralizationWe measured stimulus generalization,defined as the degree to which interven-tion effects transferred from the originalintervention situation to other situationsinvolving new intervention agents, .

settings, and tasks. Stimulus generaliza-tion thus referred to a behavior changethat occurred in spite of the fact that noplanned intervention occurred in the newsituation. We also measured the degree towhich decreases in problem behaviorgeneralized to new situations; this wasalso measured in terms of percentagereduction in problem behavior relative tobaseline.

Response GeneralizationWe measured response generalization,defined as the degree to which interven-tion effects transferred from the initialtarget(s) of intervention to other aspectsof the individual's behavior repertoire nottargeted for intervention. Let's say anintervention targets ameliorating self-injurious behavior via communicationtraining. The outcome may also demon-strate that aggressive behavior decreasedand social play increased even thoughthese two behaviors were not the focus ofintervention. The desirable side effects ofintervention (i.e., the decrease in

METHODS

aggression, and the increase in socialplay) constitute response generalization.Again, these effects were measured aspercentage change from baseline.

MaintenanceMaintenance was defined as the degree towhich intervention effects lasted overtime (intervention durability). Mainte-nance was further defined as involvingany of the following: (a) those datacollected only after the specific interven-tion had been completely terminated;(b) data collected after the interventionhad been modified (but not terminated)in some substantive way, such as adecrease in the number of formalintervention sessions per unit of time, adecrease in the number of interventioncomponents in effect (i.e., interventionfading), or a gradual reduction in theinvolvement of the intervention agent(i.e., the agent decreases the amount oftime given to intervention implementa-tion). We measured maintenance effectsas percentage reduction from baseline;when available, effects were noted at thesespecific follow-up periods: 1 to 5 months,6 to 12 months, 13 to 24 months, and 25months or more.

Lifestyle ChangeBecause the purpose of PBS is not simplyto reduce the level of problem behavior,but also to enable individuals to live morenormalized lives, a key outcome measurerelates to lifestyle change. Positive lifestylechange was defined as increased engage-ment in normative social, vocational,family, recreational, and academicactivities. Lifestyle change effects weremeasured, when available, as percentageincrease from baseline.

Social ValidityWolf (1978) argued that the impact ofinterventions cannot be gauged solely byobjective measures. Unless significantothers (e.g., parents, teachers, jobcoaches, friends, members of thecommunity) perceive the interventionand its effects to be worthwhile, theintervention would be judged as inad-equate. Considering this, social validitywas also measured as an outcome.

We examined articles to determinewhether rating scales were reported thattapped three critical dimensions of socialvalidity: feasibility, desirability, andeffectiveness. The generic feasibilitydimension involved any variant of thequestion,"Would you be able to use thisintervention strategy?" The genericdesirability question involved any variantof the question, "Would you be willing touse this intervention strategy?" Theeffectiveness dimension was subdividedinto (a) effectiveness with respect toreduction in problem behavior, and(b) effectiveness with respect to lifestylechange. Effectiveness with respect toproblem behavior involved any variant ofthe question,"Does this interventionstrategy reduce problem behavior to alevel that is acceptable to you?" Effective-ness with respect to lifestyle changeinvolved any variant of the question,"Does this intervention strategy make adifference in the lifestyle of the individualinvolved in terms of increasing opportu-nities to live, work, go to school, recreate,and socialize with typical peers andsignificant others in typical communitysettings?"

3319

CHAPTER 2

SummaryUsing the terms just delineated, we scoredthose aspects of outcome measurementrelated to the following questions:(a) Were data available on positivebehaviors resulting from the intervention(yes/no)? If so, specify the baseline andintervention data for each positivebehavior. (b) Were data available onreduction in problem behavior followingintervention (yes/no)? If so, specify thebaseline and intervention data for eachtype of problem behavior reported.(c) Regarding stimulus generalization,were anecdotal observations available(yes/no)? Were direct observation dataavailable (yes/no)? If direct observationdata were available, specify the baselineand intervention data for problembehavior. (d) Regarding responsegeneralization, were anecdotal observa-tions available (yes/no)? Were directobservation data available (yes/no)? Ifdirect observation data were available,specify the baseline and intervention datafor socially appropriate behavior andproblem behavior. (e) Regarding mainte-nance, were data on reduction in problembehavior noted at the following specifiedfollow-up periods: 1 to 5 months, 6 to 12months, 13 to 24 months, 25 months ormore (yes/no for each)? If direct observa-tion data were available, specify thebaseline and intervention data.(f) Regarding lifestyle change, was alifestyle change considered (anecdotallyor formally) as a goal of the study (yes/no)? Was there a formal intervention toimprove lifestyle (yes/no)? Was theremeasured success in producing a lifestylechange (yes/no)? If direct observationdata were available on lifestyle change,specify the baseline and interventiondata. (g) Regarding social validity, wasthere a generic feasibility question (yes/no)? Was there a generic desirability

20 34'

question (yes/no)? Was there a genericquestion regarding effectiveness withrespect to reduction in problem behavior(yes/no)? Was there a generic questionregarding effectiveness with respect tolifestyle change (yes/no)? For each of thepreceding questions, specify the pre-and postintervention data for each scale.

Literature Search andEligibility Criteria

We established initial selection criteria.Then we conducted a literature searchusing these criteria. Finally, we appliedexclusion criteria to eliminate thosearticles that did not meet desiredmethodological standards.

Initial Selection CriteriaThere were six criteria that guided theinitial selection of articles from theliterature. First, an article had to havebeen published between 1985 and 1996.All articles accessible to us by the cutoffdate of December 31,1996, were consid-ered. Some 1996 journals had delayedpublication dates in which the final issuefor 1996 was released three to fourmonths past the cutoff date. To compen-sate for the potential loss of relevant 1996articles as well as to achieve the most up-to-date review, we included a smallnumber (6) of in press articles that we hadobtained prior to the cutoff date. Thesewere classified as 1996 articles. (However,it should be noted that these 1996 articleswere no longer in press by the time thisreview was completed and, therefore, arecited in the reference list as 1997 articles.)

Second, in an attempt to ensure highstandards, only articles published in peer-reviewed journals were considered. Non-peer-reviewed manuscripts were notconsidered in the analysis in part to help

METHODS

ensure a uniformly high standard ofexperimental rigor, and in part becausesingle-subject research tradition relies onthe judgment of anonymous peerreviewers to confirm the presence of afunctional relationship. The inclusion ofnon-peer-reviewed studies (e.g., disserta-tions, clinical trials) would have imposeduncontrolled sources of error, becauseeach study would have required asurrogate peer review (by us) prior toinclusion to document the presence ofexperimental control.

Third, the article had to have beenpublished in English.

Fourth, with respect to diagnosis, weexamined all variations of DSM-III, DSM-III-R, DSM-IV, and AAMR classificationsrelated to mental retardation, autism, andpervasive developmental disorder, eitheras a primary or secondary diagnosis.Thus, relevant dual diagnoses (e.g.,anxiety disorder of childhood withmental retardation) were also retained.

Fifth, with respect to topography, thefollowing types of problem behaviorswere examined: self-injury, aggression,property destruction, and tantrums.

Sixth, with respect to intervention,all variations of stimulus- and rein-forcement-based intervention, as definedearlier, were included.

Literature SearchWe began by hand-searching all relevanteducation, psychology, and medicaljournals listed in four previous reviewsthat had included a consideration of PBS(Carr et al., 1990; Didden et al., 1997;Scotti et al., 1991; Scotti et al., 1996). Thearticles gleaned from this initial processproduced references to additionalresearch articles, review papers, books,book chapters, and newsletters. These

reference trails were, in turn, pursued.Additional reference trails were generatedwhen the following abstract and indexservices were searched by crossing thedisability diagnoses with the problembehavior topographies: Child DevelopmentAbstracts and Bibliography, CurrentContents/Social and Behavioral Sciences,ERIC, MEDLINE, Psychological Abstracts,PsychINFO, PsychLIT, PsychSCAN/MR, andthe Social Science Citation Index.

We also requested information onintervention for problem behavior fromorganizations having a stake in providingservices for people with disabilities. TheNational Information Center for Childrenand Youth With Handicaps provided uswith their list of 33 stakeholder organiza-tions that included The Association forPersons With Severe Handicaps, The ARC(formerly the Association for RetardedCitizens), Autism Society of America,Council for Exceptional Children, and TheNational Down Syndrome Society.

Finally, we requested informationfrom leading researchers (14), askingthem to send us their published and inpress papers dealing with the issue ofproblem behavior. We defined "leadingresearcher" as any individual having atleast three published articles relatedto PBS.

Using the initial selection criteria andsearch methods just explicated, weidentified 216 articles from 36 journals.

Exclusion CriteriaTo ensure the highest quality database forsubsequent analyses, we applied a numberof methodological exclusion criteria to theinitial sample of 216 articles. The applica-tion of these criteria resulted in a finalsample of 109 articles that were included,and 107 that were excluded. (It should be

35 21

CHAPTER 2

Table 1. Journals Examined and Number ofArticles Included and Excluded

No. of articlesJournal included

No. of articlesexcluded

Adult Foster Care Journal 0 1

American Journal of Medical Genetics 1 0

American Journal on Mental Retardation

(formerly American Journal of Mental Deficiency)0 9

Analysis and Intervention in Developmental Disabilities 2 3

Applied Research in Mental Retardation 0 1

Augmentative and Alternative Communication 1 0

Australia and New Zealand Journal of Developmental Disabilities 1 3

Behavior Modification 6 5

Behavior Therapy 4 0Behavioral Disorders 2 0Behavioral Interventions 3 3

Behavioral Residential Treatment 3 13

Behaviour Research and Therapy 0 1

Behavioural Psychotherapy 0 5

Child and Family Behavior Therapy 1 I

Education and Training in Mental Retardation(formerly Education & Training of the Mentally Retarded)

3 7

Education and Treatment of Children 3 I

Exceptional Parent 0 1

Journal of Applied Behavior Analysis 46 14

Journal of the Association for Persons With Severe Handicaps 10 7

Journal of Autism and Developmental Disorders 2 5

Journal of Behavioral Education 2 0Journal of Behavior Therapy and Experimental Psychiatry 4 2

Journal of Consulting and Clinical Psychology 1 0Journal of Developmental and Physical Disabilities 4 5

Journal of Intellectual Disability Research 1 1

Journal of the Multihandicapped Person 2 0Journal of Visual Impairment and Blindness 0 1

Mental Handicap Research 0 6

Mental Retardation 0 5

Research in Developmental Disabilities 3 2

School Psychology Review 3 0

Special Services in the Schools 1 0

Teaching Exceptional Children 0 I

The Irish Journal of Psychology 0 1

Topics in Early Childhood Special Education 0 3

Total 109 107

Note: The citation information for each included article appears in the Reference List, identifiedwith an asterisk(*).

22 36....

METHODS

Table 2. Criteria Used to Exclude Articles From the Database

Exclusion criterion Number of articles rejected

Absence of data

Inadequate design

Group design

Pooled data

Insufficient baseline data

Insufficient intervention data

23

28

12

33

31

7

Note: Some articles were rejected for more than one reason.

noted that some of the contributionsmade by the excluded studies will bediscussed later in this publication.) Table1 shows the breakdown of the twosamples across the 36 journals.

Table 2 lists the six exclusion criteriaused to select the final sample and thenumber of articles excluded for failing tomeet a given criterion. Some articles wereexcluded for more than one reason, andthus the total is greater than 107.

There were 23 articles excludedbecause no data were repdrted (absenceof data). These articles took the form ofnarrative case reports and extendedanecdotes.

Twenty-eight articles were excludedbecause of inadequate design. Specifically,these articles involved empirical casereports that employed an AB design (i.e.,a baseline (A) condition was followed byan intervention (B) condition). ABdesigns do not meet the internal validitycriteria enunciated in standard method-ology texts on single-subject researchdesigns (e.g., Hersen & Barlow, 1976). Incontrast, articles based on multiplebaseline, reversal, and withdrawaldesigns, all meet these criteria (e.g.,Hersen & Barlow, 1976), and wereretained for analysis.

A small number of articles (12) usedlegitimate group designs involving acomparison between experimental andcontrol groups. But these articles reportedonly group means, making it impossibleto determine how any one individualresponded to an intervention. Because allof our subsequent analyses depended onhaving individual data, the results ofthese articles could not be integrated intoour final database. Further, 6 of thesearticles either failed to report critical dataon problem behavior or pooled multiplemeasures in a way that made it impos-sible to retrieve data pertaining toproblem behavior per se.

An additional 33 articles wereexcluded because they reported onlypooled data; these articles did not uselegitimate group designs but, rather,reported averaged pre/post measures fora group of participants with no controlgroup. Again, the absence of data on anyone individual precluded the possibilityof integrating these data with the resultsobtained from other studies in whichindividual data were retrievable.

Thirty-one articles used acceptablesingle-subject designs but were nonethe-less excluded because they reported fewerthan three baseline points (often only 1

37 23

CHAPTER 2

point). Our descriptive analyses (de-scribed later) required at least threebaseline data points.

Likewise, seven articles wereexcluded because of insufficient interven-tion data (i.e., fewer than 3 interventiondata points).

Data Collection Methods

How Intervention EffectsWere Measured

One of the central issues we address is theimpact of PBS on reducing problembehavior. We always measured interven-tion effects in terms of percentagereduction of problem behavior relative tobaseline. Throughout the text,we will usethe terms percentage reduction andsuppression measure as synonyms. Manyof the articles used a reversal design; thatis, the intervention condition alternatedseveral times with the baseline condition.When this type of design was used,wedeemed the final, rather than earlier,intervention conditions most important,because the critical issue to be addressedconcerned how well an individual wasdoing at the end of intervention. Addi-tionally, because intervention frequentlyproduced a steady downward trend in thelevel of problem behavior, the overallmean for an intervention condition couldactually underrepresent the final effect.To minimize this difficulty, the mean ofthe last three intervention data points wasused so that a judgment could be madeconcerning the degree of participantimprovement at the termination ofintervention. In sum, intervention effectswere measured as the percentagereduction in problem behavior from thelast three sessions of baseline compared

24 38

against the final three interventionsessions of the final interventioncondition.

Frequently, a single participantreceived more than one type of interven-tion. For example, a participant mightfirst be exposed to functional communi-cation training, then choice, and finallyinterspersal training. That is, the personwould have received one reinforcement-based intervention followed by twostimulus-based interventions, providingan opportunity to examine three out-comes. When this situation arose, wecoded the data separately for each of thethree outcomes. In contrast, if a partici-pant had received choice in three differentphases of a reversal design, we coded onlythe final outcome because, in this situation,the same stimulus-based intervention(choice) was simply repeated in eachphase. So for a given participant, anoutcome was defined as the data associ-ated with each unique (nonrepetitive)variation of a reinforcement-based and/or

stimulus-based intervention. Table 3shows the number of outcomes perparticipant. As can be seen, out of atotal of 230 participants, 145 (63%)produced a single outcome, while 85(37%) produced more than one outcome.The total number of outcomes pooledacross participants was 366.

To calculate percentage reduction inproblem behavior (suppression measure),it was necessary to estimate the data fromeach article, point by point, thus generat-ing the baseline and intervention datapertaining to the 366 outcomes. Wemeasured, from the published tables andfigures, the last three baseline pointsprior to intervention and, as notedpreviously, the last three interventionpoints of intervention. The intervention

METHODS

Table 3. Relationship Between Outcomes and Participants

Number ofparticipants

Number of outcomesper participant

Total number ofoutcomes

145 1 145

60 2 120

13 3 39

6 4 24

2 5 10

2 6 12

0 7 0

2 8 16

Note: The total number of participants was 230, and the total number of outcomes was 366.

mean was subtracted from the baselinemean, divided by the latter, and thenmultiplied by 100 to yield the percentagereduction from baseline. In illustration,consider a participant whose baselinefrequency (last 3 data points) of aggres-sive behavior was 10,12, and 14 aggres-sive acts, and whose interventionfrequency (last 3 data points) of suchbehavior was 5,1, and 0 acts. The baselinemean was thus 12, and the interventionmean 2. The percentage reduction frombaseline was therefore (12 2) ÷ 12 x 100

= 83.3%. This method was used for datareported as a frequency or as a percentageof time samples observed. In a handful ofcases (6), data were reported as latency toproblem behavior (e.g., 10 s elapsed fromthe beginning of an observation sessionto the first instance of aggression;therefore, the latency to aggression was10 s). In this case only, the data weretransformed as follows. Any reportedlatency, irrespective of magnitude, wasscored as 1 (i.e., problem behavioroccurred after a specific time interval).The absence of problem behavior wasscored as 0. Transforming the data set to

this binary code permitted the calcula-tions already described. In illustration, ifthe baseline were 1, 1, 1 and the interven-tion data were 1, 0, 0, then the percentagereduction would be 67%.

Recall that, in addition to evaluatingthe initial reduction in problem behaviorfollowing intervention, we also examinedthe data on positive behavior, stimulusgeneralization, response generalization,and maintenance. For these four outcomemeasures, intervention data weretypically not reported; if they were, oftenfewer than three data points wereavailable (e.g., a single data point onstimulus generalization might have beenreported). To provide some indication ofeffectiveness regarding the four measures,we compared the three baseline datapoints with even one intervention pointif that was all that was available. Inillustration, if the baseline stimulusgeneralization data were 12, 14, and16 self-injurious acts and the interventionstimulus generalization datum was2 self-injurious acts, then the percentagereduction was (14 2) ÷ 14 x 100 =

85.7%.

39 25

CHAPTER 2

How Reliability WasMeasured

Three types of reliability were computedrelated to (a) adherence to initialselection and exclusion criteria;(b) agreement on scoring of categoricaland continuous data; and (c) data entry(keystroke errors).

With respect to the initial selectionand exclusion criteria, recall that theinitial selection criteria produced 216articles, and that the exclusion criteriaresulted in 109 articles being retained,and 107 articles being excluded. Werandomly chose 50 articles from the 109that were retained, and 50 articles fromthe 107 that were excluded. We then gavethese 100 articles to one of the coau-thors who was not involved in the initialselection. This coauthor was asked toperform two rating tasks: (a) to applythe six initial selection criteria (de-scribed earlier) to the 100 articles andrender a judgment as to whether thearticles met these criteria, and(b) to apply the six exclusion criteria(described earlier) to the 100 articlesand render a judgment as to whether thearticles met these criteria. The rateragreed with the original decisionthat all 100 articles met the initialselection criteria (100% reliability), andthat the 50 articles that had beenexcluded by the original coder should

4026

indeed have been excluded as per thecriteria used (100% reliability).

With respect to the categorical andcontinuous data, four coders scored thesedata from the 109 articles. Again, one ofthe coauthors not involved in the originalscoring randomly selected 7 articlesoriginally scored by each of the fourcoders and recoded all of the categoricaland continuous data from this sample of28 articles. Recall that the continuous datawere based on the following variables:positive behavior, problem behavior,stimulus generalization, response gener-alization, maintenance, and lifestylechange. All the remaining variables (e.g.,age, gender, problem behavior topography,etc.) were categorical. For the continuousdata, the Pearson product-momentcorrelation, based on point-by-pointreliability, was +0.99 (p = .000). For thecategorical data, Kappa values (Cohen,1960) ranged from .82 to 1.00. Landis &Koch (1977) characterized Kappa valuesgreater than .75 as representing excellentagreement beyond chance.

With respect to data entry (keystrokeerrors), each of the four coders reentereddata for seven randomly selected articlesthat they had previously scored. There wasa total of 81,921 keystrokes across the fourcoders, of which 94 differed from theoriginal entries, yielding an error rate ofonly 0.11%.

RESULTS

CHAPTER 3

RESULTS

As noted before, 109 published articlesthat met all the inclusion and exclusioncriteria produced data on 230 participantsinvolving 366 outcomes. We now presentthese data with respect to the researchquestions posed earlier in this review.

How WidelyApplicable is PBS?

Table 4 displays the data on the character-istics of participants involved in PBSinterventions as well as data on the type ofintervention agent and type of interven-tion setting associated with the variousparticipants. In a small number of cases,investigators failed to specify pertinentinformation (e.g., no diagnosis was given)or the description given was ambiguous(e.g., the participant was said to be"retarded" but no level of retardation wasnoted). Therefore, the numbers reportedrepresent the percentage of the databasefor which each characteristic was ad-equately specified.

Approximately twice as many malesas females were involved in PBS interven-tions. Although PBS was applied across theentire age range, preschool children wereleast likely to receive this type of interven-tion followed, in increasing order, byadolescents and adults (i.e., those 20 yearsof age and over); elementary-school-agechildren were most likely to receiveintervention. The latter two age categorieseach accounted for one third of theparticipants.

About half the participants werediagnosed as having mental retardation,and one tenth as having autism. Theremaining participants had combined

diagnoses of retardation and/or autism,frequently accompanied by additionaldiagnoses (e.g., seizure disorder, braindamage). With respect to level of retarda-tion, one third of the participants werefunctioning in the profound range, andanother third in the severe range. Theremaining participants were equallydivided between the moderate and mildrange.

The data on type of problem behaviorshowed that about one third of theparticipants displayed self-injuriousbehavior, and almost a quarter displayedaggression. Property destruction andtantrums were exhibited by only a smallpercentage of the participants. However,various combinations of the precedingfour types of problem behavior wereshown by fully one third of the partici-pants.

PBS was a little more likely to beimplemented by atypical interventionagents (e.g., psychologists, researchers)than typical agents (e.g., parents, teach-ers), although in a very small number ofcases, both atypical and typical agentscombined their efforts: In contrast, by aratio of 2:1, interventions were more likelyto occur in atypical settings (e.g., segre-gated schools, medical clinics) thantypical settings (e.g., home, community,integrated schools).

In What Ways isthe Field Evolving?

The field is evolving, as indicated by theways in which selected aspects of the dataset changed over time. These trends arepresented next.

41 27

CHAPTER 3

Table 4. Characteristics of Participants,Intervention Agents, and Settings

Characteristic Number of cases Percentage of databaseGender

Male 150 67.6Female 72 32.4

Age in years

0-4 27 11.75-12 78 33.913-19 53 23.0>20 72 31.3

Diagnosis

Mental retardation (MR) 114 51.4Autism (Aut) 25 11.3MR + Aut 34 15.3MR +/or Aut plus other diagnoses 49 22.1

Level of retardation

Mild 32 16.7Moderate 36 18.8Severe 64 33.3Profound 60 31.3

Type of problem behavior

Aggression 51 22.2Self-injurious behavior 78 33.9Property destruction 6 2.6Tantrums 11 4.8Combinations 84 36.5

Intervention agent

Typical 99 44.2Atypical 120 53.6Both typical and atypical 5 2.2

Intervention setting

Typical 79 34.3Atypical 150 65.2Both typical and atypical 1 .4

Note: Number of cases does not always sum to 230 due to missing data.

28 4'2

RESULTS

Size of the DatabaseFigure 1 shows the data on the total numberof articles, participants, and outcomesacross 4-year blocks of time. In this andall subsequent figures, the numbersdirectly above each bar graph refer to theraw data, and the height of the bars refersto the data converted into percentages.Inillustration, consider the data on articlespublished. Of the 109 articles, 24 (22%)were published between 1985 and 1988,29 (26.6%) between 1989 and 1992, and56 (51.4%) between 1993 and 1996. Thereis a substantial increase in the number ofPBS articles published over time. Itshould be noted there were no patterns tothe exclusion of articles across years. Thatis, our criteria did not result in morearticles being excluded from early yearsthan from recent years. Similar increasesare seen for both the number of partici-pants and the number of outcomes.

DemographicsFigure 2 shows the percentage of partici-pants, by gender, in each 4-year block oftime. In this and all subsequent figures(as was evident from Table 4), the datapresented are for those cases on whichinformation was available. Thus, althoughthere were 230 participants in all, thegender for 8 of them was not identified(missing data), so percentages werecomputed for 222 participants ratherthan 230. Again, for this and subsequentfigures, there was generally an increase inthe number of participants over time,because greater numbers of articles werepublished over time. The importantinformation, therefore, concerns therelative proportion of males to femalesover time. In each block of time,maleswere approximately twice as numerous as

females.

60

56

40E04.

29

(1J24

L") 20 4.)

"I "

114 180

Articles Participants Outcomes

1985-1988

1:211989-1992

1993-1996

Figure 1. Percentage of total articles (N = 109), participants (N = 230), and outcomes

(N = 366) in each 4-year block of time. Numbers over each bar graph refer to the raw data

(frequency counts).

4,3 29

CHAPTER 3

40

c.)

(73 30

ce

o 20

V-10onZio

10 4.)

0

1985-1988

33

M F

1993-1996

73

1989-1992

44

38

19

M FGender

M F

Figure 2. Percentage of total participants (N = 222) by gender in each4-year block of time. Numbers over each bar graph refer to the raw data.

25cn

CC

20

a.L.,ett

t"'' 150

E-1

0a)ori

5Lo

0.4

1985-1988

2410 20

0

21711/ .11

n1.

13

CZN ONO A

1989-199225

12

Age (Years)

1993-199636

15

3330

71IM. 1==i INII'cl" CT 0

I ,C" ("A0 I I A1r1 rn

Figure 3. Percentage of total participants (N = 230) by age in each 4-year block of time.Numbers over each bar graph refer to the raw data.

30 44

RESULTS

Figure 3 shows the percentage ofparticipants, by age, in each 4-year blockof time. Generally, elementary-school-agechildren (5-12 years old), adolescents (13-19 years old), and adults (20 years andolder) were well represented in each timeblock. Preschool children were generallyless well represented.

Figure 5 shows the percentage ofparticipants by level of mental retarda-tion. Participants associated withprofound and severe levels were generallymore numerous in each block of timethan those with moderate and mild levelsof retardation.

30

at's 20

1985-1988

1913

;)0ca 0-

1989-199234

13 10

1993-199661

r=4

Diagnosis

19

0as

26

Figure 4. Percentage of total participants (N = 222) by diagnosis in each 4-year block of time.

Numbers over each bar graph refer to the raw data. MR = mental retardation; Aut = autism;

Both = MR + Aut; Plus = MR eWor Aut plus other diagnoses.

Figure 4 shows the percentage ofparticipants, by diagnosis, over time.Participants with a diagnosis of mentalretardation were most numerous in eachtime block with no clear trends evidentfor the other diagnoses.

Figure 6 shows the percentage ofparticipants by types of problem behav-ior. In the first time block, aggression,self-injurious behavior, and combinationsof problems were equally numerous. Overtime, self-injurious behavior andcombinations gradually became morenumerous than aggression. Propertydestruction and tantrums remained atlow levels across blocks of time.

; 4531

CHAPTER 3

20

1:164

0.1 15

a 100on

a)

0.4

1985-1988

13 13

u2

1989-1992

14 14

"MEI.

12 12

33

37

'mm

1993-1996

19 18

1MM 1MM.P S Mod M P S Mod M P S Mod M

Level of Retardation

Figure 5. Percentage of total participants (N = 192) by level of retardation in each 4-year blockof time. Numbers over each bar graph refer to the raw data. P = profound; S = severe; Mod =moderate; M = mild.

20(#)

15st,71

4-d10

1985-19880.1)tit) 15 14 15

z

3

1989-199226

1993-199643 43

2

WO co a) E-4 ..0 on = (:;) E-, -0 1:'.0 as L1 E-,di ) (7) C: 5 " § d° (.7 f'-' § bA p..4 c

"t CI) tV V VType of Problem Behavior

Figure 6. Percentage of total participants (N = 230) by type of problem behavior in each 4-yearblock of time. Numbers over each bar graph refer to the raw data. Agg = aggression; SIB = self-injurious behavior; PD = property destruction; T = tantrums; Comb = combinations of types.

32 46

RESULTS

30

c° 20at.

`.4`

ti)or10

Z1

1985-1988

223

5

-

1989-199241

19

- "IMMLf=1. al. =E-4 0d

1993-1996

57- 54

:

Type of Intervention Agent

Figure 7. Percentage of total participants (N = 224) by type of intervention

agent in each 4-year block of time. The numbers overeach bar graph refer to the

raw data. Typ = typical intervention agents; Atyp = atypical intervention

agents; Both = both typical and atypical agents.

Figure 7 shows the percentage ofparticipants involved with different typesof intervention agents. Except for 1989-1992, typical and atypical agents wereinvolved to the same degree. The involve-ment of both typical and atypical agentsfor the same participant was rare.

Figure 8 shows the percentage ofparticipants by type of interventionsetting. Although participants were mostlikely to be seen in atypical settings ineach block of time, the gap between theuse of atypical and typical settingssteadily narrowed over time. The use ofboth atypical and typical settings for thesame participant was rare.

4 33

CHAPTER 3

34

U)

'12

0.1 20

i-O

to 10

O

30

1985-198838

1993-199665

1989-199247 47

20

71.

Type of Intervention Setting

Figure 8. Percentage of total participants (N = 230) by type of interventionsetting in each 4-year block of time. The numbers over each bar graph refer to theraw data. Typ = typical intervention settings; A typ = atypical interventionsettings; Both = both typical and atypical settings.

50

§ 40

711 300

ty 20a)cis

104)1.

1985-1988

45

1989-1992

68

36

1993-1996153,

27

_LIYes No Yes No Yes No

Prior Assessment Carried Out

Figure 9. Percentage of total outcomes (N = 366) by prior assessment in each4-year block of time. The numbers over each bar graph refer to the raw data.

48

RESULTS

Figure 10. Percentage of total outcomes (N = 266) by typeof assessment in each 4-year block of

time. Numbers over each bar graph refer to the raw data. 10 = indirect observation; DO = direct

observation; FA = functional analysis; Comb = combinations of 2 or more types of assessment.

AssessmentFigure 9 shows the percentage of out-comes for which a prior assessment was(yes) or was not (no) carried out. Therewas a dramatic increase, over time, in theproportion of outcomes that wereassociated with a prior assessmentcompared to those that were not.

Figure 10 shows the percentage ofoutcomes for which a specific type ofassessment was completed (i.e., indirectobservation only, direct observation only,functional analysis only, and combina-tions of the preceding types of assess-ment). Over time, combined assessmentsand functional analysis became propor-tionately greater-than indirect and directobservation.

Additional analyses permitted ananswer to the question of whether

assessment practices varied with respectto type of intervention agent, type ofintervention setting, and whetherintervention was subsequently carried outin all relevant contexts. With respect totype of intervention agent, for 257outcomes it was possible to determineboth the type of agent involved and thetype of assessment involved. We con-trasted the assessment data generatedfrom formal functional analysis (FA-based) with the data based on less formalindirect and direct observation (not FA-based). FA-based procedures consisted ofFA alone or FA used in combination withindirect or direct observation. Non-FA-based procedures excluded the use of FAand involved indirect and/or directobservation only. There were 200 out-comes that.were FA-based. Of these,typical agents were associated with 73

49 35

CHAPTER 3

outcomes (36.5% of the total), atypicalagents with 124 outcomes (62.0% of thetotal), and both typical and atypicalagents with 3 outcomes (1.5% of thetotal). There were 57 outcomes that werenot FA-based. Of these, typical agentswere associated with 41 outcomes (71.9%of the total), atypical agents with 14outcomes (24.6% of the total), and bothtypical and atypical agents with 2outcomes (3.5% of the total). In sum, FA-based outcomes were more likely to beassociated with atypical agents, and non-FA-based outcomes were more likely to beassociated with typical agents.

With respect to type of interventionsetting, for 266 outcomes it was possibleto determine both the type of settinginvolved and the type of assessmentinvolved. There were 203 outcomes thatwere FA-based. Of these, typical settingswere associated with 64 outcomes (31.5%of the total), and atypical settings with139 outcomes (68.5% of the total). Therewere 63 outcomes that were not FA-based.Of these, 32 outcomes (50.8% of the total)were associated with typical settings, 30outcomes (47.6% of the total), withatypical settings, and 1 outcome (1.6% ofthe total) with both typical and atypicalsettings. In sum, FA-based outcomes weremore likely to be associated with atypicalsettings, and non-FA-based outcomeswere equally likely in typical and atypicalsettings.

With respect to intervention in allrelevant contexts, for 237 outcomes it waspossible to determine whether interven-tion in all relevant contexts subsequentlyoccurred, and the type of assessmentpreceding the intervention. There were179 outcomes that were FA-based. Ofthese, 28 outcomes (15.6% of the total)were associated with intervention in allrelevant contexts, but 151 outcomes

5036

(84.4% of the total) were not. Fifty-eightoutcomes were not FA-based. Of these, 29outcomes (50% of the total) wereassociated with intervention in allrelevant contexts, and an equal numberwere not. In sum, FA-based outcomeswere less likely to be associated withintervention in all relevant contexts, andnon-FA-based were associated half thetime with intervention in all relevantcontexts and half the time without suchintervention.

Overall, then, the use of functionalanalysis as an assessment tool is moreclosely associated with atypical agents,atypical settings, and a failure to inter-vene in all relevant contexts. In contrast,non-FA-based assessment is more closelyassociated with typical agents but is notdifferentially associated with type ofsetting or the presence versus absence ofintervention in all relevant contexts.

Figure 11 shows the percentage ofoutcomes associated with each type ofbehavioral motivation for the 250outcomes in which motivation (function)could be determined. In each block oftime, escape predominated over othersingle motivations, and the degree towhich this was the case increased overtime. There was also some indication of aproportionate increase over time in thenumber of outcomes associated withmultiple motivations (combinations ofmotivations).

Finally, there were only 11 outcomes(out of the total of 266) for whichassessment was repeated over time.Between 1985 and 1988, 0 outcomesinvolved repeated assessments; between1989 and 1993, there was 1 such outcome(.3% of the total); and between 1993 and1996, there were 10 (3.8% of the total).Repeated assessment became morecommon over time, though the sample

RESULTS

Figure 11. Percentage of total outcomes (N = 250) by typeof motivation in each 4-year block of

time. Numbers over each bar graph refer to the raw data. Att = attention; Esc = escape; Tan =

tangible; Sen = sensory; Comb = 2 or more types of motivation.

size was very small. Our additionalanalyses allowed an answer to thequestion of when assessment was most/least likely to be repeated. Specifically, for246 outcomes, it was possible to deter-mine the type of setting and interventionagent for which the assessment was notrepeated. With respect to settings,assessment was not repeated for 79outcomes occurring in typical settings(32.1% of the total), 166 outcomesoccurring in atypical settings (67.5% ofthe total), and 1 outcome occurring inboth a typical and atypical setting (.4% ofthe total). With respect to interventionagents, assessment was not repeated for106 outcomes occurring with typicalagents (43.1% of the total), 136 outcomesoccurring with atypical agents (55.3% ofthe total), and 4 outcomes occurring withboth typical and atypical agents (1.6% of

the total). For 11 outcomes, it waspossible to determine the type of settingand intervention agent for which theassessment was repeated. With respect tosettings, assessment was repeated for 10outcomes occurring in typical settings(90.9% of the total), I outcome occurringin atypical settings (9.1% of the total),and 0 outcomes occurring in both atypical and atypical setting. With respectto intervention agents, assessment wasrepeated for 8 outcomes occurring withtypical agents (72.7% of the total), 2outcomes occurring with atypical agents(18.2% of the total), and 1 outcomeoccurring with typical and atypicalagents (9.1% of the total). In sum, then,assessment was most likely to be repeatedwith typical agents and settings, andmost likely not to be repeated withatypical agents and settings.

51. 37

CHAPTER 3

40

c)E0 300To'

E-1 20w0cu

02, 10a-4

cs.

ST/RF

1985-1988 1989-1992

5450 50

14 14

4

1993-1996

91

24

1993-1996

Non-PBS127

1989-1992

79

1985-1988

58

H Fa. P-T-rz4 cr) c:4 rf) f:4

ac)

Type of Intervention

4

CU 0.7,0

25

0>1.1z

53

Figure 12. Percentage of total outcomes (N = 366) by type of intervention in each 4-year block oftime. The numbers over each bar graph refer to the raw data. ST = stimulus-based intervention;RF = reinforcement-based intervention; Both = combined stimulus-based and reinforcement-based intervention; Non-PBS = nonpositive behavior support procedure was/was not (Yes/No) partof the intervention.

InterventionFigure 12 shows the percentage ofoutcomes associated with various typesof interventions. The relative proportionof stimulus-basedversus reinforcement-based intervention reversed over time.Initially, reinforcement-based interven-tion predominated (1985-1988); then thetwo types of intervention were equallyapplied (1989-1992); and finallystimulus-based intervention predomi-nated (1993-1996). Combined interven-tion (i.e., both stimulus-based andreinforcement-based) showed no cleartrend relative to the other two types.Additional analyses permitted an answerto the question of when combined

385.2

intervention versus noncombinedintervention (i.e., stimulus-based only orreinforcement-based only) was most/leastlikely to occur. Specifically, there were 42outcomes for which combined interven-tion occurred. Of these, 26 (61.9% of thetotal) were carried out by typicalagentsand only 15 (35.7% of the total) werecarried out by atypical agents. There were324 outcomes for which noncombinedintervention occurred. Of these, therewere 9 outcomes for which type of agentcould not be determined, leaving adatabase of 315 outcomes. For theremaining cases (315 outcomes), thepattern previously described wasreversed. Specifically, only 128 (40.6% of

RESULTS

the total) were carried out by typicalagents, but 181 (57.5% of the total) werecarried out by atypical agents. There were7 outcomes involving both typical andatypical agents together (2.2% of thetotal).

Data were also available on type ofsetting. Of the 42 outcomes for whichcombined intervention occurred, 22(52.4% of the total) were carried out intypical settings, and 19 (45.2% of thetotal) were carried out in atypicalsettings. In 1 outcome combined inter-vention was carried out in both typicaland atypical settings (2.3% of the total).In sharp contrast, for the 324 outcomesinvolving noncombined intervention, 104(32.1% of the total) were carried out intypical settings and 220 (67.9% of thetotal) were carried out in atypicalsettings. In sum, then, combined inter-vention was most likely to be conductedby typical agents and in typical settings,and noncombined intervention was mostlikely to be conducted by atypical agentsand in atypical settings.

One final comparison concerns therelationship between type of interventionand its use in all relevant contexts. Therewere 322 outcomes for which it waspossible to determine both the type ofintervention used and whether itoccurred in all relevant contexts.There were 285 outcomes associated withnoncombined intervention. Of these, 50(17.5% of the total) were associated withintervention in all relevant contexts, and235 (82.5% of the total) were not. Therewere 37 outcomes associated withcombined intervention. Of these, 26(70.3% of the total) were associated withintervention in all relevant contexts, and11 (29.7% of the total) were not. In sum,combined intervention was most likely tooccur in all relevant contexts, whereas

noncombined intervention was leastlikely to occur in all relevant contexts.

Figure 12 also displays the data foroutcomes associated with interventionsthat included a non-PBS component (yes)in addition to a PBS component versusoutcomes not associated with interven-tions that included a non-PBS component(no) (i.e., the intervention had only PBScomponents). Consider the first block oftime (1985-1988). There were 24 out-comes associated with interventions thatincluded a non-PBS component, but 58outcomes associated with interventionsthat did not include a non-PBS compo-nent. This pattern of results, namely, thepredominance of outcomes for which PBSwas the only component of the interven-tion package, was evident across eachblock of time.

Figure 13 shows the percentage ofoutcomes associated with different typesof systems change. With respect towhether significant others were/were not(yes/no) required to change theirbehavior as a component of intervention,it is clear that the presence of such changepredominated in all three time blocks; thedegree of this predominance dramaticallyincreased in the final time block. Withrespect to whether environmentalreorganization was/was not (yes/no) acomponent of intervention, it is clear thatthe presence of such change was rareover time.

Figure 14 shows the percentage ofoutcomes in which intervention was/wasnot (yes/no) carried out in all relevantcontexts. Over time, there was a dramaticincrease in the number of outcomesassociated with a failure to intervene inall relevant contexts, but only a modestincrease in those associated withintervention in all relevant contexts.Additional analyses allowed us to

53 39

CHAPTER 3

50

5o 4064O

30

20-0

on

b. 10a

SIG

1989-19921985-1988

5953

245

V) 0Z

1993-1996

156

24

ENV

1985-1988

79

3

1989-199298

1993-1996

169

V) 0()")

V)

Z0

'1?

Type of Systems Change

V) 0Z

11

11116ce) 0

Z

Figure 13. Percentage of total outcomes (N = 366) by type of systems change in each 4-year blockof time. The numbers over each bar graph refer to the raw data. SIG = significant others wererequired to change their behavior; ENV= environmental reorganization was undertaken.

determine when intervention was most/least likely to be carried out in all relevantcontexts. Specifically, for 241 outcomes, itwas possible to determine the type ofsetting and type of intervention agent forwhich intervention was not (no) carriedout in all relevant contexts. With respectto settings, intervention was not carriedout in all relevant contexts for 70 out-comes occurring in typical settings(29.0% of the total), and 171 outcomesoccurring in atypical settings (71.0% ofthe total). With respect to agents, suchintervention was not carried out for 82outcomes occurring with typical agents(34.0% of the total), 157 outcomesoccurring with atypical agents (65.1% ofthe total), and 2 outcomes occurring withboth typical and atypical agents (1.0% ofthe total). For 72 outcomes, it was

40 54

possible to determine the type of settingand type of intervention agent for whichintervention was (yes) carried out in allrelevant contexts. With respect to settings,intervention was carried out in allrelevant contexts for 42 outcomesoccurring in typical settings (58.3% ofthe total), 29 outcomes occurring inatypical settings (40.3% of the total), and1 outcome occurring in both typical andatypical settings (1.4% of the total). Withrespect to agents, such intervention wascarried out for 62 outcomes occurringwith typical agents (86.1% of the total), 5outcomes occurring with atypical agents(6.9% of the total), and 5 outcomesoccurring with both typical and atypicalagents (6.9% of the total). In sum, then,intervention in all relevant contexts wasmost likely to occur with typical agents

RESULTS

40

E30

O

E. 20

b.13cet

4cu 10o-1

1985-198851

1989-199265

14

1993-1996130

40

aIntervention in All Relevant Contexts

Figure 14. Percentage of total outcomes (N = 322) by intervention in all relevantcontexts in each 4-year block of time. The numbers over each bar graph refer tothe raw data. Yes = intervention was carried out in all relevant contexts; No =intervention was not carried out in all relevant contexts.

and settings, and most likely not to occurwith atypical agents and settings.

OutcomesFigure 15 shows the percentage ofoutcomes associated with diversemeasures of generalization. Typically, datawere based on small numbers of out-comes. The percentage of outcomes forstimulus generalization showed no trendover time, but that for response generali-zation showed a slight increase over time.The percentage of outcomes reported formaintenance for 1 to 5 months' durationrose steadily over time, but that for 6 to 12months, and 13 to 24 months showed notrend. Likewise, no trends were seen forthe percentage of outcomes associated

with a stated goal of lifestyle change, norwas there a trend with respect to the useof planned intervention intended toproduce lifestyle change. Finally, thedirect measurement of lifestyle changefollowing intervention showed a smallincrease over time.

Figure 16 shows the percentage ofoutcomes associated with diversemeasures of social validity. In all cases,data were based on very small numbersof outcomes. The percentage of outcomesfor which there was a feasibility questionshowed a modest increase over time, butpercentage for the desirability questionshowed no trend. Likewise, we noted asmall increase for the question concern-ing the acceptability of the level of

55 41

CHAPTER 3

0

14

1

9

52

19

31

26

42

1=IMIMIMM

24_.,

10 10

SG RG MI M6

16

5k

19

2 riNILI 0

M13 LG

Type of Generalization

1985-19881989-19921993-1996

9

5

01:1 X01LI LS

Figure 15. Percentage of total outcomes (N = 366) by type of generalization in each 4-year blockof time. The numbers over each bar graph refer to the raw data. SG = stimulus generalization; RG= response generalization; M1, M6, and M13 = maintenance after 1-5 months, 6-12 months, and13-24 months respectively; LG = stated goal of lifestyle change; LI = lifestyle change interventionimplemented; LS = lifestyle change success measured.

Figure 16. Percentage of total outcomes (N = 366) by type of social validation in each 4-yearblock of time. The numbers over each bar graph refer to the raw data. Feas = feasibility;Des = desirability; Acc = acceptability; Life = lifestyle change.

42 56

RESULTS

30

(1g

20

00

az 10Ez

VI

23

viojA

7\ .11

Changes in Frequency

40

30

20

10

VI

37

111.

VI

A

31

vio3In

r-A-4

25

22

oVI 74

.75

A ;Changes in Percentages

Figure 17. Number of outcomes for whichchanges in positive behavior were measured.The bar graph above represents the data foroutcomes measured in terms of changes infrequency (N = 43), and the bar graph belowrepresents the data for outcomes measured interms of changes in percentages (N = 122).Numbers over each bar graph refer to thenumber of outcomes associated with eachcategory of change.

problem-behavior reduction, but therewas no trend concerning whethersignificant others perceived the interven-tion to have produced meaningfullifestyle change.

How Effective is PBS?

Changes in Positive BehaviorPBS intervention involves strategiesdesigned to make socially desirableresponses (positive behaviors) moreprobable. We identified data on changesin positive behavior for 165 of the 366outcomes (45.1% of the total). Becausethe baseline for positive behavior was, ina substantial number of cases, 0, it wasnot possible to compute a percentageincrease in positive behavior relative tobaseline (because division by 0 isimpermissible). Therefore, as Figure 17shows, difference scores based on the rawdata were used. For example, if the datawere reported as frequencies (as theywere in 43 cases), we simply subtractedthe mean of the last three baseline points(typically 0) from the mean of the lastthree intervention points. In illustration,if the baseline mean was 0 communica-tive acts, and the intervention mean was30 communicative acts, then the differ-ence score (i.e., increase in communica-tive acts) was 30. In Figure 17, this dataoutcome would contribute to the bargraph category >10 & 50. We used thesame formula for data reported aspercentages (as they were in 122 cases).In illustration, if the baseline mean was10% intervals of cooperative behavior,and the intervention mean was 90%intervals of cooperative behavior, then thedifference score (i.e., increase in percent-age of intervals containing cooperativebehavior) was 80%. Figure 17 displaysboth the frequency and percentage data

57 43

CHAPTER 3

on positive behavior. In either category,frequency or percentages, interval dataare not directly comparable. For example,a frequency increase of 50 communica-tive acts is not "twice as good" as afrequency increase of 25 cooperative acts.The two types of behavior are notequivalent. Nonetheless, the data inFigure 17 are presented to provideinformation to answer the generalquestion of whether positive behaviorsdid indeed increase following PBS. Thefigure shows clearly that there were only ahandful of cases in which the level ofpositive behavior remained the same ordecreased following intervention.Overwhelmingly, for both the frequencyand percentage data, positive behaviorsincreased following the use of PBS, albeitthe degree of increase varied widely frommodest to substantial.

Distribution of OutcomeEffectiveness for PBS

InterventionsTable 5 shows the distribution of outcomeeffectiveness in terms of percentage

reduction in problem behavior frombaseline. Across all PBS interventions(i.e., irrespective of whether they werestimulus-based or reinforcement-based),approximately two-thirds of the outcomes(68.0%) were associated with substantialreductions in problem behavior of 80% ormore from baseline levels. A similarpattern was seen when the total databasefor intervention was subdivided: out-comes that were stimulus-based versusthose that were reinforcement-based.Specifically, 66.5% of the outcomes thatwere stimulus-based and 71.6% of theoutcomes that were reinforcement-baseddemonstrated 80% or more reduction inproblem behavior from baseline levels. Incontrast, across the intervention catego-ries, a minimal number of outcomes wereassociated with small reductions (i.e., lessthan 20% reduction from baseline levels).For approximately 6 to 8% of the out-comes, depending on the type of inter-vention, an increase in problem behaviorwas noted. It would be interesting toknow whether there are specific variablesthat predict the few instances in whichPBS produced minimal or negative

Table 5. Distribution of Outcome Effectiveness:Reduction in Problem Behavior

Frequency of outcomes by percentage reduction in problem behaviorIncrease in

100% 90- 80- 60- 40- 20- 0- problem Total99% 89% 79% 59% 39% 19% behavior outcomes

All PBS-based outcomes 97 92 60 51 23 10 6 27 366

ST-based outcomes 44 52 35 31 12 6 5 12 197

RF-based outcomes 67 48 36 25 12 4 2 17 211

Note: PBS = any intervention that had an ST and/or RF component; ST = any stimulus-basedintervention (i.e., ST with/without RF, with/without non-PBS); RF= any reinforcement-basedintervention (i.e., RF with/without ST, with/without non-PBS). Thus, outcomes associated with STand RF in combination would appear once under ST-based and again under RF-based.

5844

RESULTS

effects. But the articles reviewed providedinsufficient information to make such adetermination. A reasonable workinghypothesis is that intervention failures aremore a function of systems than oftechniques. Thus, uncooperative staff, toomany staff changes, lack of respite forparents, and insufficient time on the partof teachers are all plausible examples ofthe types of systemic factors likely toimpact negatively on interventionoutcomes, a point taken up again later. Atpresent, the existence of unfavorableoutcomes should serve as a prompt forresearchers to expand their search forthose systems variables that appear toinfluence outcome.

Success Rates for PBSInterventions Pooled Across

OutcomesTable 6 shows the success rates forvarious types of PBS interventions. Astringent criterion was used to definesuccess: An outcome had to reflect a 90%or greater reduction in problem behaviorfrom baseline levels to be considered asuccess. The success rate was generallywithin 5 points of 50%, irrespective of thetype of intervention employed.

Almost three-quarters of theinterventions (72.1%) did not include anon-PBS component; however, thepresence versus absence of a non-PBS

Table 6. Success Rates for PBS InterventionsPooled Across Outcomes

Pooled dataNumber ofoutcomes

Percentage ofgrand total

Numberof successes

Percentage ofof successes

All interventions (grand total) 366 100 189 51.6

All that excluded non-PBS 264 72.1 136 51.5

All that included non-PBS 102 27.9 53 52.0

All single interventions 324 88.5 167 51.5

All ST interventions 155 42.3 74 47.7

All ST excl. non-PBS 129 35.2 59 45.7

MI ST incl. non-PBS 26 7.1 15 57.7

All RF interventions 169 46.2 93 55.0

All RF excl. non-PBS 104 28.4 62 59.6

All RF incl. non-PBS 65 17.8 31 47.7

All combined interventions 42 11.5 22 52.4

All ST + RF that excl. non-PBS 31 8.5 15 48.4

All ST + RF that incl. non-PBS 11 3.0 7 63.6

Note: ST = stimulus-based interventions; RF = reinforcement-based interventions; excl. =excluding; incl. = including. Success was defined as a 90% or more reduction in problem behaviorfrom baseline.

59 45

CHAPTER 3

Table 7. Types of Non-PBS Interventions Used

Type of non-PBS Number of outcomes

Extinction

DRO

Punishment

Reprimand

Forced compliance

Response cost

Overcorrection

Timeout

Brief restraint

Water mist

58

6

49

4

6

8

6

7

17

1

Note: The total number of outcomes added up to more than 102, because some outcomes wereassociated with several types of non-PBS.

component had little effect on interven-tion success (i.e., each intervention typeproduced success rates close to 50 %).Asubstantial majority of interventions(88.5%) used stimulus-based strategiesor reinforcement-based strategies, but notboth together. The two strategies wereused about equally often. The successrates for the two were comparable (i.e.,47.7% for stimulus-based, and 55.0%for reinforcement-based). Only a smallminority of interventions (11.5%)combined stimulus-based withreinforcement-based strategies. Thesecombined interventions also producedsuccess rates of about 50%. The oneexception to this general finding involvedcombined interventions that included oneor more non-PBS interventions; theseproduced a 63.6% success rate. However,this rate was based on the smallestnumber of outcomes for any of the PBSintervention categories reported and may,therefore, reflect sampling error ratherthan a unique property of this type ofintervention.

46 60

Because non-PBS interventions wereassociated with approximately onequarter (102) of the 366 outcomes, it ishelpful to clarify the specific nature ofthese interventions. Table 7 presents abreakdown of the types of non-PBSinterventions used. About half theoutcomes were associated with the use ofextinction and half with the use ofpunishment. A handful were associatedwith DRO. The punishment category wasfurther broken down into seven proce-dures. Of these, only brief physicalrestraint was associated with more thaneight outcomes.Among the sevenpunishment procedures, only water misthas been considered in the literature(Scotti et al., 1991) as a highly intrusiveprocedure, and it was associated withonly 1 outcome of the 102 generated.

RESULTS

Table 8. Generalization Measures of Outcome Effectiveness

Frequency of outcomes by percentage reduction in problem behavior

100% 90-99%

80-89%

60-79%

40-59%

20-39%

0-19%

Increase inproblembehavior

Totaloutcomes

Stimulus generalizationoutcomes 5 8 3 2 1 0 0 1 20

Frequency of outcomes by percentage increase in appropriate behavior

100% 90-99%

80-89%

60-

79%

40-59%

20-

39%0-

19%

Decrease inappropriatebehavior

Totaloutcomes

Response generalizationoutcomes 0 1 2 3 4 1 1 1 13

Stimulus and ResponseGeneralization

Stimulus generalization refers to thedegree to which intervention effectstransferred from the original interventionsituation to other situations involvingnew intervention agents, settings, andtasks. Anecdotal reports of stimulusgeneralization were noted for 29 out-comes. Data-based reports of stimulusgeneralization were noted for an addi-tional 24 outcomes (i.e., in 6.6% of thetotal database). However, in 4 of thesecases, the baseline data were inadequatefor computation purposes (i.e., fewer than3 data points were reported). Thedistribution of the remaining 20 out-comes is shown in Table 8 (top half). Ascan be seen, in a number of cases,problem behavior decreased frombaseline levels, sometimes substantially,in situations that were not a direct targetof intervention.

Response generalization refers to thedegree to which intervention effects

transferred from the initial behavioraltarget of intervention to other aspects ofan individual's behavior repertoire thatwere not targeted for intervention. In thearticles examined, response generaliza-tion consisted exclusively of changes in avariety of appropriate behaviors (e.g.,social skills, academics). Anecdotalreports of response generalization werenoted for 25 outcomes. Data-basedreports of response generalization werenoted for an additional 26 outcomes (i.e.,in 7.1% of the total database). However, in13 of these cases, baseline data wereinadequate (i.e., fewer than 3 data pointswere reported), or baseline data werereported but postintervention data werenot. The distribution of the remaining 13outcomes is shown in Table 8 (bottomhalf). As can be seen, in a number ofcases, modest increases in appropriatebehavior were observed relative tobaseline, even though such behavior wasnot a target of intervention.

6147

CHAPTER 3

Table 9. Maintenance Measures of Outcome Effectiveness

Frequency of outcomes by percentage reduction in problem behaviorIncrease in

100% 90-99%

80-89%

60-79%

40-59%

20-39%

0-19%

problem Totalbehavior outcomes

1 5 months 62 6 8 14 1 2 1 5 99

6 12 months 19 9 5 5 4 2 0 0 44

13 24 months 5 0 1 0 1 0 0 0 7

Note: There were no maintenance outcomes reported for follow-up periods of 25 months or more.

MaintenanceAs noted earlier, maintenance is definedas the degree to which interventioneffects were documented to last over time(intervention durability). Table 9 displaysthe outcome data for maintenance,measured as percentage reduction inproblem behavior from baseline, noted atspecific follow-up periods, namely, at 1 to5 months, 6 to 12 months, 13 to 24months, and 25 months or more. Thenumber of outcomes noted for each of thefollow-up periods just mentioned was 99,44, 7, and 0, indicating a dramaticdecrease in available data with eachsuccessive increase in maintenanceduration. In general, most outcomesclustered at the high end of the percent-age reduction in problem behavior. Thatis, good maintenance effects wereobserved for a substantial majority ofoutcomes. Using a success criterion of90% or more reduction in problembehavior from baseline, the success ratesat 1 to 5 months, 6 to 12 months, and 13to 24 months were 68.7%, 63.6%, and71.4% respectively, suggesting thatintervention effects were quite durable, atleast for those cases in which mainte-nance outcome data were available.

48 62

Lifestyle ChangeAs the ultimate purpose of PBS is toenable individuals to live more normal-ized lives, lifestyle change is an importantindex of effectiveness. Surprisingly, thedatabase for this outcome measure wasextremely small. Lifestyle change was astated intervention goal for only 24 out ofthe 230 participants in the sample (i.e.,10.4% of the sample). A formal interven-tion directed specifically at improvinglifestyle was recorded for 8 out of the 230participants (i.e., 3.5% of the sample).Finally, success in improving lifestyle wasmeasured for only 6 participants (i.e.,2.6% of the sample). Of these, anecdotal(nonquantified) improvement was notedfor 4 participants. Data (percentageimprovement from baseline) were takenon only 2 participants, and showed a 100%improvement with respect to increasedengagement in community activities.

Social ValiditySocial validity refers to whether signifi-cant others (e.g., parents, teachers, jobcoaches) perceive the intervention and itseffects to be worthwhile. (It should benoted that no study involved asking

RESULTS

persons with disabilities what theythought about the social validity of theinterventions that they experienced.)Again, various aspects of social validitywere assessed for a very small number ofparticipants. With respect to the genericfeasibility question (e.g.,"Would you beable to use this intervention strategy?"),data were available on 14 of the 230participants. In 8 cases, these data wereanecdotal; teachers (for example) statedthat they had continued to use theintervention over time. For the remaining6 participants, Likert-scale data indicatedthat whereas intervention agents thoughtthey would seldom/never be able to usethe intervention before they had beentrained, afterwards they felt they wouldvery much/always be able to use theintervention.

In terms of the generic desirabilityquestion (e.g.,"Would you be willing touse this intervention strategy?"), datawere available on 12 participants. In 7cases, these data were anecdotal state-ments from support people affirming thedesirability of the intervention used overall other proposed or previously at-tempted interventions. For the remaining5 participants, Likert-scale data wereavailable. These data showed that, prior totraining, support people felt they wouldchoose to implement the interventionnever/not at all; following training, theywould choose the intervention verymuch/always.

With respect to the generic accept-ability question (e.g.,"Does the interven-tion strategy reduce problem behavior toa level that is acceptable to you?"), datawere available for 29 participants. In 4cases, these data were anecdotal. Supportpeople were interviewed and assertedthat each of the participants had "im-proved" In 25 cases, data were available

r.

but the metric used varied greatly fromstudy to study. For example, in 6 cases,support people were asked to rate thedegree to which (strongly agree/agree/uncertain/disagree/strongly disagree)they felt that the participant's problembehavior had improved. In 4 of thesecases, 65% of the support people stronglyagreed/agreed; in the other 2 cases, 83%strongly agreed/agreed. In 2 cases,support people were asked to rateproblem behavior on a 6-point scale(where 1 represented low improvementand 6, high improvement). In 1 case, therating improved from 1.8 to 2.9; in theother case, from 2.2 to 4.1. In 13 addi-tional cases, significant others were askedto rate problem behavior on a 7-pointscale (where 1 represented low improve-ment and 7, high improvement). For 8 ofthese cases, the ratings improved from apreintervention mean of 2.1 to a post-intervention mean of 5.1. For 5 of thesecases, the postintervention mean was 5.9,but no preintervention data were reported,complicating interpretation. Finally, in 4cases, significant others were asked to usea 10-point scale (where 1 was little/noimprovement and 10 was maximumimprovement/no problem behavior). Thepreintervention mean was 2.1, and thepostintervention mean was 7.2.

For the generic lifestyle changequestion (e.g.,"Does this interventionstrategy produce effects that increaseopportunities to live, work, go to school,recreate, and socialize with typical peersand significant others in typical commu-nity settings?"), data were available for8 participants. In 6 of these cases, lifestylechange was rated on a 7-point scale(where 1 represented little improvementand 7, major improvement). Across 3 ofthese cases, the preintervention mean was2.3, and the postintervention mean, 6.0.

6:3 49

CHAPTER 3

In the other 3 cases, the postinterventionmean was 3.2, but there were nopreintervention data, again complicatinginterpretation. In the remaining 2 cases,significant others were interviewed byphone 2 years after intervention andasked whether they agreed/disagreed thatfamily interactions had improved as aconsequence of intervention and that thechild was more accepted in the cornmu-nity; 86% of the interviewees agreed withthe preceding statement.

What FactorsModulate the

Effectiveness of PBS?The data on intervention effectiveness,just examined, make clear that stimulus-based and reinforcement-based interven-tions produced very similar outcomes. Ofcourse it is poSsible that each type ofintervention might, nonetheless, producea unique interaction with modulatorvariables such as demographic character-istics or type of assessment used.However, from a clinical perspective,there are no pure interventions. That is,clinically, stimulus-based interventionalways encompasses changes in reinforce-ment parameters (e.g., redesigning acurriculum will also produce shifts in theallocation of reinforcers across behav-iors), and reinforcement-based interven-tion always encompasses changes instimulus parameters (e.g., a reinforcer is,itself, a stimulus that can be discrimina-tive for specific responses). These factsled to our decision to pool all the data forthe two types of intervention in order tostudy how various factors modulate theeffectiveness of PBS-based interventionas a whole. Because clinical practicealways involves combining elements ofstimulus- and reinforcement-based

50 64

intervention, we assume that pooling thedata across the two types of interventionprovides one plausible estimate of theoutcomes of comprehensive positivebehavior support applied to an indi-vidual.

Influence of DemographicVariables

Table 10 displays the relationship betweendifferent demographic variables andintervention effectiveness across all PBS-based outcomes. The first entry, forgender, shows that the percentage ofsuccess was not strongly influenced bygender per se; both males and femalesshowed equivalent success rates (i.e.,within a few points of 50 %).

The data on age show considerablevariation, from a success rate of 4.1.1% foradults (i.e., participants over 20 years ofage) to a success rate of 63.7% foradolescents, with preschool and school-age children falling between these twoextremes. The relationship between age andsuccess rate appears linear up to adoles-cence, and drops off sharply thereafter.

With respect to diagnosis, thesuccess rate was highest for thosediagnosed with a combination of mentalretardation plus autism (59.1%), andlowest for those diagnosed with autismalone (43.2%). The other two diagnosticcategories were associated with successrates that fell between those just given. Itis interesting to note that the diagnosticcategories associated with the smallestnumber of outcomes (autism, and mentalretardation plus autism) were also thetwo whose success rates deviated mostsharply from the typical finding (in thissynthesis) of a 50% success rate, suggest-ing the influence of sampling error raiherthan formal diagnosis per se.

RESULTS

Table 10. Relationship Between Demographic Variablesand Outcome Effectiveness

CharacteristicNumber ofoutcomes

Number ofsuccesses

Percentage ofsuccesses

Gender

Male 235 128 54.5

Female 117 60 51.3

Age in years

0-4 37 17 45.9

5-12 129 70 54.3

13-19 88 56 63.7

>20 112 46 41.1

Diagnosis

Mental retardation (MR) 201 106 52.7

Autism (Aut) 37 16 43.2

MR + Aut 44 26 59.1

MR +/or Aut plus other diagnoses 72 35 48.6

Level of retardation

Mild 51 23 45.1

Moderate 53 35 66.0

Severe 100 53 53.0

Profound 110 58 52.7

Type of problem behavior

Aggression 90 50 55.5

Self-injurious behavior 132 74 56.1

Property destruction 13 5 38.5

Tantrums 13 8 61.6

Single type 248 137 55.2

Combination of types 118 52 44.1

Note: Number of outcomes does not always sum to 366 due to missing data.

Level of retardation produced apattern of results similar to those justnoted for diagnosis. Specifically, the tworetardation levels associated with thegreatest number of outcomes (severe andprofound) had success rates within a fewpoints of 50%. Those associated with the

fewest number of outcomes (mild andmoderate) had success rates that deviatedfrom the 50% rate (i.e., 45.1% and 66.0%respectively), again suggesting theinfluence of sampling error rather thanlevel of retardation per se.

51

CHAPTER 3

Table 11. Relationship Between Assessment Variablesand Outcome Effectiveness

FactorNumber ofoutcomes

Number ofsuccesses


Type of assessment

Informal observation 19 8 42.1

Formal direct observation 10 6 60.0

Functional analysis 105 64 61.0

Combined assessments 132 79 59.8

Assessment conducted 266 157 59.0

No assessment conducted 100 32 32.0

Type of function

Attention 32 20 62.5

Escape 122 75 61.5

Tangible 27 15 55.6

Sensory 17 4 23.5

Multiple 52 33 63.5

Assessment repeated

Yes 11 4 36.4

No 255 155 60.0

Assessment information used

Yes 231 135 58.4

No 35 22 62.9

Type of problem behavior producedsuccess rates near 50% for those behav-iors associated with numerous outcomes(i.e., aggression and self-injuriousbehavior), and considerable deviationsfrom 50% for those behaviors associatedwith few outcomes (i.e., propertydestruction and tantrums). Importantly,it was possible to cumulate all outcomesinvolving a single type of problembehavior (e.g., aggression alone, self-injurious behavior alone, etc.) andcompare these to all outcomes involving

6652

combinations of problem behavior (e.g.,aggression plus self-injurious behavior,property destruction plus aggression,etc.). The success rate for outcomes basedon single types was 55.2%, whereas therate for combinations was only 44.1%.Because this comparison involved largenumbers of outcomes, the approximately11% spread between the two success ratessuggests that successfully intervening oncombinations of problem behavior maybe somewhat more difficult than inter-vening on single types of problem behavior.

RESULTS

Influence of Assessment.Variables

Table 11 displays the relationship betweendifferent assessment variables andintervention effectiveness. The datasuggest that conducting an assessmentcan have a considerable impact onsuccess rates. Thus, conducting somekind of assessment (i.e., informalobservation, formal direct observation,functional analysis, or any combination ofthese three) was associated with a successrate of 59.0%, whereas a failure toconduct assessment was associated with asuccess rate of 32.0%. Functional analysisand combined assessment each yieldedsuccess rates close to 60%, as did formaldirect observation. Informal observationproduced the lowest success rate (42.1%);however, both informal observation andformal direct observation were based onvery small numbers of cases, and theseresults may reflect sampling error.

The success rates associated withtype of function were always within a fewpoints of 60% irrespective of whether thefunction was attention, escape, tangible,or multiple types (i.e., the problembehavior was maintained by more thanone factor). The one exception to thisfinding involves the sensory function,which produced a dramatically lowersuccess rate (23.5%). The latter successrate may reflect the fact that sensoryfunctions are nonsocial in nature,whereas all the other functions aresocially mediated. Alternatively, thedifference may merely reflect samplingerror given the fact that the data onsensory function were based on a verysmall number of outcomes.

Table 11 also addresses the issue ofwhether repeating functional assessmentover time influences success rates. The

data showed that repeated assessmentswere associated with lower success rates(36.4%) than single assessments (60.0%).However, the former rate is difficult tointerpret because it is based on only 11outcomes, again plausibly reflectingsampling error. Alternatively, this findingmay reflect the fact that repeatedassessments were more likely if the initialintervention proved unsuccessful or if thefactors controlling the behavior weremore complex.

Finally, the data show that successrates were equivalent and high whether ornot the assessment data were used todesign an intervention. However, thefailure to use the assessment data was acharacteristic of only 35 outcomes. Inaddition, it was not possible to determinefrom the published reports whether mereknowledge of assessment data nonethe-less influenced choice of intervention insubtle ways not articulated by theinvestigator.

Systems ChangeTable 12 shows that the two systems-change variables may have had an effecton success rates. Thus, when significantothers (e.g., teachers, parents, jobcoaches) altered their own behavior aspart of a systematic intervention, thesuccess rate was 55.2%, which was higherthan the 41.8% success rate associatedwith interventions that did not involvebehavior change on the part of significantothers. Likewise, environmental reorgani-zation was associated with a success rateof 65.0%, higher than the rate obtainedwithout such organization (50.9%).However, the latter finding must beinterpreted cautiously, because there wereso few outcomes associated with environ-mental reorganization.

53

CHAPTER 3

Table 12. Relationship Between Systems-Change Variables and Outcome Effectiveness


Number ofsuccesses


Significant others change

Yes 268 148 55.2No 98 41 41.8

Environmental reorganization

Yes 20 13 65.0No 346 176 50.9

Table 13. Relationship Between Ecological Validityand Outcome Effectiveness


Number ofsuccesses


Intervention agent

Typical 154 94 61.0Atypical 196 87 44.3

Intervention setting

Typical 126 61 48.4Atypical 239 128 53.5

Intervene in all relevant contexts

Yes 76 42 55.3No 246 127 51.6

Note: Data based on combinations of both typical and atypical intervention agents andsettings were omitted, because there were too few outcomes.

Ecological ValidityWith respect to ecological validity, thedata (Table 13) show that the type ofintervention agent may make a difference.Typical agents (e.g., parents, teachers)produced a success rate of 61.0%,whereas atypical agents (e.g., psycholo-gists, researchers) had a success rate ofonly 44.3%. With respect to type ofsetting, the success rates for typical (e.g.,

54 68

home, integrated school) versus atypical(e.g., clinic, institution) were roughlycomparable. Comparability in successrates was also seen for interventions thattook place in all relevant contexts versusthose that did not.

MedicationIn light of the well-developed pharmaco-therapy literature alluded to earlier (Reiss

RESULTS

& Aman, 1998; Schaal & Hackenberg,1994; Schroeder & Tessel, 1994; Thomp-son et al., 1991), it seems plausible thatmedication might be another variablethat modulates the effectiveness of PBS.However, in many of the studies exam-ined, no reference was made to medica-tion, so its use is an unknown. In stillother studies, reference was made tomedication, but the dosage level was notindicated nor was the duration of druguse specified. A minority of studies

provided details on dosage level andduration but did not report data on howmedication and PBS interrelated. Thus,the unique contributions made bymedication versus PBS are not known atthis time. More important, perhaps, thepossibility that combinations of PBS andmedication produce synergistic effectshas not been examined systematicallyand is clearly a topic that merits futureattention from researchers.

69 55

DISCUSSION

CHAPTER 4

DISCUSSIONThe results of our synthesis bear onseveral important issues: (a) potentialbiases in the retrieved literature,(b) inferences based on the results,(c) effectiveness of PBS, and (d) implica-tions for future research.

Potential Biases in theRetrieved Literature

Rigor Versus Relevance?Our exclusion criteria (Table 2) weredesigned to produce a database that metthe highest standards of methodologicalrigor currently articulated in the field.These standards, however, reflect a strongbias in favor of demonstrations ofexperimental control (internal validity),sometimes at the expense of demonstra-tions of generality (external validity).These standards emphasize the analysisof cause and effect relationships at thelevel of single individuals with a view tounderstanding process variables. Thisdoes not mean, however, that the single-subject designs that predominate in thefield are inherently unable to address theissue of external validity.

It will be helpful to review, briefly, thedifference between group methods andsingle-subject designs with respect to theissue just raised. Traditional groupmethods rely on sampling procedures toestablish external validity. Greatergenerality is possible for individuals whoare similar to research participants onrelevant characteristics, and generallyexternal validity is less reliable forindividuals who are different fromresearch participants on those character-

istics. In the case of single-subjectdesigns, the investigator's ability togeneralize to other individuals is depen-dent on an understanding of what factorscontrolled a participant's behavior inbaseline (Wolery & Ezell, 1993). Once aninvestigator understands what controlsbaseline responding, then applyinginterventions to other participants whosebehavior is controlled by variables similarto those identified in the baseline willlikely produce greater external validitythan that achievable in the absence ofsuch similarity.

Single-subject research, now almostfour decades old, has been invaluable inidentifying critical variables that controlproblem behavior, raising issues concern-ing assessment and intervention, andsuggesting broad guidelines forremediation. It is also true, however, thatthis style of rigorous experimentalresearch is easiest to carry out in con-trolled situations. This contributed to thelarge number of outcomes in ourdatabase that involved atypical interven-tion agents (especially researchersandother expert professionals), atypicalsettings (particularly noncommunitysettings in which key variables could beeasily monitored and manipulated), andrestricted venues for intervention (i.e., alack of intervention in all relevantcontexts). Short-term process studiesconducted in restricted contexts also tendto underemphasize repeated assessment,multicomponent intervention, andmeasures of generalization, maintenance,and validated lifestyle change. Yet, as wenoted earlier, these are precisely the

7 0 57

CHAPTER 4

dimensions that define relevance(external validity), because without themthere is no demonstration that anintervention (a) is readily applicable inthe community or (b) changes peoples'lives in broad, meaningful ways. In viewof what we have just outlined, it is notsurprising that the journal best known forits emphasis on experimental analysis(Journal of Applied Behavior Analysis)accounted for almost half (42%) of thearticles included in the database (Table 1)whereas the journal best known for itsapplied community orientation (Journalof The Association for Persons With SevereHandicaps) accounted for only 9% of thearticles included.

Have researchers chosen rigor overrelevance? It seems so. Is this choice theonly one possible for the field? Wethink not.

Rigor Over Relevance?With respect to the issue of rigor overrelevance, there is much research,including some in our database, that doesin fact demonstrate a systematic concernfor external validity and all the variablesassociated with it. In addition, many ofthe excluded articles, while not meetingour methodological criteria, easily metimportant criteria for external validity,and for that reason will be highlightedshortly. Finally, our review of the researchliterature should not be seen as an exactreflection of current clinical practice.There is a vast nonresearch (practitioner)literature that reports successful, exter-nally valid applications of PBS (e.g.,Copeland, 1997; Hays, 1997; Jones, 1997;Kincaid, 1992, 1996; Lucyshyn & Albin,1993; Metlen, Majure, & Stroll-Reisler,1996; Metz, 1992a,1992b, 1992c; TheFamily Connection staff, DeVault, Krug, &Fake, 1996; Tifft, 1996; Turnbull &Turnbull, 1996; Virginia Institute for

58 71

Developmental Disabilities, 1996). Thechallenge therefore is how best to viewthe bias toward controlled experimentalresearch so that new perspectives can becreated that build constructively on thatbias while at the same time movingbeyond it.

The bias toward controlled experi-mental research is a legitimate beginningfor the evaluation of the PBS approach.Recall that two questions critical to thissynthesis (and discussed shortly) involve(a) whether PBS is effective and (b) whatfactors modulate its effectiveness. Thesequestions can be definitively answeredonly through careful experimentalanalysis that includes appropriatecontrols to ensure internal validity. In theabsence of these controls, interpretationof the database would be speculative atbest. Nonetheless, the portions of thedatabase that emphasize internal validitywhile minimizing the focus on externalvalidity are a useful beginning for thefield, because they increase our confi-dence that (a) PBS procedures do havedemonstrable effectiveness and(b) certain variables are causal inmodulating that effectiveness.

The portions of the experimentaldatabase, however scant, that movebeyond internal validity issues todemonstrate ecological validity show thatit is possible to produce meaningfulchange under more naturalistic condi-tions. The question then becomes howbest to extend the database undernaturalistic conditions. For this, one mustturn to an examination of some of theexcluded research articles that we willreview shortly. At this point in thedevelopment of the field, it is clear thatmore controlled experimental research incommunity settings is a legitimate futuredirection for the extension of PBS.

DISCUSSION

The nonresearch literature thatfocuses on consumer needs and interestshelps to clarify the gap between theresearch literature (both included andexcluded studies) and what key consum-ers value, thereby providing a heuristicfor moving beyond what researcherspresently offer.

In sum, the biases in the databaseserve a useful function in establishing atrustworthy foundation which, togetherwith the heuristic elements inherent inthe excluded research articles andnonresearch literature, provide directionfor the further development of the PBSapproach.

Is Rigor Over Relevance theOnly Choice?With respect to whether one must choosebetween rigor and relevance, thoughtfulconsideration suggests that this issueneeds to be restated. Specifically, there isno absolute standard for rigor. Rigor is a

function of context. Currently the "goldstandard" for rigor is the laboratoryexperiment, but, as we have seen, thisstandard is generally unrealistic incomplex community settings.

Rigor needs to be defined in terms ofits contextual parameters. Analogdemonstrations ought to be subjected tocriteria for rigor that characterizelaboratory research, in which there is anemphasis on highly controlled situationsthat permit cause and effect statements.In contrast, applied community demon-strations ought to be subjected to criteriafor rigor that reflect the realities ofcarrying out research in situations thattypically offer fewer opportunities formanipulating one variable at a time, thatfrequently contain multiple interactingvariables, that require nonresearchers/

nonexperts to implement assessment andintervention, that demand that interven-tion be conducted over protractedperiods of time, and that entail alteringsocial systems to produce validatedlifestyle change.

These applied community criteria forrigor have not been systematicallyarticulated in the literature, and, there-fore, represent an evolving feature of thefield, one that, at its heart, is tantamountto developing a new applied science. From

a researcher's perspective, this taskrepresents the greatest challenge offered

by the PBS approach.

It is important to note that thelaboratory-style experiment remains theonly method for determining, definitively,cause and effect relationships. In thiscase, advances in basic knowledge willstill depend on using some variant of thetraditional experiment. However,multidimensional projects that focus onintervention efficacy need to be evaluatedas well, and, although they may not produce

new basic knowledge, they generate atype of applied knowledge that is crucialto advancing the practice of PBS innaturalistic contexts. Some of the researcharticles excluded from our databaserepresent an important step in movingbeyond the current bias favoring experi-mental studies, and toward incorporatingthe best information generated by thosestudies into a research style that focuseson issues of external validity.

External Validity ofExcluded Studies

Of the 107 articles excluded for one ormore methodological reasons, a numberwere noteworthy in that they demon-strated the viability of PBS in meetingimportant criteria for external validity

7 259

CHAPTER 4

(e.g., Cameron, Ainsleigh, & Bird, 1992;Dadson & Horner, 1993; Homer, Close etal., 1996; Horner, Vaughn, Day, & Ard,1996; Lucyshyn, Olson, & Homer, 1995;Malette et al., 1992; Northup et al., 1994;Smith, 1985; Smith & Coleman, 1986).Although we did not examine thisliterature using the detailed analysesapplied to the included studies, it may behelpful here to review one representativestudy that reflects a concern with externalvalidity criteria.

One illustrative study by Homer,Close et al. (1996) involved PBS interven-tion for 12 adults who had been institu-tionalized for many years. These indi-viduals were identified by staff as amongthe most challenging in the entireinstitution. The approach taken involveda number of PBS interventions whoseinternal validity had been well estab-lished in numerous experimental studiesof the type included in our database.Thus, the intervention adopted was builton a solid foundation of experimentalresearch. Importantly, however, numerousvariables associated with external validityconcerns were combined with the coreinterventions in order to produce anapproach that proved viable in addressingthe wider needs of the individualsinvolved. Specifically, the entire interven-tion was implemented by typical inter-vention agents (support staff) rather thanby experts or researchers. A supportedliving model was put in place that focusedexclusively on home and community(typical settings) as the site of interven-tion. Evaluation occurred over a period of4 years (long-term maintenance).Because of the changing nature of theliving situation, repeated assessmentswere routinely conducted, and interven-tion was carried out in all relevantcontexts. As the natural contexts involved

multivariate control of problem behavior,the intervention itselfwas multicompo-nent and included reorganizing theenvironment (rearranging the physicalsetting and schedules), changing taskfeatures, teaching new skills, alteringreinforcing consequences for positivebehavior as well as applying extinction toproblem behavior, and enhancing healthand safety factors. The results of thiscomprehensive PBS strategy demon-strated decreases in major problembehaviors, greater physical and socialinclusion, and stability or improvement inhealth and safety. Further, both familiesand direct support staff validated thesechanges with respect to variables thatincluded community integration, socialrelationships, problem behavior, andoverall quality of life. Thus, validatedlifestyle change was evident.

This study, like the others citedearlier, is instructive for two reasons.First, it demonstrates that a comprehen-sive PBS approach that is responsive toexternal validity concerns can yieldimprovements in problem behavior whileproducing validated lifestyle change.Second, it demonstrates the need toconsider anew how we define rigorousapplied science; if we do not, we willcontinue to underestimate the scope andvalue of PBS applications in communitycontexts. It is encouraging to note thatemerging research addresses the issue ofwhat constitutes rigorous community-based science (e.g., Lucyshyn, Albin, &Nixon, 1997). It is becoming clear thatsuch research will almost certainly needto redefine what is acceptable in thedomains of assessment, intervention, andoutcomes, a point to which we shallreturn later.

At this point, it is useful to examinevarious features of the database while

60 7 3

DISCUSSION

keeping in mind the external validitycriteria we have just discussed. In thismanner, gaps in the field can begin to beidentified and then elaborated upon inthe final section of the discussion thatdeals with the differing perspectives andneeds of consumers (nonresearchers).

Drawing Inferencesfrom the Results

In this section, we use the database inorder to address the first four researchquestions posed earlier.

How Widely ApplicableIs PBS?

Several researchers have raised thelegitimate question of whether PBS mightbe limited in the scope of its applicability.Specifically, they have cautioned that theapproach has, in the past, often focusedon individuals who are relatively easy todeal with because they are young andhave minor disabilities, high levels ofcognitive functioning, and mild problembehavior (Axelrod, 1987; Feldman, 1990;Mulick & Linscheid, 1988). The presentdata (Table 4) make clear, however, thatthe field has developed to the point thatdifficult cases are now being addressed.Thus, (a) over half the cases in ourdatabase were adolescents or adults;(b) serious disabilities (i.e., mentalretardation, autism) were well repre-sented; (c) fully two-thirds of the casesfell within the severe/profound level ofretardation; and (d) the types of problembehavior involved were among the mostserious ones seen clinically (aggression,self-injury). There is, however, one areawhere the scope of applicability of PBS isas yet unclear, specifically, culturaldiversity. Very few studies mentionminority status as a participant charac-teristic. More significantly, no studies

investigated whether cultural parametersinfluenced the design of the PBS ap-proach used. Yet cultural factors havebeen identified as one important aspectof planning interventions for people withdisabilities (Harry, Allen, & McClaughlin,1995; Harry, Grenot-Scheyer et al., 1995).The systematic exploration of culturalfactors and PBS use is a topic that awaitsresearch scrutiny.

Another concern is that the putativecomplexity of PBS may preempt itsapplicability to natural (typical) settingsunless there is extensive involvement ofhighly trained professionals (Paisey,Whitney, & Hislop, 1990).Again, however,the data (Table 4) show that almost halfof the intervention agents were notexperts; they were parents, teachers, jobcoaches, and the like (typical agents).Further, in one-third of the cases,intervention took place in typical settings(homes, integrated schools), therebydemonstrating that the use of PBS wasnot restricted to the kinds ofspecializedsettings (e.g., clinics, segregated schools)normally associated with expert profes-sionals.

In What Ways Is theField Evolving?

General Status of PBSHistorically, the field of developmentaldisabilities has shown a tendency toembrace new intervention approacheswith great enthusiasm and then abandonthem with equal enthusiasm as the initialpromise of the interventions failed tosurvive scientific scrutiny (Arendt,MacLean, & Baumeister, 1988; Green &Shane, 1994; Willemsen-Swinkels,Buitelaar, Nijhof, & van Engeland, 1995).Is PBS yet another passing fad? The data(Figure 1) suggest that this approach is

7461

CHAPTER 4

here to stay. Over the 12-year period ofthe review, there was a clear increase inthe number of articles published as wellas the number of participants andoutcomes involved. This increase isespecially noteworthy in view of periodiccriticisms of PBS alluded to earlier in thisreview (e.g., see Repp & Singh, 1990).

DemographicsPBS procedures have remained widelyapplicable over time to individuals withsignificant problem behavior (Figures 2-6). During the 12-year period reviewed,very young children (preschool age)made up only a small percentage of thecases; most individuals who participatedin PBS interventions were older(elementary-school age to adult) and,therefore, presumablymore difficult todeal with. Likewise, people diagnosed ashaving mental retardation or autism, andthose with cognitiVe functioning in theprofound or severe range were present insubstantial and stable numbers over time,again attesting to the continued applica-bility of PBS to individuals who aregenerally viewed as providing seriouschallenges. Importantly, while theproportion of cases involving aggressivebehavior remained stable, those involvingdangerous self-injurious behavior andmultiple problem behaviors (combina-tions) actually increased over time. Insum, then, PBS procedures continued tobe applied to populations of people whoare generally viewed as posing thegreatest challenges.

As noted previously, there has beensome concern in the field that PBSprocedures are being implementedprimarily by highly trained experts(atypical intervention agents) in highlyspecialized (atypical) settings (Paisey etal.,1990; Scotti et al., 1996). Our database

62 75

(Figure 7) demonstrated no clear trendwith respect to type of intervention agent.Interestingly though, in the last timeblock (1993-1996), there were equalnumbers of typical and atypical interven-tion agents, a fact that showed thecontinued heavy involvement of typicalintervention agents at a time when thenumber of PBS studies rose dramatically.Just as important, there appears to be aclear trend in the field toward applyingPBS more often in typical settings. Thus,the ratio of typical to atypical settingshas, over time, been approaching parity(Figure 8). The field appears to be movingtentatively toward greater use of PBS innatural contexts.

AssessmentFor many years, investigators havestressed the importance of conducting anassessment prior to designing anintervention, but they have also noted therelative lack of such assessment in thepublished literature (Carr et al., 1990;Scotti et al., 1991). More recently, reviewsof the entire spectrum of interventionsfor problem behavior have detected aclear movement toward greater use ofassessment methods (Scotti et al., 1996;Didden et al., 1997). Our databaseconfirms that these general trends in thefield are seen also in the PBS literature.There has been a dramatic increase overtime in the proportion of outcomesassociated with prior assessment (Figure9). To extend previous reviews, we brokedown the information obtained so that wecould examine trends with respect tospecific types of assessment (Figure 10).We found large increases in the use offunctional analysis and combinations ofassessment procedures. The sole use ofindirect or direct observation remained ararity. Our data also showed that func-tional analysis was most likely to be

DISCUSSION

associated with atypical interventionagents, atypical settings, and a lack ofintervention in all relevant contexts, allindices of lower ecological validity. Incontrast, informal and/or direct observa-

tion were generally associated withhigher levels of ecological validity.

These data suggest that functionalanalysis is unlikely to be viewed as afeasible method of assessment bypractitioners operating in naturalisticcontexts. One recent survey (Desrochers,Hi le, & Williams-Moseley, 1997) supportsthis notion. Specifically, 300 practitionerswere asked to rank the degree of useful-

ness of a number of assessment proce-dures. ABC (antecedent-behavior-consequences) analysis (i.e., a directobservation procedure) and interviewwith staff/relatives (i.e., an indirectobservation procedure) ranked first andsecond respectively. Functional analysisranked a very distant fifth in usefulness.Practitioners felt that a lack of environ-mental control, insufficient time, diffi-culty collecting data, lack of expertise,and environmental complexity all madefunctional analysis an impractical and,therefore, seldom used method in real-lifesettings. Taken as a whole, these consider-ations argue for the development andrefinement of a new generation ofnonexperimental assessment proceduresthat are user-friendly, practical, andfeasible, while retaining functionalanalysis primarily'as a tool for research-ers and occasionally to be used by expertsfor cases in which less formal methods havefailed. There is, of course, a 'related strongneed for personnel preparation programsand associated texts (e.g.; Sne11,1993) that

can assist future professionals in con-ducting valid and practical assessments.

Our review of the. PBS literature isalsb in accord with analyses of motivation

documenting that escape is the singlemost commonly identified function forproblem behavior (Scotti et al., 1996). Inaddition, our database shows that thisfeature is stable in that escape wasassociated with the greatest number ofoutcomes in each successive time block(Figure 11). These data may reflect thedecreasing emphasis in the field oncustodial care and a greater emphasis oneducation, involvement in meaningfulactivities, and work (Emerson et al.,1994). Because escape-motivatedproblem behavior typically occurs inresponse to demands (e.g.,academictasks, home chores, work activities), theshift from low-demand custodialsituations to higher-demand school andcommunity-based situations may haveprovided many more opportunities forescape-motivated problem behavior tomanifest itself.

Interestingly, the proportion ofoutcomes involving multiple motivationsis accelerating over time. The greaterinvolvement of people with disabilities intypical settings, noted earlier, may berelated to the increase in outcomesassociated with multiple motivation; thecomplexity of school, work, and homesituations (typical settings) compared tothe more restricted, less complex atypicalsettings may provide increased opportu-nities for multiple factors (e.g., escape,attention, tangibles) to exert theirinfluence on problem behavior.

The data on whether assessment wasrepeated provide some important insightson how the field is progressing. Thus,although repeated assessment wasassociated with only 11 outcomes, 10 ofthese occurred in the most recent timeblock. Significantly, repeated assessmentwas most likely to be carried out intypical settings and by typical interven-

7663

CHAPTER 4

tion agents. In contrast, a lack of repeatedassessment was most likely to be associ-ated with atypical agents and atypicalsettings. These findings make sense givenrecommendations that assessment shouldbe repeated whenever there are importantchanges in an individual's life situation(Carr et al., 1994; O'Neill et al., 1997a &b). In typical settings (e.g., home, school),ongoing change is the norm; in atypicalsettings, ongoing change may be less of afactor, because such settings are oftenmore restricted in terms of activityschedules, social relationships, andreinforcer accessibility. Although the 2:1ratio of atypical to typical settings seen inthe database (Table 4) is consistent withthe lopsided ratio of nonrepeatedassessment to repeated assessment, theextremely low frequency of the latter doesnonetheless represent a relative short-coming of the field with respect toembracing one standard of best practice.

InterventionThe data (Figure 12) show that bothstimulus-based and reinforcement-basedinterventions have remained an enduringfeature of the field. However, there hasbeen an important change toward apredominance of stimulus-basedprocedures. This change likely reflects asurge of interest, more generally in thefield, in making education a priority forpeople with disabilities. Educationallyrelevant and popular procedures, such asinterspersal training, choice, curricularmodification, errorless learning, andprompting, are all stimulus-based andintegral to the promotion of learning inacademic and related contexts (Luiselli &Cameron, 1998).

As noted earlier, the PBS philosophydoes not view people with disabilities asbeing passive recipients of intervention

64 77

for problem behavior. Instead, they areseen as participating in systems thatprecipitate problem behaviors, and suchsystems are in need of change.

The data on systems change (Figure13) suggest a dramatic increase over timein the requirement for behavior changeon the part of significant others.Caregivers and support persons are mostrecently (1993-1996) frequently requiredto change aspects of their own behavioras one component of intervention. Thisfinding reflects the reciprocity ofbehavior change that is one hallmark ofPBS intervention.

In contrast, the data on environmen-tal reorganization, an additional aspect ofsystems change, fail to show an increasingtrend but remain at low levels. Thisfinding may reflect the abiding predomi-nance of atypical settings over typicalsettings (Figure 8) as the site for interven-tion. In typical settings, a major priorityis commonly given to selecting engagingactivities, enriching lifestyle throughcommunity integration, providing choice,systematically providing respite services,and offering other opportunities relatedto broad environmental reorganization/restructuring (Emerson et al., 1994).Atypical settings are very often morerestrictive and institutional, limitingopportunities for such types of broadchange. Thus, the imbalance that favorsthe use of atypical settings over typicalsettings likely contributed greatly to thelack of focus on environmental reorgani-zation characterizing the current data-base. The accelerating interest in usingtypical settings as well as the currentpaucity of data on environmentalreorganization should, together, promptthe field to explore and analyze thisaspect of systems change further as onemore element of best practice.

DISCUSSION

Some of what has just been dis-cussed also bears on the finding that theapplication of intervention in all relevantcontexts was uncommon (Figure 14).Further analyses demonstrated thatintervention in all relevant contexts wasmost likely to be associated with typicalagents and settings, and least likely withatypical agents and settings. Other datashowed that intervention in all relevantcontexts was closely associated withcombined interventions (i.e., stimulus-based plus reinforcement-based),whereas its absence was closely associ-ated with the use of single interventions(i.e., either stimulus-based or reinforce-ment-based). One explanation is thatatypical agents and settings are oftenused in analog (laboratory-style) researchdesigned to study basic principles ratherthan to demonstrate broad clinicalchanges; so the scope of such interven-tions may be quite limited (i.e., not allrelevant contexts are involved and singleinterventions are common). In contrast,with typical agents and settings, the focusis often on demonstrating and evaluatingthe use of interventions across allpertinent aspects of an individual's life, sosuch interventions may more likelyinvolve all relevant contexts and theapplication of combined interventions.The paucity of data on intervention in allrelevant contexts may reflect a stronginterest, at least in the literature samplereviewed, in elucidating basic principlesat the expense of demonstrating broad-spectrum behavior change. Because thelatter activity is at least as important asthe former, one priority for the fieldwould be to redress this imbalance.

The pattern we have been describing,involving typical versus atypical agentsand settings, is seen once again in theanalysis of combined interventions (i.e.,

those involving both stimulus-based andreinforcement-based as part of the sameintervention package). Figure 12 showsno trend for the increased or decreaseduse of combined interventions. However,

additional analyses make clear thatcombined interventions, when they didoccur, were more likely to be conductedby typical intervention agents and intypical settings. Noncombined interven-tions were more likely to be conducted byatypical agents and in atypical settings.This pattern also conforms to thedistinction between analog research andapplied research in the community. In theformer type of laboratory-style research,one typically avoids combiningmultiplevariables, favoring instead the manipula-tion of as few variables as possible todemonstrate cause-and-effect relation-ships unambiguously. In contrast, thelatter type of research is more concernedwith demonstrating the efficacy ofintervention packages (i.e., the mostpotent combination of interventions).

In sum, then, the data in Figures 12-14 yield a fairly consistent picture in termsof the very different research and clinicalagenda represented by typical agentsworking in typical settings versus atypicalagents working in atypical settings.

A final aspect of the interventiondata pertains to the use of non-PBSprocedures (Figure 12). These procedureswere much less frequently used than PBSthroughout the entire time periodreviewed. This finding implies that, for aclear majority of outcomes, PBS by itselfwas viewed by investigators as an adequateapproach for dealing with serious problembehavior; the addition of non-PBSprocedures was not seen as a necessity.Subsequent analyses, discussed shortly,corroborate and amplify this conclusion.

.7365

CHAPTER 4

OutcomesFor many years, PBS advocates, amongothers, have argued strongly that reduc-tion in problem behavior is an insuffi-cient outcome unless such reduction isassociated with broader changes thatinclude stimulus and response generali-zation, maintenance, and improvementsin lifestyle (Carr et al.,1994; Horner,Dunlap, & Koegel, 1988; Koegel et al.,1996; Meyer & Evans, 1989; Schalock,1990; Turnbull & Turnbull, 1996). Ourdatabase (Figure 15) suggests that thiscall for best practices has gone largelyunheeded. Relatively few outcomesinvolved demonstrations of broad change.Thus, no trends were seen for stimulusgeneralization and only a small increasewas seen for response generalization.Demonstrations of short-term mainte-nance (less than 6 months) became morefrequent over the period reviewed;however, demonstrations of longer-termmaintenance dropped sharply in fre-quency as the duration of follow-up wasextended. Significantly, no outcomes weretracked for more than 2 years, thusleaving unanswered the question ofwhether PBS can produce permanentchange. Of most concern, however, is thefinding that a focus on lifestyle changewas uncommon, with no compellingupward trend. Studies meeting ourrigorous inclusion criteria appeared moreconcerned with demonstrating initialefficacy and with understanding ofprocess, and less concerned with moregeneralized change including improve-ments in lifestyle. This gap in theliterature helps define a critical researchpriority for the future.

The lack of focus on broad change isparalleled by a lack of focus on social

66 79

validity (Figure 16). Social validity datawere available for only a small number ofoutcomes. Modest increases over timewere seen for questions dealing with thefeasibility of intervention and theacceptability of intervention outcomes,but no trends were seen for questionspertaining to the desirability of specificintervention procedures and the effective-ness of intervention in producing lifestylechange. The paucity of data on socialvalidation may be linked to the paucity ofdata on broad change. That is, to theextent that investigators have been largelyconcerned with demonstrating interven-tion effects in somewhat circumscribedcontexts, it makes little sense to asksocial validity questions that are intendedto tap consumer satisfaction across abroad range of circumstances. A greaterfuture emphasis on studies of broadbehavior change should lead as well to agreater focus on whether such change issocially valid.

How Effective Is PBS?

Changes in Positive BehaviorOur database (Figure 17) documented,true to the definition of PBS, increases inpositive behavior. Data available forslightly less than half the outcomesdemonstrated modest to substantialincreases in socially desirable responses(positive behaviors) as a consequence ofintervention. These data are importantbecause they show that the field hasmoved beyond an exclusive concern withreducing problem behavior per se and hasbecome more committed to enhancingconstructive aspects of individuals'repertoires as a general strategy fordealing with problem behavior.

DISCUSSION

Distribution of OutcomeEffectiveness: What Constitutesa Success?There were many instances in which PBSwas an effective strategy for reducing thelevel of serious problem behavior (Table5). The two generic variations of PBS,stimulus -based and reinforcement-based,were about equally effective in producingsuccessful outcomes (Tables 5 & 6). It isimportant to understand, however, thatthe field has not yet reached a consensuson what constitutes a success. As can benoted in the distribution of outcomeeffectiveness (Table 5), the degree ofsuccess depends on where one sets thecutoff criterion. Across all PBS-basedoutcomes, a criterion of 100% reduction(total elimination of problem behavior)produced a success rate of 26.8%, onequarter of all the outcomes. Few investi-gators set so demanding a criterion.Instead, in the present paper, as well as inthe earlier review the senior authorconducted for the National Institutes ofHealth (Carr et al., 1990), a 90% or morereduction criterion was established thatyielded a success rate of 51.6%, one halfof all the outcomes. Other investigators(Mulick & Kedesdy, 1988) have used an80% or more reduction criterion, whichin the present case would produce asuccess rate of 68.0%, two thirds of all theoutcomes. An 80% reduction criterionmight, at first glance, appear unaccept-ably low. Yet clinically, if an individualwho has been biting him- or herself anaverage of 100 times a day for manymonths were to decrease biting abruptlyto 20 times a day, caregivers might ratethe outcome as a qualified success. Incontrast, even a 90% reduction in a lesschallenging self-injurious behavior (e.g.,trichotillomania: hair pulling) might beviewed as unimpressive and relatively

unsuccessful. The point is that othervariables besides percentage reductioncontribute to the definition of interven-tion success. In this vein, the manydemonstrations in the literature ofdramatic decreases in problem behaviorduring 10-minute analog sessionsconducted by researchers are not, a priori,superior to more modest decreasesreported for interventions appliedthroughout the day in natural settings bytypical caregivers, a serious concern thatneeds more systematic attention in thefuture.

Table 5 also shows that for 33outcomes (9% of the total), problembehavior decreased minimally (only 0-19% from baseline) or increased. Ourprevious review (Carr et al., 1990)likewise noted negative outcomes in asmall minority of cases. Such findingsserve to underscore the point that nointervention, including PBS, should beassumed to be universally effective. Asnoted later, negative outcomes may reflectthe influence of variables (e.g., non-optimal social systems, physiologicalfactors) that attenuate the impact of PBSstrategies. Identifying such variablesrepresents yet another research priorityfor the field.

Success Rates PooledAcross OutcomesWith all the caveats just discussed inmind, we used a 90% reduction criterionas our definition of success. Using thiscriterion, it was clear (Table 6) that thesuccess rate was invariant and notdependant on the type of interventionemployed. Generally, about half theoutcomes could be categorized assuccesses, irrespective of whether theywere associated with stimulus-based orreinforcement-based intervention and/or

&' 067

CHAPTER 4

different combinations of these with non-PBS procedures. However, furtheranalyses of these data highlight some ofthe complexities that are not initiallyapparent from examination of Table 6alone.

Single Versus Multiple InterventionsThe fact that single interventions had asuccess rate of 51.5%, and combinationinterventions had a success rate of 52.4%,might make it appear as if not much isgained by combining stimulus-based andreinforcement-based intervention into acomprehensive approach. However, asnoted earlier, single interventions weremost likely to be carried out by atypicalagents (e.g., research psychologists) inatypical settings (e.g., institutions)whereas combined interventions weremost likely to be carried out by typicalagents (e.g., parents) in typical settings(e.g., homes). The degree of controlavailable in typical settings involvingtypical agents is, arguably, substantiallyless than that achievable by experts(atypical agents) operating in morerestricted (atypical) settings. Therefore,one might expect poorer outcomesassociated with the former as comparedto the latter. Because this result was notthe case (i.e., the success rates were aboutthe same), one could conclude thatcombined intervention might be particu-larly useful in dealing with situations inwhich natural (typical) interventionagents must address the day-to-day(typical) settings in which many peoplewith disabilities live.

The Role of ConsequencesWhereas PBS interventions are proactive(i.e., take place in the absence of problembehavior and act to prevent futureoccurrences of such behavior), non-PBSinterventions are reactive (i.e., take place

68

at the moment problem behavior isoccurring and act as direct consequencesfor such behavior). The data (Table 6)suggest that non-PBS interventions thatstress consequences may not always benecessary for dealing with problembehavior because (a) for almost threequarters of the database (72.1% of theoutcomes), PBS procedures were usedalone and (b) the rate of success wasvirtually the same with or without non-PBS (i.e., 51.5% successes with PBS alone,and 52.0% with non-PBS). Again,however, consideration of a number offactors suggests that this conclusionneeds to be tempered and qualified. Herewe will discuss two key factors, the role ofpunishment and extinction as responseconsequences.

Punishment:A Form of CrisisManagementMost practitioners and caregivers areconcerned about what they can do at themoment of crisis, when dangerousbehavior is occurring, to mitigate itseffects. Our database (Table 7) shows thatapproximately half of the non-PBSinterventions took the form of punish-ment, albeit, with one exception (watermist), punishments that are usuallyregarded as relatively nonintrusive. Thefact that generally innocuous non-PBSprocedures (added to the PBS proceduresalready in place) were sufficient withoutthe addition of more highly intrusiveprocedures should be encouraging tothose opposed to the use of the lattermore controversial measures. As has beenpointed out elsewhere (Carr et al., 1994),intervention agents frequently feel theneed for crisis management procedureseven when committed to a PBS approach.The perceived need for these procedurescannot easily be dismissed. The impor-

DISCUSSION

tant message from our database, however,is that when additional non-PBS proce-dures seemed necessary, they wererelatively benign, and relatively rare.

Extinction: A Crucial Consequence?Our database (Table 7) also showed thatapproximately half of the non-PBSinterventions took the form of extinction.There are conceptual grounds forassuming that extinction may not alwaysbe necessary, but there are additionalgrounds that support its use. Conceptu-ally, both the Matching Law and thenotion of response efficiency mayplausibly obviate the use of extinction.The Matching Law states that the relativerate of a response is a function of therelative rate of reinforcement for thatresponse (McDowell, 1982). Thus, if anindividual receives all his or her rein-forcement contingent on aggressivebehavior, that behavior would occur at ahigh frequency. But suppose, through PBSintervention, the individual acquiredmany new skills (responses) that alsoprovided access to reinforcement. Nowthis person would not be solely depen-dent on aggression as a means foraccessing reinforcers. Thus, even ifaggressive behavior were reinforced at thesame level as it had been prior to PBSintervention, the relative rate of thisbehavior would now be much lowerbecause of the presence of many newskills (responses) that generate additionalreinforcement. So the absence of anextinction contingency for aggressivebehavior would not necessarily matter,because many available competingpositive responses reduced the relativerate of reinforcement for aggression withthe result that aggression would now beless frequent.

Response efficiency is another factor.Data show that if two responses both

access the same reinforcer, the moreefficient of the two is likely to predomi-nate (Homer & Day, 1991). Let us say anindividual acquired a new response (e.g.,communication) as a result of PBSintervention. But then the individualcontinued to receive reinforcement foraggressive behavior (i.e., no extinctioncontingency). If the new response wasmore efficient than the aggressivebehavior, the new response wouldpredominate, and aggression wouldbecome less frequent.

The possible role of the MatchingLaw and response efficiency in decreasingincidents of problem behavior in theabsence of extinction may explain why sofew outcomes in the database wereassociated with the use of extinction.Investigators may simply have found it tobe unnecessary. Nonetheless, a variety ofconsiderations support its use. First,many investigators, while not providingformal data on the use of extinction,made passing references to "ignoring"problem behavior or "not rewarding" it,both statements implying that somedegree of extinction was in effect. Thus,the number of outcomes associated withextinction (Table 7) may represent anunderestimate. Second, and morecritically, the Matching Law itself suggestswhy extinction may be important.Specifically, if in addition to strengthen-ing new skills that access the reinforcerspreviously associated with problembehavior alone, one were also to discon-tinue the reinforcement of the problembehavior itself (extinction), then onewould be greatly increasing the relativerate of reinforcement for the new(positive) skills. This strategy would alsohave the effect of increasing the relativefrequency of positive behavior. In thiscrucial respect, extinction could, on

82 69

CHAPTER 4

conceptual grounds, be viewed as animportant procedure because it directlypotentiates the display of positivebehavior. On these grounds alone, itwould be prudent to include extinction aspart of any PBS-based approach.

Stimulus Generalization,Response Generalization,and MaintenanceThese three measures of generalizationwere typically associated with smallnumbers of outcomes. Yet, with a 90%reduction criterion for problem behavior,about two-thirds of the outcomes weresuccesses for stimulus generalization andfor the various durations of maintenance.Response generalization was much lesssuccessful, with small gains in appropri-ate behavior characteristic of most of theoutcomes. The scant attention paid tothese three measures of generalizationwas criticized in our earlier review of PBSintervention (Carr et al., 1990). Unfortu-nately, this gap in best practices remainsan enduring feature of the field; thereremains a strong bias toward demonstrat-ing experimental control of problembehavior in restricted situations ratherthan investigating more widespreadchange.

Full implementation of the PBSapproach would significantly increase thedatabase pertinent to generalizationmeasures. Specifically, a greater focus onassessing and remediating problembehavior in all relevant contexts (not justanalog settings) would, of necessity, teachspecific alternatives to problem behaviorin many different situations (multipleexemplar training), a factor known topromote stimulus generalization (Stokes& Baer, 1977). Likewise, extensive

assessment and intervention for deficient

70 83

behavior repertoires would teach andstrengthen a wide variety of responses(communication, self-management,social skills) whose increase might wellchange many other behaviors in a

response generalization paradigm (Carr,1988). Finally, because optimum PBSintervention is not saltatory but, rather,continuous, ongoing, and permanent,maintenance is potentially a naturalconsequence of the approach. To state thematter differently, the hallmark of thePBS philosophy is to remediate deficien-cies in all contexts (i.e., deficient environ-ments and deficient skill repertoires).Once this remediation has occurred,agents take steps to ensure that thecontexts do not deteriorate to their initiallevels of deficiency. In this manner,maintenance is promoted indefinitely.

Lifestyle Change andSocial ValidityThe ultimate goals of PBS interventionare to enable individuals to live morenormalized lives (lifestyle change) and tohave key consumers (e.g., caregivers,support persons, and, if possible, thepeople with disabilities themselves) agreethat the intervention and its effects areworthwhile (social validity). A strongmovement developing in the fieldemphasizes the centrality of these goalsin service provision and remediationefforts (Dennis, Williams, Giangreco, &Cloninger, 1993; Hughes, Hwang, Kim,Eisenman, & Killian, 1995; Risley, 1996;Sands, Kozleski, & Goodwin, 1991;Schalock, 1990, 1996; Turnbull &Turnbull, 1996).Yet the database does notreflect this emphasis. Specifically, only asmall number of outcomes focused onthese issues. Data on successful lifestylechange were taken on only 8 out of the

DISCUSSION

230 participants in our sample, and thesesuccessful outcomes were partiallyvalidated by a likewise small sample ofconsumers, none of whom was an actualparticipant. Clearly, a large gap existsbetween the stated goals for PBS inter-vention and the intensive and systematicempirical exploration of these goals.

Several factors have hindered thedevelopment of a larger database. First, asnoted earlier, our rigorous inclusioncriteria resulted in the elimination of anumber of studies (on methodologicalgrounds) that did demonstrate validatedlifestyle change. The interpretability ofthese studies was problematic, however,primarily because the standard of rigorthat exists in many parts of the fieldemphasizes a degree of experimentalcontrol that is most easily achieved inanalog situations. Perhaps that is why somany outcomes were associated withrestricted (atypical) settings in which theintervention was carried out by expert(atypical) intervention agents in re-stricted circumstances (i.e., not allrelevant contexts). Tightly controlling thecontext in this manner greatly increasesthe prospects of achieving internalvalidity but does so at the expense ofdeveloping a truly applied technologythat has the multidimensional propertiesneeded for addressing goals larger thanexperimental control, namely, validatedlifestyle change. As noted previously, rigoris not an absolute construct but, rather, afunction of context. Thus, the standardsdefining rigor for analog studies thatstress the analysis of process will almostcertainly have to be different from thestandards defining rigor for naturalistic,community-based studies that stress theanalysis of outcome.

What Factors Modulate theEffectiveness of PBS?

Demographic VariablesAlthough success rates varied withinmany of the demographic variables (e.g.,on the age variable, success rates wereconsiderably higher for adolescents thanfor adults), comparison of the presentdatabase (Table 10) with that obtained inprevious related reviews (e.g., Carr et al.,1990; Scotti et al., 1991) does not revealany replicated trends. One could notconclude, for instance, that problembehavior in adults is consistently moredifficult to remediate than problembehavior in adolescents. In addition, theliterature contains no conceptual basis forinferring that variations in demographicvariables, such as gender, age, diagnosis,retardation level, or topography (type) ofproblem behavior, ought to bear asystematic relationship to interventionsuccess.

We intentionally included anadditional variable not explored inprevious reviews: the relative difficulty indealing with single types of problembehavior versus combinations of types.Data indicated that the latter wasassociated with lower success rates; thismay be due in part to the increasedchallenge of intervening on multipletypes of problem behavior. It may be thatmultiple problem behaviors servemultiple functions, thereby requiring thedesign of interventions that are morecomplex and, therefore, more prone tofailure. Alternatively, the great effortrequired to deal with diverse problembehavior may render support people lesslikely to persevere in intervention efforts,resulting in poor outcomes. In any case,

8471

CHAPTER 4

the analysis of single versus combinationproblem behavior has not been ad-equately explored in the literature; thepossibility that this factor may modulateintervention effectiveness makes itworthy of study.

Influence of AssessmentVariablesOne of the most striking findings wasthat the success rate for interventionsbased on some form of functionalassessment was almost twice that ofinterventions not based on a priorassessment (Table 11). This findingappears robust, having been noted also inseveral previous reviews (Carr et al., 1990;Didden et al., 1997; Scotti et al., 1991).The relationship between assessment andsuccessful outcomes suggests that thefield needs to develop assessmenttechnology further; as noted, the pre-dominant tool is currently functionalanalysis, and this type of assessmentrequires a level of expertise and a degreeof experimental control that is frequentlyabsent in community-based settingsinvolving typical intervention agents. Wereturn to this point when we considerwhat gaps in knowledge exist in the fieldand how best to address them.

With respect to type of function, itwas clear that outcome success wasuniformly substantial (i.e., an approxi-mately 60% success rate) across thevarious functions with one notableexception: The sensory function wasassociated with an extremely low successrate. Although this finding might simplyreflect sampling error based on the smallnumber of outcomes associated with thesensory function, there is a noteworthyalternative explanation. Specifically, allthe other functions were socially medi-ated; the sensory function was not. One

72 85

interpretation of this finding is that PBSintervention is not pertinent to problembehavior that lacks social mediation. Amore optimistic interpretation is that thefield has not as yet adequately addressednonsocially mediated problem behaviorwithin a PBS framework; if it were to doso, success rates would rise. Certainly,there are instances of problem behaviorrelated to physiological variables thatwould logically appear to be outside thepurview of a PBS approach (Gardner &Whalen, 1996; Guess & Siegel-Causey,1995; Lowry & Sovner, 1992; Schroeder &Tessel, 1994). For example, it has beenknown for some time that certain geneticconditions such as Lesch-Nyhan syn-drome may cause self-injurious behavior(Seegmiller, 1972). Whether sensory-based behavior is likewise outside thePBS approach is currently an empiricalquestion. One working hypothesispostulates that extending PBS interven-tions so as to provide nonproblem-behavior alternatives for accessingrelevant sensory stimuli might under-mine sensory-based problem behavior,thereby improving success rates (Favell etal., 1982; Prosser, 1988; Smith, 1986).

Finally, the small sample of outcomesassociated with repeated assessment isnot in keeping with best practices. If PBSis truly an approach designed to deal withproblem behavior across all relevantcontexts for protracted periods of time, itstrains credibility to believe that a singleassessment would suffice to understandthe totality of problem behavior acrosschanging circumstances over many years.One interpretation of the relatively lowsuccess rate associated with repeatedassessment is that the field has not yetexplored this issue with the sameintensity and detail as it has explored theissue of short-term analog assessment in

DISCUSSION

restricted settings. Developing a largerdatabase on this topic could help identifyand refine the factors critical for ensuringmeaningful repeated assessment.

Systems ChangeRecall that a critical aspect of PBSintervention is the remediation ofdeficient contexts in an effort to normal-ize them. This requires change in systemsand not just in the person with disabili-ties. The data (Table 12) show that changein the behavior of significant others wasfrequent and associated with successrates higher than when the significantothers were not required to alter theirbehavior. One mechanism related to thisgreater success may involve the degree ofsupport the significant others provide fornewly enhanced skills displayed by theperson with disabilities. Thus, if theperson with disabilities begins to usecommunication rather than problembehavior as a way of accessing attention,then greater responsivity to bids forattention ensures that the new communi-cative behavior will be strengthened andbetter able to compete with problembehavior. Likewise, being able to "read"when a person with disabilities is seekingattention nonverbally (e.g., via furtiveglances toward the significant other) andthen responding to these nonverbal cuesprior to the display of problem behaviorrepresents another change in behaviorthat serves, in this instance, to preventfurther display of problem behavior. Thetype and magnitude of behavior changeshown by significant others as well astheir relation to intervention success havebeen studied only minimally in theliterature. Our database implies that itwould be fruitful to explore such changefurther by focusing on issues of reciproc-ity (i.e., both parties in a social exchange

alter their behavior) rather than the moretraditional emphasis on unilateralstrategies (i.e., only the person with theproblem behavior is the focus of interven-tion, and the primary goal is for thatperson to change).

The data on environmental reorgani-zation, though scant, suggest that thismolar approach can produce highersuccess rates than those approaches fromwhich it is excluded or minimized. Herethe effective mechanism may relate to thefact that broad environmental change(e.g., altering personnel, reorganizingactivity schedules, changing the physicalproperties of the home and/or school)often provides an array of discriminativestimuli and setting events that potentiatethe display of positive behaviors (Carr,Carlson, Langdon, Magito-McLaughlin, &Yarbrough, 1998) that, in turn, competewith the problem behavior. Importantly,our earlier discussion noted that ourdatabase includes a preponderance ofatypical settings involving atypicalintervention agents who fail to addressproblem behavior in all relevant contexts.This fact may account for the smalldatabase related to environmentalreorganization, because any approachthat minimizes naturalistic variablessimultaneously limits the scope ofopportunities available for institutingmolar systems change. The solution tothis dilemma lies in new researchpriorities that stress the inclusion ofnaturalistic variables the presence ofwhich would motivate researchers toanalyze and evaluate broad environmen-tal change. The conceptual underpinningsfor this expanded approach are clearlyevident in the literature on ecologicalsystems theory (Bronfenbrenner, 1989).This theory is compatible with theoperant perspective but also suggests

86 73

CHAPTER 4

additional useful practices for under-standing the individual's behavior incontext and for planning interventionsoriented toward broad systems change.The developmental literature providesevidence that for parents the ongoingpresence of informal social support thatincludes (a) multiple, high-quality, stablerelationships and (b) assistance frommembers of a broader interpersonalnetwork correlates strongly with parents'viewing their child's behavior as lesstroublesome or difficult (Dunst, Trivette,& Jodry, 1997). Observations such asthese should spur the field into examin-ing problem behavior from the perspec-tive of broader ecological variables.

Ecological ValidityOne might anticipate that the higherdegree of experimental control thatfrequently characterizes interventions inrestrictive situations (i.e., those involvingatypical settings and interventions thatdo not occur in all relevant contexts)might produce higher success rates,particularly when the intervention agentis an expert/professional. The database(Table 13) does not confirm this expecta-tion. Instead, typical agents are associatedwith higher success rates. Further, thesuccess rates involving typical versusatypical settings, and the presence versusabsence of intervention in all relevantcontexts, are roughly comparable. Onecould argue that atypical agents workwith individuals displaying more difficultproblem behavior, thus providing anexplanation for their lower success rates.However, as there is no consensus in thefield as to what metric should be used togauge level of difficulty, one cannot knowwhether this interpretation is correct. Inaddition, it is important to note that whentypical agents implemented intervention,

74 8'7

they did so only after they had referredthe case to one or more experts/profes-sionals who then helped them structurethe intervention. In other words, theatypical agents were still involved albeitin a consultative role rather than asimplementors.

It is encouraging that typicalintervention agents were able to producegood success rates, as ultimately these arethe people who must carry the brunt ofintervention in day-to-day situations. It iseven more encouraging that the generallyless controlled, less restrictive situationsrepresented by typical settings producedsuccess rates comparable to thoseobtained in atypical settings. Finally, it ismost encouraging that intervention in allrelevant contexts, truly a best practice,produced success rates at least as good asthose obtained with more circumscribedintervention that did not address allrelevant contexts. In sum, the ecologicaldata suggest that the involvement oftypical intervention agents may increasesuccess rates and that implementingintervention in less restrictive circum-stances (i.e., typical settings, all relevantcontexts) does not typically decrease thesuccess rate much below the 50% levelobtained generally. Apparently, with PBSintervention, it is possible to have bothsuccessful outcomes and ecologicalvalidity.

Implications forFuture Research

We have now reviewed the database forstudies that met our methodologicalcriteria for inclusion. Earlier we noted thebias in this database toward articles thatemphasized internal validity concerns.Although some portions of the databasedid address external validity concerns, it

DISCUSSION

was clear that the bulk of the literatureemphasizing these concerns is found inthe excluded research articles and, to agreat extent, among clinical reports anddescriptive accounts provided bypractitioners, parents, trainers, and othersconcerned with day-to-day support innaturalistic contexts. In Chapter 1, wealluded to the gap between the needs andinterests of researchers and those ofnonresearchers (consumers). Havingreviewed the database, we are now in aposition to weigh its strengths andweaknesses in light of nonresearcherconcerns so we can formulate an agendathat addresses knowledge gaps bybuilding on what is known while franklyacknowledging what is not known. Thisissue, of course, is the basis for the fifthand final research question to which wenow turn our attention.

How Responsive Is the PBSLiterature to the Needs of

Consumers (Nonresearchers)?Four priorities stand out in the literatureas pertaining especially to the needs ofconsumers: (a) comprehensive lifestylesupport, (b) long-term change,(c) practicality and relevance, and(d) consumer support issues. We willexamine each of these in turn.

Comprehensive LifestyleSupportFamilies and friends of people withdevelopmental disabilities are focused onissues pertaining to family life, jobs,community inclusion, supported living,and expanding social relationships(Risley, 1996; Ruef, 1997; Turnbull & Ruef,1996, 1997; Turnbull & Turnbull, 1996), inshort, comprehensive lifestyle support.The database, as a whole, does not reflect

this priority. First, only a tiny minority ofoutcomes involved goals of lifestylechange, implementation of lifestyleintervention, or reports of successfullifestyle change (Figure 15). Second, fewdata were reported on stimulus andresponse generalization (Table 8),measures that are indicative of broadchange. Third, the preponderance ofatypical settings (Figure 8), and therelative lack of intervention in all relevantcontexts (Figure 14) are not responsive tofamily perspectives that stress the needfor good adjustment to real-life settingsand round-the-clock support to achievethis goal. Fourth, the paucity of interven-tions that combine stimulus-based andreinforcement-based interventions into acomprehensive multicomponent ap-proach (Table 6) also indicates that theemphasis is not on providing interven-tions that demonstrate efficacy acrosscomplex, changing contexts such as thosethat typify community settings. Fifth, thecall for extensive environmental reorgani-zation as a crucial strategy for ensuringimprovement in lifestyle (Risley, 1996)has not been heeded. Few outcomes wereassociated with this type of systemschange (Table 12).

Yet the gaps just noted need notremain a permanent feature of theresearch literature. Even in the currentdatabase, there were indications that thissituation can improve. However, althoughthey were in the minority, there weredemonstrations of successful lifestylechange as well as stimulus and responsegeneralization. Also, some investigatorshave, with good effect, intervened in allrelevant contexts, used combined(multicomponent) interventions, andreorganized environments. There is also aclear trend toward implementing PBSmore often in typical settings. Taken as a

88 75

CHAPTER 4

whole, these data show that it is possibleto respond to consumer needs forcomprehensive lifestyle change. Indeed,this embryonic research base, whencombined with similar information fromthe excluded research studies as well asnumerous reports in the nonresearchliterature, fully justifies a major shift inresearch priorities toward analysis andevaluation of comprehensive lifestylesupport. As noted, all the elements of thisapproach are present in the literature.Generally lacking at the moment are (a)rules for their systematic combinationand (b) the scientific scrutiny of thoserules that is necessary to prove efficacy.

Long-Term ChangeConsumers tend to be concerned aboutproblem behavior over long periods oftime. Families note that advocacy effortsnecessary to achieve comprehensivelifestyle change can often take years(Nickels, 1996; Turnbull & Turnbull,1996). Both teachers and families statethat transitioning individuals frompreschool to elementary and high schooland then to the workplace and supportedliving requires a lifespan perspective thatviews the successful management ofproblem behavior as a never-endingprocess responsive to different challengesat different stages of life (Turnbull, 1988;Vandercook,York, & Forest, 1989). Notsurprisingly, then, the overall short-term,uncoordinated nature of programs andsupports has all parents, even those whoare currently satisfied with their pro-grams, worried about the future (Ruef,1997; Turnbull & Ruef, 1997). Thedatabase reveals a substantial gapbetween the needs of consumers for long-term demonstrations of efficacy and theinterests of researchers who followindividuals for short periods of time,most typically for less than 6 months and

876 9

in no case for more than 2 years (Table 9).

There are, however, both empiricaland conceptual grounds for believing thatthe gap can be closed. First, the databaseincludes a limited number of demonstra-tions of successful maintenance effectslasting up to 2 years. There is no a priorireason for assuming that the effectscannot be further extended, especiallygiven the excluded research studies andclinical reports, some of which note long-term maintenance. At a conceptual level,one might expect long-term effects if thePBS appioach were implemented in amanner more consistent with its generalphilosophy. To the extent that deficientenvironments and deficient skillscontinue to be identified over time, whichis almost always the case when onefollows an individual over many years inchanging life circumstances, PBSstrategies would have to be added and/ormodified. In other words, interventionnever stops. This view is in contrast tomany traditional studies in whichmaintenance is defined as durablesuccess following intervention cessation(Carr et al., 1990). In a truly comprehen-sive PBS approach, maintenance would beguaranteed because intervention wouldnever stop. Interestingly, an added benefitof such a long-term strategy is that, overtime, the individual is supported in manydifferent situations (a feature that wouldenhance stimulus generalization) and istaught many different skills (a feature thatwould enhance response generalization).Thus, comprehensive changes are likely tobe facilitated over the protracted periodsof time that PBS is in effect. In short,maintenance and comprehensive lifestylechange are intertwined variables.

Practicality and RelevanceUnless consumers view a research findingas being usable in their particular

DISCUSSION

circumstances, it is unlikely that it willguide day-to-day efforts to effect mean-ingful, long-term change. The gapbetween research and practice is high-lighted by the fact that researchersgenerally focus on methodological rigor,whereas consumers generally focus onpracticality and relevance (Carnine,1997).

Both interviews (Turnbull & Ruef,1996) and focus groups (Ruef, 1997)conducted with families indicate thatrelevance is most often defined aspertaining to lifestyle change. However, asnoted before, this topic is dealt with onlymarginally in the database. Families arealso concerned with a number ofpractical issues (Ruef, 1997) that gobeyond the simple desire for informationabout the nature of specific interventions:(a) How many people are necessary toimplement PBS, and can a parent do italone? (b) How much training does aparent need, and must one become anexpert to be successful? (c) What is theaverage number of hours per weeknecessary to effectively implement PBS inthe first 6 months? After 6 months? After1 year? (d) Can one be successful byimplementing only part of the PBSapproach? (e) Are there shortcuts (e.g.,simpler versions of functional assess-ment), and how does one sustain theenergy levels necessary for long-termefforts? (f) How can parents effectivelyimplement PBS when their own fears,embarrassment, and anger get in the way(maintaining composure)? (g) How canparents do an effective job in the face ofnegative reactions from friends, acquain-tances, family, and others in the commu-nity? The fact that the database rarelyconfronts these issues demonstrates theexistence of a wide research-to-practicegap that argues in favor of a future researchagenda centering on practicality criteria.

With respect to other consumergroups, the issues may vary, but the focuson practicality and relevance does not.Teachers, for example, regard propertraining for managing problem behavioras a top priority (Pearman, Huang, &Mellblom, 1997). Again, their concern isnot just with the specifics of interventionbut, rather, with how well interventionpractices will fit into the system in whichthey work (Ruef, 1997): (a) How can PBSbe implemented in a general-educationclassroom? (b) How does PBS vary as afunction of grade level (elementary,middle school, high school)? (c) Whatstructural modifications are required forimplementation in special-educationversus general-education settings?(d) What systems prerequisites must besatisfied for PBS to be a viable option(e.g., does it matter if you have one childwith a disability in a class of typicalchildren or a class composed entirely ofchildren with disabilities)? (e) Doesimplementing PBS with one student takeaway time spent with other students(fairness)? The database rarely touchesthese topics, a fact that is particularlyevident from the small number ofoutcomes associated with broad environ-mental reorganization, a key facet ofsystems change (Table 12). That teacherswidely perceive research to be irrelevantto their needs is also evident from datashowing that fewer than 1% of thenation's 4 million teachers participate inthe American Federation of TeachersEducation Research and Developmentprogram in which emphasis is placed onusing research to make informededucational decisions to guide practice(Billups,1997).

Other groups of consumers such asfriends of people with disabilities, policymakers, and people with disabilitiesthemselves raise questions about the

90 77

CHAPTER 4

complexity of systems change and therelevance of current research for dealingwith this complexity (Morrissey, 1997;Ruef, 1997).

Several developments in the fieldpoint to a way of closing the research-to-practice gap to enhance practicality andrelevance. Specifically the ParticipatoryAction Research (PAR) approach seesresearchers and consumers as collabora-tors (Meyer & Evans, 1993; Turnbull,Friesen, & Ramirez, 1998). Traditionally,researchers have defined the issues,formulated a plan for investigating theissues, and occasionally disseminated theresults to consumers (but more often toother researchers). The PAR model, incontrast, views consumers as having anactive rather than passive role. Consum-ers can play many roles that includehelping to define the issues, assisting inthe design of those aspects of the researchthat enhance ecological and socialvalidity, and providing consultation onhow to package the results so that theyare more readily usable by other consum-ers. Policy makers have called for thistype of collaboration for several yearsnow (Lloyd, Weintraub, & Safer, 1997;Malouf & Schiller, 1995). However, withvirtually no exceptions, the PAR modelwas not a feature associated with ourdatabase. Nonetheless, it is encouragingto note that in a recent issue of theJournal of Special Education devoted toresearch in severe disabilities, multipleresearchers independently endorsed theidea of making the PAR model a criticalcomponent of a future research agendafor the field (Browder, 1997; Nietupski,Hamre-Nietupski, Curtin, & Shrikanth,1997; Reichle, 1997). In sum, althoughparents and teachers, for example, willtypically not have the expertise toenhance the technical design aspects of

78 91

research, they can and should play amajor role in enhancing ecologicalandsocial validity and in identifying andresolving barriers to achieving successfuloutcomes.

Although the database on socialvalidity was scant (Figure 16), it issomewhat encouraging to note that thedata were generally positive with respectto issues of feasibility, desirability,acceptability, and lifestyle change,suggesting that, at least for the limitedsample of consumers involved, PBS wasviewed as practical and relevant. The fieldneeds to build on this base by developingresearch models that reflect the majorelements of the PAR approach. In thisregard, recent work on goodness-of-fitmodels (Albin, Lucyshyn, Horner, &Flannery, 1996) seem especially relevant.This research strategy stresses the notionthat interventions must be congruentwith contextual variables involvingparticipant characteristics, characteristicsof the people who will be implementingthe plan, and systems features related tothe environment in which the plan is tobe implemented. Consumer needs andpriorities are critical. Recent researchdemonstrates clearly that a focus ongoodness-of-fit not only produces sociallyvalued intervention outcomes but is thelogical translation of the PAR philosophyinto scientific methodology (Lucyshyn eta1.,1995,1997).

Consumers Want Support, TooIt is easy to misinterpret PBS as referringonly to the person with disabilities,namely, how best to support that personthrough skills training, environmentalreorganization, and a focus on improvinglifestyle. However, as we noted, PBS refersto systems change broadly conceived, thatis, all elements of the system including

DISCUSSION

the needs of other people who mustsupport the person with disabilities. Thus,a critical theme that has emerged in theliterature is that consumers such asparents and teachers require support.

Some researchers have arguedforcefully that professional supportshould be family-centered and entailcomprehensive systems change involvingall relevant parties and not just theperson with disabilities (Albin et al.,1996; Dunlap & Robbins, 1991). Nonethe-less, parents continue to report a lack offamily-centered service delivery systemsand an overall lack of support fromprofessionals paid to deliver services(Wheeler, 1996). Few examples exist inthe literature that demonstrate supportfor the people (families) who must carrythe weight of supporting others (c.f.,Lucyshyn et al., 1995, 1997; The FamilyConnection staff et al., 1996; Turnbull &Turnbull, 1996).

Interestingly, although both parentsand teachers value advice and trainingfrom professionals, they particularlyvalue parent-to-parent and teacher-to-teacher mentoring (Gersten &Brengelman, 1996; Gersten, Morvant, &Brengelman, 1995; Ruef, 1997; Santelli,Turnbull, Marquis, & Lerner, 1993, 1995).Teachers also value emotional supportfrom principals and other administratorsas a way of coping with high job stress inworking with challenging populations

(Fimian, 1986; Littrell, Billingsley, &Cross, 1994).

In some limited respects, thedatabase touches on several of theconcerns just described. Thus, the factthat typical intervention agents workingin typical settings are sometimesintervening in all relevant contexts (Table13) is a step in the right direction,because this pattern implies family-centered and school-centered services.However, as noted previously, this patternis associated with only a minority ofoutcomes in the database. More signifi-cant perhaps, the database does not show,in any systematic way, a concern for thebroader needs of consumers that involvesissues such as peer mentoring, emotionalsupport, and stress reduction. Instead, apremium is placed on teaching specificintervention strategies at the expense of afocus on the broader systems issuesrelated to supporting consumers who willbe responsible for facilitating lifestylechanges for decades (i.e., parents often dothis for 50 or more years, and siblings canhave this role for even longer). This gap inthe research literature can be addressedonly by enlarging the conception of bestpractices to include the systematic assess-ment of consumer needs and by acting onthose needs by designing supports for thesupportersfamilies, teachers, job coaches,and other community-based staff.

9279

SUMMARY

CHAPTER 5

SUMMARYWe now summarize the findings with thegreatest clinical and conceptual signifi-cance as they relate to the five researchquestions posed at the beginning of thisreview.

How WidelyApplicable is PBS?

1. The approach is widely applicable topeople with serious disabilities whoexhibit serious problem behavior.

2. The approach can be applied bytypical intervention agents in typicalsettings. Its use is not restricted toexperts operating in specializedcircumstances although a majority ofoutcomes are still associated with thispattern.

In What Ways isthe Field Evolving?

1. PBS is not a fad. It has been showingsteady and dramatic growth especiallyover the most recent time periodreviewed.

2. Outcomes involving the most difficultproblem behavior (e.g., SIB andcombinations of problem behaviors)have become more numerous.

3. Generally, typical and atypicalintervention agents are about equallylikely to conduct PBS intervention.With respect to intervention settings,atypical settings are more likely to bethe venue for intervention. However,there is a steady increase over time inthe use of typical settings, and the gapbetween the use of typical versusatypical settings is closing.

4. There has been a dramatic increaseover time in the use of assessmentprior to planning intervention.

5. Both formal functional analysis andcombination assessments havebecome more numerous. There isconcern, however, as to whetherfunctional analysis is a practicalmethod in naturalistic contexts.

6. There is a clear focus on escape-motivated problem behavior and onproblem behaviors that are multiplymotivated.

7. Repeated assessments have remaineduncommon. When they do occur, theyare likely to be carried out by typicalintervention agents and in typicalsettings.

8. Stimulus-based intervention hasgradually become more common thanreinforcement-based intervention.

9. Combination interventions do notshow an increasing trend. When theydo occur, they are likely to be carriedout by typical agents and in typicalsettings.

10. With respect to systems change,significant others are increasinglylikely to alter their behavior as part ofthe intervention. No trend is seen,however, for environmental reorgani-zation, a critical aspect of systemschange.

11. Intervention in all relevant contextsdoes not show a trend. When it doesoccur, it is likely to be done by typicalagents and in typical settings.

12. There is an increase over time inoutcomes associated with short-term

93 81

CHAPTER 5

maintenance (5 months or less) butnone for longer-term maintenance.

13. Outcomes associated with generaliza-tion, lifestyle change, and socialvalidity show no trends and are few innumber.

How Effective is PBS?1. Modest to substantial increases in

positive behavior are typicallyobserved following the application ofPBS intervention.

2. With respect to reduction in problembehavior, about one-half (using acriterion of 90% or more decrease inproblem behavior from baseline) totwo-thirds (using an 80% criterion) ofthe outcomes are successes.

3. Typically, the success rate does notchange as a function of whetherstimulus-based intervention andreinforcement-based intervention areused alone or in combination witheach other, nor does it change whennon-PBS interventions are added.

4. With respect to maintenance effects,about two-thirds of the outcomes aresuccesses (using a 90% reductioncriterion). However, the database issmall and gets smaller as the durationof follow-up is lengthened.

5. There are demonstrations of success-ful lifestyle change and good socialvalidity, but these data are reportedonly for a very small minority ofoutcomes.

9482

What Factors Modulatethe Effectiveness of PBS?

1. Intervention for combinations ofproblem behavior produces lowersuccess rates than interventions forsingle types of problem behavior.

2. The success rate (using a 90%reduction criterion) for interventionsbased on a prior functional assess-ment was almost twice that obtainedwhen this type of assessment was notconducted.

3. The success rate (90% criterion) wasgreater following interventions thatinvolved systems change (althoughthe database for one aspect of systemschange, environmental reorganization,was very small).

4. The success rate (90% criterion)associated with typical agents washigher than that obtained withatypical agents. The success rates intypical versus atypical settings werecomparable.

How Responsive isthe PBS Literature to the

Needs of Consumers(Nonresearchers)?

Comprehensive lifestyle support is amajor goal of families, but thedatabase rarely addressed this issue.

2. Families are most concerned withlong-term behavior change. Thedatabase, in a minority of cases,demonstrated such change. However,no outcomes involved follow-up of

1.

SUMMARY

longer than 2 years, which posesdifficulties for the vast majority offamilies that think of maintenance interms of decades, rather than months.

3. Consumers (e.g., parents, teachers)judge interventions in terms of theirpracticality and relevance and areconcerned with how well interventionplans mesh with the realities of thecomplex social systems in which theconsumers must function. The

database, more concerned with issuesof rigor and demonstrations ofexperimental control, generally failedto focus on larger consumer goals.

4. Consumers are concerned withobtaining support for themselves inaddition to support for people withdisabilities. This topic was not a focusof systematic research in the studiesincluded in the database.

95

83

RECOMMENDATIONS

CHAPTER 6

RECOMMENDATIONS

For ResearchersThe main recommendation to researchersis that we need a new applied science. Fora truly applied science to develop,researchers need to address consumerneeds more systematically and morefrequently. A robust applied science willrequire changes in assessment andintervention practices, a redefinition ofwhat constitutes a significant outcome,and measurement procedures thataddress these new priorities.

With respect to assessment, methodsmust be developed that are user-friendly,are feasible in the community, and yieldaccurate information. Functional analysismeets the last criterion but not the firsttwo; it therefore needs to be supple-mented and often replaced with a newgeneration of assessment tools that meetall three criteria and can be implementedrepeatedly as circumstances warrant.

With respect to advancing interven-tion practices, researchers need to focuson consumer goals pertaining to compre-hensive lifestyle support, long-termchange, and direct support to consumersthemselves. These goals can mostplausibly be met by anincreasingemphasis on multicomponent interven-tions that are linked to assessmentinformation, broad reorganization ofcontext (systems change), an emphasis onecological validity (typical agents andtypical settings), intervention in allrelevant contexts, and the application ofall these practices over protracted periodsof time. The consumer goals just notedtogether with the goals of practicality andrelevance can be best met by focusing on

how well intervention practices fit withspecific contexts (goodness-of-fit), andthis determination can be made throughthe active participation of consumers inthe research process as advisers and/orcollaborators (Participatory ActionResearch).

Finally, with respect to outcomes, thePAR model needs to be adapted as well inorder to define, in specific terms, thedimensions of new outcome measuresdefined primarily in terms of long-term,socially validated, comprehensive lifestylechange and only secondarily in terms ofreductions in problem behavior.

For Service ProvidersThe main recommendation to serviceproviders is to fix problem contexts, notproblem behavior. Problem contexts (i.e.,environmental deficiencies and skilldeficits) are the fertile ground from whichproblem behavior springs. Therefore,whenever problem behavior is identifiedas a clinical issue, service providers oughtfirst and foremost to structure interven-tion so that it reflects a knowledge(derived from systematic assessment) ofwhat is wrong with the individual'senvironment (e.g., educational practices,scheduling issues, lack of control) andskill repertoire (e.g., lack of communica-tion, poorly developed social skills,insufficient self-management). Suchassessment information can then be usedto redesign the environment and enhancethe individual's adaptive skill repertoire.The primary effect of this approach is tostrengthen positive behaviors. Animportant but secondary effect is toproduce a decrease in problem behavior.

96 85

CHAPTER 6

Reflecting consumer needs, serviceproviders need to offer hands-on, ongoing(rather than episodic) support to families,teachers, and staff over long periods of time.This strategy has the dual effect of support-ing the consumers themselves andgradually permitting a transfer of supporttactics from the professional serviceprovider to typical intervention agents.

The PAR model for researchers is alsoappropriate for service providers. Specifi-cally, service providers need to transformtheir role from experts who unilaterallyselect goals and structure interventions, tocollaborators who, in consort with consum-ers, define the dimensions of comprehensivelifestyle change (thereby ensuring socialvalidity) in a manner that represents a goodfit with the reality of day-to-day contextualconstraints (thereby ensuring practicalityand relevance).

For SocialPolicy Advocates

The main recommendation for socialpolicy advocates is that regulationsdefining quality of services need tomandate standards of best practice. Thesestandards should involve or address:

repeated functional assessments thatidentify, on an ongoing basis, theenvironmental and behavioraldeficiencies that are the root cause ofproblem behavior;

direct linkage between assessment infor-mation and the design of interventions;

intervention in all relevant contexts, astrategy that almost invariably meansthe use of multicomponent interven-tions geared toward altering systems,not just discrete behaviors;

ecologically valid relevant contexts(i.e., typical agents carry out interven-tion in typical settings);

86 97,

the long-term perspectives ofconsumersby designing andredesigning interventions as changesin life circumstances warrant; that is,intervention plans must have alifespan orientation rather than acrisis management orientation;

consumers being an integral part ofthe system by constructing interven-tions that respond to the personalneeds and concerns of consumers(goodness-of-fit) thereby ensuringpracticality and relevance;

social validity issues, definingoutcome goals in terms of compre-hensive lifestyle change and supportand not just reduction in problembehavior.

For the GovernmentThe main recommendation for govern-ment is that resources should be investedto ensure the continued development andevaluation of a truly applied science ofPBS that is sensitive to consumer needs.Government can strengthen this processby (a) developing grant competitions thatrequire adherence to best practices,emphasizing the demonstration ofsocially valid comprehensive lifestylechange rather than microanalysis ofcause-and-effect processes in situationsthat lack ecological validity; (b) creating anationally accessible database on PBSthat is updated periodically so thatconsumers, advocates, policy analysts,and researchers can study the database tosee what progress is being made, what thegaps are, and what future directionsmight be fruitful; (c) convening periodicstate-of-the-art conferences that definewhere we are and how PBS needs tochange to keep abreast of new develop-ments in policy, advocacy, consumerneeds, and research findings.

REFERENCES

REFERENCES

References marked with an asteriskindicate studies included in the synthesis.

Albin, R. W., Lucyshyn, J. M., Homer,R. H., & Flannery, K. B. (1996).Contextual fit for behavior supportplans. In L. K. Koegel, R. L. Koegel, &

G. Dunlap (Eds.), Positive behavioralsupport: Including people with difficultbehavior in the community (pp. 8198). Baltimore: Paul H. Brookes.

*Albin, R. W, O'Brien, M., & Homer, R. H.(1995). Analysis of an escalatingsequence of problem behaviors: Acase study. Research in DevelopmentalDisabilities, 16, 133-147.

Aman, M. G., Singh, N. N., Stewart, A. W,& Field, C. J. (1985). Psychometriccharacteristics of the AberrantBehavior Checklist. American Journalof Mental Deficiency, 89, 492-502.

Arendt, R. E., MacLean, W. E., &Baumeister, A. A. (1988). Critique ofsensory integration therapy and itsapplication in mental retardation.American Journal on Mental Retarda-tion, 92, 401-411.

Axelrod, S. (1987). Doing it withoutarrows. [Review of the book Alterna-tives to punishment: Solving behaviorproblems with non-aversive strategies].The Behavior Analyst, 10, 243-251.

Baer, D. M., Wolf, M. M., & Risley, T. R.

(1968). Some current dimensions ofapplied behavior analysis. Journal ofApplied Behavior Analysis, 1, 91-97.

Bailey, J. S., & Pyles, D.A. M. (1989).Behavioral diagnostics. In E. Cipani(Ed.), The treatment of severe behaviordisorders (Monograph No. 12, pp. 85-107). Washington, DC: AmericanAssociation on Mental Retardation.

*Ballard, K. D., & Medland, J. L. (1986).Collateral effects from teachingattention, imitation, and toy interac-tion behaviors to a developmentallyhandicapped child. Child and FamilyBehavior Therapy, 7(4), 47-60.

*Bambara, L. M., Koger, E, Katzer, T., &Davenport, T. A. (1995). Embeddingchoice in the context of daily routines:An experimental case.study. Journal ofThe Association for Persons WithSevere Handicaps, 20, 185-195.

Bannerman, D. J., Sheldon, J. B., Sherman,J. A., & Harchik, A. E. (1990). Balanc-ing the right to habilitation with theright to personal liberties: The rightsof people with developmentaldisabilities to eat too many doughnutsand take a nap. Journal of AppliedBehavior Analysis, 23, 79-89.

Barrett, R. P. (Ed.). (1986). Severe behaviordisorders in the mentally retarded. NewYork: Plenum.

Bijou, S. W., & Baer, D. M. (1961). Childdevelopment I: A systematic andempirical theory. Englewood Cliffs, NJ:Prentice-Hall.

Bijou, S.W., Peterson, R. F., & Ault, M. H.(1968).A method to integratedescription and experimental fieldstudies at the level of data andempirical concepts. Journal of AppliedBehavior Analysis, 1, 175-191.

98 87

REFERENCES

Billingsley, B. S., & Cross, L. H. (1991).Teachers' decisions to transfer fromspecial to general education. Journalof Special Education, 24, 496-511.

Billups, L. H. (1997). Response to bridgingthe research-to-practice gap. Excep-tional Children, 63, 525-527.

Bird, E, Dores, P. A., Moniz, D., &Robinson, J. (1989). Reducing severeaggressive and self-injurious behav-iors with functional communicationtraining. American Journal on MentalRetardation, 94, 37-48.

*Blum, N. J., Mauk, J. E., McComas, J. J., &Mace, F. C. (1996). Separate andcombined effects of methylphenidateand a behavioral intervention ondisruptive behavior in children withmental retardation. Journal of AppliedBehavior Analysis, 29, 305-319.

Bodfish, J. W., & Madison, J. T. (1993).

Diagnosis and fluoxetine treatment ofcompulsive behavior disorder ofadults with mental retardation.American Journal on Mental Retarda-tion, 98, 360-367.

*Bowman, L. G., Fisher, W.W., Thompson,R. H., & Piazza, C. C. (1997). On therelation of mands and the function ofdestructive behavior. Journal ofApplied Behavior Analysis, 30, 251-265.

Braddock, D., Hemp, R., Fujiura, G.,Bachelder, L., & Mitchell, D. (1990).The state of the states in developmentaldisabilities. Baltimore: Paul H.Brookes.

Bronfenbrenner, U. (1989). Ecologicalsystems theory. In R. Vasta (Ed.),Annals of Child Development (Vol. 6,pp. 187-249). Greenwich, CT: JAIPress.

88 do

Browder, D. M. (1997). Educating studentswith severe disabilities: Enhancing theconversation between research andpractice. The Journal of SpecialEducation, 31, 137-144.

Brown, F (1991). Creative daily schedul-ing: A nonintrusive approach tochallenging behaviors in communityresidences. Journal of The Associationfor Persons With Severe Handicaps, 16,75-84.

Browning, R., & Stover, D. (1971).Behavior modification in childtreatment: An experimental andclinical approach. Chicago: AldineAtherton.

Bruyere, S. M. (1993). Participatory actionresearch: Overview and implicationsfor family members of persons withdisabilities. Journal of VocationalRehabilitation, 3(2), 62-68.

Bucher, B., & Lovaas, 0. I. (1968). Use ofaversive stimulation in behaviormodification. In M. R. Jones (Ed.),Miami symposium on the prediction ofbehavior, 1967: Aversive stimulation(pp. 77-145). Coral Gables, FL:University of Miami Press.

Busk, P. L., & Serlin, R. C. (1992). Meta-

analysis for single-case research. In T.Kratochwill and J. Levin (Eds.), Singlecase research design and analysis: Newdirections for psychology and educa-tion (pp. 187-212). Hillsdale, NJ:Lawrence Erlbaum.

Cameron, M. J., Ainsleigh, S. A., & Bird,F. L. (1992). The acquisition ofstimulus control of compliance andparticipation during an ADL routine.Behavioral Residential Treatment, 7,327-340.

REFERENCES

*Cameron, M. J., Luiselli, J. K., McGrath,M., & Carlton, R. (1992). Stimuluscontrol analysis and treatment ofnoncompliant behavior. Journal ofDevelopmental and Physical Disabili-ties, 4, 141-150.

*Campbell, R. V., & Lutzker, J. R. (1993).Using functional equivalence trainingto reduce severe challenging behavior:A case study. Journal of Developmentaland Physical Disabilities, 5, 203-216.

Carnine, D. (1997). Bridging the research-to-practice gap. Exceptional Children,63, 513-521.

Carr, E. G. (1977). The motivation of self-injurious behavior: A review of somehypotheses. Psychological Bulletin, 84,

800-816.

Carr, E.G. (1988). Functional equivalenceas a mechanism of response generali-zation. In R. H. Horner, R. L. Koegel, &G. Dunlap (Eds.), Generalization andmaintenance: Lifestyle changes inapplied settings (pp. 194-219).Baltimore: Paul H. Brookes.

Carr, E.G. (1993). Behavior analysis is notultimately about behavior. TheBehavior Analyst, 16, 47-49.

Carr, E.G. (1994). Emerging themes in thefunctional analysis of problembehavior. Journal of Applied BehaviorAnalysis, 27, 393-399.

*Carr, E.G., & Carlson, J. I. (1993).Reduction of severe behavior prob-lems in the community using amulticomponent treatment approach.Journal of Applied Behavior Analysis,26, 157-172.

Carr, E. G., Carlson, J. I., Langdon, N. A.,Magito McLaughlin, D., & Yarbrough,

S. C. (1998). Two perspectives onantecedent control: Molecular andmolar. In J. K. Luiselli & M. J. Cameron(Eds.), Antecedent control: Innovativeapproaches to behavioral support (pp.3-28). Baltimore: Paul H. Brookes.

*Carr, E. G., & Durand, V. M. (1985).Reducing behavior problems throughfunctional communication training.Journal of Applied Behavior Analysis,18,111 -126.

Carr, E.G., Levin, L., McConnachie, G.,Carlson, J. I., Kemp, D. C., & Smith,

C. E. (1994). Communication-basedintervention for problem behavior. Auser's guide for producing positivechange. Baltimore: Paul H. Brookes.

Carr, E. G., & Lovaas, 0. I. (1983). Contin-gent electric shock as a treatment forsevere behavior problems. In S.Axelrod & J. Apsche (Eds.), Punish-ment: Its effects on human behavior(pp. 221-245). New York: AcademicPress.

Carr, E. G., & McDowell, J. J. (1980). Socialcontrol of self-injurious behavior oforganic etiology. Behavior Therapy, 11,

402-409.

*Carr, E.G., & Newsom, C. (1985).Demand-related tantrums:Conceptualization and treatment.Behavior Modification, 9, 403-426.

Carr, E. G., Newsom, C. D., & Binkoff, J. A.

(1976). Stimulus control of self-destructive behavior in a psychoticchild. Journal of Abnormal ChildPsychology, 4, 139-153.

10.089

REFERENCES

Carr, E. G., Newsom, C. D., & Binkoff, J. A.(1980). Escape as a factor in theaggressive behavior of two retardedchildren. Journal of Applied BehaviorAnalysis, 13, 101-117.

Carr, E. G., Robinson, S., Taylor, J. C., &Carlson, J. I. (1990). Positive ap-proaches to the treatment ofseverebehavior problems in persons withdevelopmental disabilities: A reviewand analysis of reinforcement andstimulus-based procedures. (Mono-graph No. 4). Seattle: The Associationfor Persons With Severe Handicaps.

Carr, E. G., & Smith, C. E. (1995). Biologi-cal setting events for self-injury.Mental Retardation and Developmen-tal Disabilities Research Reviews, 1,94-98.

Cataldo, M. E (1991). The effects ofpunishment and other behavior-reducing procedures on the destructivebehaviors of persons with developmen-tal disabilities. In Treatment of destruc-tive behaviors in persons with develop-mental disabilities (NIH Publication No.91-2410, pp. 231-341). Washington, DC:National Institutes of Health.

*Cataldo, M. F., Ward, E. M., Russo, D. C.,Riordan, M., & Bennett, D. (1986).Compliance and correlated problembehavior in children: Effects of contin-gent and noncontingent reinforcement.Analysis and Intervention in Develop-mental Disabilities, 6, 265-282.

*Charlop, M. H., Kurtz, P. E, & Milstein,J. P. (1992). Too much reinforcement,too little behavior: Assessing taskinterspersal procedures in conjunc-tion with different reinforcementschedules with autistic children.Journal of Applied Behavior Analysis,25, 795-808.

90 1 0 1

*Clarke, S., Dunlap, G., Foster-Johnson, L.,Child's, K. E., Wilson, D., White, R., &Vera, A. (1995). Improving theconduct of students with behavioraldisorders by incorporating studentinterests into curricular activities.Behavioral Disorders, 20(4), 221-237.

Cohen, J. (1960).A coefficient of agree-ment for nominal scales. Educationaland Psychological Measurement, 20,37-46.

*Cooper, L. J., Wacker, D. P., Thursby, D.,

Plagmann, L.A., Harding, J., Millard,T., & Derby, M. (1992). Analysis of theeffects of task preferences; taskdemands, and adult attention on childbehavior in outpatient and classroomsettings. Journal of Applied BehaviorAnalysis, 25, 823-840.

Copeland, K. (1997). Worming our wayinto the system, or as the worm turns.The Communicator, 8(2), .20-22.

Dadson, S., & Homer, R. H. (1993).Manipulating setting events todecrease problem behaviors: A casestudy. Teaching Exceptional Children,25(3), 53-55.

*Day, H. M., Homer, R. H., & O'Neill, R. E.(1994). Multiple functions of problembehaviors: Assessment and interven-tion. Journal of Applied BehaviorAnalysis, 27, 279-289.

*Day, R. M., Rea, LA., Schussler, N. G.,Larsen, S. E., & Johnson, W. L. (1988).

A functionally based approach to thetreatment of self-injurious behavior.Behavior Modification, 12, 565-589.

Dennis, R. E., Williams, W., Giangreco,M. E, & Cloninger, C. J. (1993). Qualityof life as a context for planning andevaluation of services for people withdisabilities. Exceptional Children, 59,499-512.

REFERENCES

Derby, K. M., Wacker, D. P., Sasso, G.,Northup, J., Cigrand, K., & Asmus, J.

(1992). Brief functional assessmenttechniques to evaluate aberrantbehavior in an outpatient setting: Asummary of 79 cases. Journal ofApplied Behavior Analysis, 25,

713-721.

Desrochers, M. N., Hile, M. G., &Williams-Moseley, T. L. (1997). Survey

of functional assessment proceduresused with individuals who displaymental retardation and severeproblem behaviors. American Journalon Mental Retardation, 101, 535-546.

Didden, R., Duker, P. C., & Korzilius, H.(1997). Meta-analytic study ontreatment effectiveness for problembehaviors with individuals who havemental retardation. American Journalon Mental Retardation, 101, 387-399.

Donnellan, A. M., LaVigna, G. W., Negri-Shoultz, N., & Fassbender, L. L. (1988).Progress without punishment: Effectiveapproaches for learners with severebehavior problems. New York: Teachers

College Press.

*Donnellan, A. M., LaVigna, G. W.,Zambito, J., & Thvedt, J. (1985). Atime-limited intensive interventionprogram model to support commu-nity placement for persons withsevere behavior problems. Journal ofThe Association for Persons WithSevere Handicaps, 10, 123-131.

*Ducharme, J. M., Lucas, H., & Pontes, E.(1994). Errorless embedding in thereduction of severe maladaptivebehavior during interactive andlearning tasks. Behavior Therapy, 25,

489-501.

*Ducharme, J. M., Pontes, E., Guger, S.,Crozier, K., Lucas, H., & Popynick, M.(1994). Errorless compliance toparental requests II: Increasingclinical practicality through abbrevia-tion of treatment parameters.Behavior Therapy, 25, 469-487.

*Ducharme, J. M., & Popynik, M. (1993).Errorless compliance to parentalrequests: Treatment effects andgeneralization. Behavior Therapy, 24,

209-226.

Dunlap, G., dePerczel, M., Clarke, S.,Wilson, D., Wright, S., White, R., &Gomez, A. (1994). Choice making andproactive behavioral support forstudents with emotional and behav-ioral challenges. Journal of AppliedBehavior Analysis, 27, 505-518.

*Dunlap, G., Foster-Johnson, L., Clarke, S.,Kern, L., & Childs, K. E. (1995).Modifying activities to producefunctional outcomes: Effects on theproblem behaviors of students withdisabilities. Journal of The Associationfor Persons With Severe Handicaps, 20,248-258.

*Dunlap, G., Kern-Dunlap, L., Clarke, S., &

Robbins, E R. (1991). Functionalassessment, curricular revision, andsevere problems. Journal of AppliedBehavior Analysis, 24, 387-397.

Dunlap, G., & Koegel, R. L. (1980).Motivating autistic children throughstimulus variation. Journal of AppliedBehavior Analysis, 13, 619-627.

Dunlap, G., & Robbins, F. R. (1991).Current perspectives in servicedelivery for young children withautism. Comprehensive Mental Health

Care, 1,177-189.

102 91

REFERENCES

Dunlap, G., Robbins, F. R., & Darrow, M.A.(1994). Parents' reports of theirchildren's challenging behaviors:Results of a statewide survey. MentalRetardation, 32, 206-212.

Dunst, C. J., Trivette, C. M., & Jo dry, W.(1997). Influences of social support onchildren with disabilities and theirfamilies. In M. J. Guralnick (Ed.), Theeffectiveness of early intervention (pp.499-522). Baltimore: Paul H. Brookes.

Durand, V. M. (1990). Functional communi-cation training: An intervention programfor severe behavior problems. New York:Guilford.

*Durand, V. M. (1993). Functional commu-nication training using assistivedevices: Effects on challenging behaviorand affect. Augmentative and Alterna-tive Communication, 9, 168-176.

*Durand, V. M., & Carr, E.G. (1991).

Functional communication training toreduce challenging behavior: Mainte-nance and application in new settings.Journal of Applied Behavior Analysis, 24,251-264.

*Durand, V. M., & Carr, E.G. (1992). Ananalysis of maintenance followingfunctional communication training.Journal of Applied Behavior Analysis, 25,777-794.

Durand,V. M., & Crimmins, D. B. (1988).Identifying the variables maintainingself-injurious behavior. Journal ofAutism and Developmental Disorders,18,99-117.

Durand,V. M., & Crimmins, D. B. (1992).The motivation assessment scale (MAS)administration guide. Topeka, KS:Monaco & Associates.

10392

*Durand,V. M., & Kishi, G. (1987).

Reducing severe behavior problemsamong persons with dual sensoryimpairments: An evaluation of atechnical assistance model. Journal ofThe Association for Persons WithSevere Handicaps, 12, 2-10.

*Dyer, K., Dunlap, G., & Winterling, V.(1990). Effects of choice making onthe serious problem behaviors ofstudents with severe handicaps.Journal of Applied Behavior Analysis,23, 515-524 .

*Egan, P. J., Zlomke, L. C., & Bush, B. R.(1993). Utilizing functional assess-ment, behavioral consultation andvideotape review of treatment toreduce aggression: A case study.Special Services in the Schools, 7,27-37.

Emerson, E., McGill, P., & Mansell, J.(Eds.). (1994). Severe learningdisabilities and challengingbehaviours. London: Chapman & Hall.

Evans, I. M., & Meyer, L. H. (1985). Aneducative approach to behaviorproblems: A practical decision modelfor interventions with severelyhandicapped learners. Baltimore: PaulH. Brookes.

Evans, I. M., & Scotti, J. R. (1989).

Defining meaningful outcomes forpersons with profound disabilities. InF. Brown & D. Lehr (Eds.), Persons

with profound disabilities: Issues andpractices (pp. 83-107). Baltimore: PaulH. Brookes.

The Family Connection staff, DeVault, G.,Krug, C., & Fake, S. (1996). Why doesSamantha act that way? ExceptionalParent, 26(9), 43-47.

REFERENCES

Favell, J. E., McGimsey, J. F., & Schell, R. M.

(1982). Treatment of self-injury byproviding alternate sensory activities.Analysis and Intervention in Develop-mental Disabilities, 2, 83-104.

Feldman, M. A. (1990). Balancingfreedom from harm and right totreatment for persons with develop-mental disabilities. In A. C. Repp & N.N. Singh (Eds.), Perspectives on the useof nonaversive and aversive interven-tions for persons with developmentaldisabilities (pp. 261-271). Sycamore,IL: Sycamore.

Fimian, M. J. (1986). Social supportand occupational stress in specialeducation. Exceptional Children, 52,436-442.

*Fisher, W W., Piazza, C. C., Bowman,L. G., Hanley, G. P., &Adelinis, J. D.(1997). Direct and collateral effectsof restraints and restraint fading.Journal of Applied Behavior Analysis,30, 105-119.

*Fisher, W. W., Piazza, C. C., Cataldo, M. F.,

Harrell, R., Jefferson, G., & Conner, R.

(1993). Functional communicationtraining with and without extinctionand punishment. Journal of AppliedBehavior Analysis, 26, 23-36.

*Fisher, W W., Piazza, C. C., & Page, T. J.

(1989). Assessing independent andinteractive effects of behavioral andpharmacologic interventions for aclient with dual diagnoses. Journal ofBehavior Therapy and ExperimentalPsychiatry, 20, 241-250.

*Flannery, K. B., & Horner, R. H. (1994).

The relationship between predictabil-ity and problem behavior for studentswith severe disabilities. Journal ofBehavioral Education, 4, 157-176.

Foxx, R. M. (1982). Decreasing behaviorsof severely retarded and autisticpersons. Champaign, IL: ResearchPress.

Foxx, R. M. (1990). "Harry": A ten yearfollow-up of the successful treatmentof a self-injurious man. Research inDevelopmental Disabilities, 11, 67-76.

Foxx, R. M., Bittle, R. G., & Faw, G. D.

(1989). A maintenance strategy fordiscontinuing aversive procedures: A52-month follow-up of the treatmentof aggression. American Journal onMental Retardation, 94, 27-36.

*Friman, P. C., Barnard, J. D., Altman, K.,& Wolf, M. M. (1986). Parent andteacher use of DRO and DRI to reduceaggressive behavior. Analysis andIntervention in DevelopmentalDisabilities, 6, 319-330.

Fuchs, D., & Fuchs, L. S. (1990). Makingeducational research more important.Exceptional Children, 57, 102-107.

Gardner, W. I., Cole, C. L., Berry, D. L., &Nowinski, J. M. (1983). Reduction ofdisruptive behaviors in mentallyretarded adults: A self-managementapproach. Behavior Modification, 7,76-96.

Gardner, W. I., & Sovner, R. (1994). Self-injurious behavior. Willow Street, PA:

VIDA.

Gardner, W. I., & Whalen, J. P. (1996).Discussion: A multimodal behavioranalytic model for evaluating theeffects of medical problems onnonspecific behavioral symptomsin persons with developmentaldisabilities. Behavioral Interventions,11, 147-161.

104 93

REFERENCES

Gersten, R., & Brengelman, S. U. (1996).The quest to translate research intoclassroom practice. Remedial andSpecial Education, 17, 67-74.

Gersten, R., Morvant, M., & Brengelman,S. (1995). Close to the classroom isclose to the bone: Coaching as ameans to translate research intoclassroom practice. ExceptionalChildren, 62, 52-66.

*Grace, N., Cowart, C., & Matson, J. L.(1988). Reinforcement and self-control for treating a chronic case ofself-injury in Lesch-Nyhan syndrome.Journal of the MultihandicappedPerson, 1, 53-59.

Green, G., & Shane, H. C. (1994). Science,reason, and facilitated communica-tion. Journal of The Association forPersons With Severe Handicaps, 19,151-172.

Guess, D., Helmstetter, E., Turnbull, H. R.III, & Knowlton, S. (1987). Use ofaversive procedures with persons whoare disabled: An historical review andcritical analysis. (Monograph No. 2).Seattle: The Association for PersonsWith Severe Handicaps.

Guess, D., & Siegel-Causey, D. (1995).Attractor dimensions of behavior statechanges among individuals withprofound disabilities. AmericanJournal on Mental Retardation, 99,642-663.

Hall R. V., Lund, D., & Jackson, D. (1968).

Effects of teacher attention on studybehavior. Journal of Applied BehaviorAnalysis, 1, 1-12.

94 105

*Hanley, G. P., Piazza, C. C., & Fisher,W. W. (1997). Noncontingentpresentation of attention andsubstitute stimuli in the treatmentof attention-maintained destructivebehavior. Journal of Applied Behav-ior Analysis, 30, 229-237.

Haring, T. G. (1996). The role of researchin the continuing development ofthe field of severe disabilities.Journal of The Association forPersons With Severe Handicaps, 21,6-8.

Haring, T. G., & Kennedy, C. H. (1990).Contextual control of problembehavior in students with severedisabilities. Journal of AppliedBehavior Analysis, 23, 235-243.

Harris, S. L., & Ersner-Hershfield, R.(1978). Behavioral suppression ofseriously disruptive behavior inpsychotic and retarded patients: Areview of punishment and itsalternatives. Psychological Bulletin,85, 1352-1375.

Harry, B., Allen, N., & McLaughlin, M.(1995). Communication versuscompliance: African-Americanparents' involvement in specialeducation. Exceptional Children, 61,364-377.

Harry, B., Grenot-Scheyer, M., Smith-Lewis, M., Park, H-S., Xin, F, &Schwartz, I. (1995). Developingculturally inclusive services forindividuals with severe disabilities.Journal of The Association forPersons With Severe Handicaps, 20,99-109.

REFERENCES

Hays, C. D. (1997). A new perspective. ThePositive Behavior Support Newsletter,1(3), 2-3.

Hersen, M., & Barlow, D. H. (1976). Singlecase experimental designs. New York:

Pergamon.

Hess, P. M., & Mullen, E. J. (Eds). (1995).Practitioner-research partnership:Building knowledge from, in, and forpractice. Annapolis, MD: NASW Press.

Horner, R. (1980). The effects of anenvironmental "enrichment" programon the behavior of institutionalizedprofoundly retarded children.Journal of Applied Behavior Analysis,13, 473-491.

*Horner, R. H., & Budd, C. M. (1985).Acquisition of manual sign use:Collateral reduction of maladaptivebehavior, and factors limitinggeneralization. Education and Trainingof the Mentally Retarded, 20, 39-47.

Horner, R. H., Close, D. W., Fredericks,H. D. B., O'Neill, R. E., Albin, R. W.,

Sprague, J.R., Kennedy, C. H.,Flannery, K. B., & Heathfield, L. T.(1996). Supported living for peoplewith profound disabilities and severeproblem behaviors. In D. H. Lehr & EBrown (Eds.), People with disabilitieswho challenge the system (pp. 209-240). Baltimore: Paul H. Brookes.

*Horner, R. H., & Day, H. M. (1991). Theeffects of response efficiency onfunctionally equivalent competingbehaviors. Journal of Applied BehaviorAnalysis, 24, 719-732.

*Horner, R. H., Day, H. M., & Day, J. R.

(1997). Using neutralizing routines toreduce problem behaviors. Journal ofApplied Behavior Analysis, 30, 601-614.

*Horner, R. H., Day, H. M., Sprague,J. R., O'Brien, M., & Heathfield, L. T.(1991). Interspersed requests: Anonaversive procedure forreducing aggression and self-injury during instruction. Journalof Applied Behavior Analysis, 24,

265-278.

Horner, R. H., Diemer, S., & Brazeau, K.(1992). Educational support forstudents with severe problembehaviors in Oregon: A descriptiveanalysis from the 1987-1988school year. Journal of TheAssociation for Persons With SevereHandicaps, 17, 154-169.

Horner, R. H., Dunlap, G., & Koegel,R. L. (Eds.). (1988). Generalizationand maintenance in appliedsettings. Baltimore: Paul H.

Brookes.

Homer, R. H., Dunlap, G., Koegel, R. L.,Carr, E.G., Sailor, W., Anderson, J.,Albin, R. W, & O'Neill, R. E. (1990).Toward a technology of`nonaversive" behavioral support.Journal of The Association forPersons With Severe Handicaps, 15,125-132.

*Horner, R. H., Sprague, J. R., O'Brien,M., & Heathfield, L. T. (1990). Therole of response efficiency in thereduction of problem behaviorsthrough functional equivalencetraining: A case study. Journal ofThe Association for Persons WithSevere Handicaps, 15, 91-97.

10695

REFERENCES

Horner, R. H., Vaughn, B. J., Day, H. M., &Ard, W. R. (1996). The relationshipbetween setting events and problembehavior: Expanding our understand-ing of behavioral support. In L. K.Koegel, R. L. Koegel, & G. Dunlap(Eds.), Positive behavioral support:Including people with difficult behaviorin the community (pp. 381-402).Baltimore: Paul H. Brookes.

Hoshmand, L. T., & Polkinghorne, D. E.(1992). Redefining the science-practice relationship and professionaltraining. American Psychologist, 47,55-66.

Huberman, M. (1990). Linkage betweenresearchers and practitioners: Aqualitative study. American Educa-tional Research Journal, 27, 363-391.

Hughes, C., Hwang, B., Kim, J. H.,Eisenman, L. T., & Killian, D. J. (1995).Quality of life in applied research: Areview and analysis of empiricalmeasures. American Journal onMental Retardation, 99, 623-641.

*Hunt, P., Alwell, M., Goetz, L., & Sailor, W.(1990). Generalized effects of conver-sation skill training. Journal of TheAssociation for Persons With SevereHandicaps, 15, 250-260.

Individuals With Disabilities EducationAct., Pub. L. No. 105-117, 20 U.S.C. §1401 et seq. (1997).

Iwata, B. A., Dorsey, M. F, Slifer, K. J.,

Bauman, K. E., & Richman, G. S.(1982). Toward a functional analysisof self-injury. Analysis and Interven-tion in Developmental Disabilities, 2,3-20.

10796

*Iwata, B. A., Pace, G.M., Kalsher, M. J.,

Cowdery, G. E., & Cataldo, M. F. (1990).Experimental analysis and extinctionof self-injurious escape behavior.Journal of Applied Behavior Analysis,23, 11-27.

*Iwata, B. A., Pace, G.M., Willis, K. D.,

Gamache, T. B., & Hyman, S. L. (1986).Operant studies of self-injurious handbiting in the Rett's syndrome.American Journal of Medical Genetics(Suppl. 1), 157-166.

*Jansma, P., & Combs, C. S. (1987). Theeffects of fitness training and rein-forcement on maladaptive behaviorsof institutionalized adults, classifiedas mentally retarded/emotionallydisturbed. Education and Training inMental Retardation, 22, 268-279.

*Jayne, D., Schloss, P. J., Alper, S., &Menscher, S. (1994). Reducingdisruptive behaviors by trainingstudents to request assistance.Behavior Modification, 18, 320-338.

*Jones, D. B., & Van Houten, R. (1985). Theuse of daily quizzes and publicposting to decrease disruptivebehavior of secondary schoolstudents. Education and Treatment ofChildren, 8, 91-106.

Jones, S. (1997, May). The FriendlyConnection works at E. C. Glass.Positive Behavior Support Times,1( 1), 1.

*Karsh, K. G., Repp, A. C., Dahlquist, C. M.,& Munk, D. (1995). In vivo functionalassessment and multi-elementinterventions for problem behaviorsof students with disabilities inclassroom settings. Journal ofBehavioral Education, 5, 189-210.

REFERENCES

Kaufman, M., Schiller, E., Birman, B., &Coutinho, M. (1993). A federalperspective on improving practices,programs, and policies in specialeducation. Evaluation and ProgramPlanning, 16, 263-269.

Kazdin, A. E. (1980). Behavior modifica-tion in applied settings (Rev. ed.).Homewood, IL: Dorsey Press.

*Kehle, T. J., Clark, E., Jenson, W. R., &

Wampold, B. E. (1986). Effectivenessof self-observation with behaviordisordered elementary schoolchildren. School Psychology Review, 15,

289-295.

*Kemp, D. C., & Carr, E.G. (1995).Reduction of severe problem behaviorin community employment using anhypothesis-driven multicomponentintervention approach. Journal of TheAssociation for Persons With SevereHandicaps, 20, 229-247.

*Kennedy, C. H. (1994). Manipulatingantecedent conditions to alter thestimulus control of problem behavior.Journal of Applied Behavior Analysis,27,161-170.

*Kennedy, C. H., & Haring, T. G. (1993).Combining reward and escape DRO toreduce the problem behavior ofstudents with severe disabilities.Journal of The Association for PersonsWith Severe Handicaps, 18, 85-92.

*Kennedy, C. H., & Itkonen, T. (1993).Effects of setting events on theproblem behavior of students withsevere disabilities. Journal of Applied

Behavior Analysis, 26, 321-327.

*Kennedy, C. H., & Souza, G. (1995).Functional analysis and treatment ofeye poking. Journal of AppliedBehavior Analysis, 28, 27-37.

Kincaid, D. K. (1992, March). Positivebehavior support training.Morgantown, WV: University Affili-ated Center for DevelopmentalDisabilities (Shawnee Hills Commu-nity Mental/Health RetardationCenter).

Kincaid, D. (1996). Person-centeredplanning. In L. K. Koegel, R. L. Koegel,

& G. Dunlap (Eds.), Positive behavioralsupport: Including people with difficultbehavior in the community (pp. 439-466). Baltimore: Paul H. Brookes.

*Knapczyk, D. R. (1988). Reducingaggressive behaviors in special andregular class settings by trainingalternative social responses.Behavioral Disorders, 14, 27-39.

*Koegel, R. L., Dyer, K., & Bell, L. K.

(1987). The influence of child-preferred activities on autisticchildren's social behavior. Journal ofApplied Behavior Analysis, 20, 243-

252.

Koegel, L. K., Koegel, R. L., & Dunlap, G.

(Eds.). (1996). Positive behavioralsupport: Including people with difficultbehavior in the community. Baltimore:Paul H. Brookes.

*Koegel, L. K., Koegel, R. L., Hurley, C., &

Frea, W. D. (1992). Improving socialskills and disruptive behavior inchildren with autism through self-management. Journal of AppliedBehavior Analysis, 25, 341-353.

*Koegel, R. L., Koegel, L. K., & Suratt, A.

(1992). Language intervention anddisruptive behavior in preschoolchildren with autism. Journal ofAutism and Developmental Disorders,22,141-153.

108 97

REFERENCES

*Konarski, E. A., & Johnson, M. R. (1989).The use of brief restraint plusreinforcement to treat self-injuriousbehavior. Behavioral ResidentialTreatment, 4, 45-52.

*Lalli, J. S., Browder, D. M., Mace, F. C., &Brown, D. K. (1993). Teacher use ofdescriptive analysis data to implementinterventions to decrease students'problem behaviors. Journal of AppliedBehavior Analysis, 26, 227-238.

*Lalli, J. S., & Casey, S. D. (1996). Treat-ment of multiply controlled problembehavior. Journal of Applied BehaviorAnalysis, 29, 391-395.

*Lalli, J. S., Casey, S., & Kates, K. (1995).Reducing escape behavior andincreasing task completion withfunctional communication training,extinction, and response chaining.Journal of Applied Behavior Analysis,28, 261-268.

*Lam, A. L., Cole, C. L., Shapiro, E. S., &

Bambara, L. M. (1994). Relative effectsof self-monitoring on-task behavior,academic accuracy, and disruptivebehavior in students with behaviordisorders. School Psychology Review,23, 44-58.

Lancioni, G. E., & Hoogeveen, E R. (1990).Non-aversive and mildly aversiveprocedures for reducing problembehaviours in people with develop,mental disorders. Mental HandicapsResearch, 3, 137-160.

Landis, J. R., & Koch, G. G. (1977). Themeasurement of observer agreementfor categorical data. Biometrics, 33,159-174.

10998

*Larson, J. L., & Miltenberger, R. G.

(1992). The influence of antecedentexercise on problem behaviors inpersons with mental retardation: Afailure to replicate. Journal of TheAssociation for Persons With SevereHandicaps, 17, 40-46.

Lather, P. (1986). Research as praxis.Harvard Educational Review, 56, 257-277.

LaVigna, G. W., & Donnellan, A. M. (1986).Alternatives to punishment: Solvingbehavior problems with non-aversivestrategies. New York: Irvington.

Lee, V. L. (1988). Beyond behaviorism.

Hillsdale, NJ: Lawrence Erlbaum.

Lennox, D. B., & Miltenberger, R. G.

(1989). Conducting a functionalassessment of problem behavior inapplied settings. Journal of TheAssociation for Persons With SevereHandicaps, 14, 304-311.

Lennox, D. B., Miltenberger, R. G.,Spengler, P., & Erfanian, N. (1988).Decelerative treatment practices withpersons who have mental retardation:A review of five years of the literature.American Journal on Mental Retarda-tion, 92, 492-501.

*Lerman, D. C., Iwata, B. A., Smith, R. G.,& Vollmer, T. R. (1994). Restraintfading and the development ofalternative behavior in the treatmentof self-restraint and self-injury.Journal of Intellectual DisabilityResearch, 38, 135-148.

Littrell, P. C., Billingsley, B. S., & Cross,L. H. (1994). The effects of principalsupport on special and generaleducators' stress, job satisfaction,school commitment, health, andintent to stay in teaching. Remedialand Special Education, 15, 297-310.

REFERENCES

Lloyd, J. W., Weintraub, E J., & Safer, N. D.

(1997).A bridge between research andpractice: Building consensus. Excep-tional Children, 63, 535-538.

Lovaas, 0. I., Freitag, G., Gold, V. J., &

Kassorla, I. C. (1965). Experimentalstudies in childhood schizophrenia:Analysis of self-destructive behavior.Journal of Experimental ChildPsychology, 2, 67-84.

Lowry, M.A., & Sovner, R. (1992). Severebehavior problems associated withrapid cycling bipolar disorder in twoadults with profound mental retarda-tion. Journal of Intellectual DisabilityResearch, 36, 269-281.

Lucyshyn, J. M., & Albin, R. W. (1993).Comprehensive support to families ofchildren with disabilities and behaviorproblems: Keeping it "friendly" InG. H. S. Singer & L. E. Powers (Eds.),Families, disability, and empowerment(pp. 365-407). Baltimore: Paul H.Brookes.

*Lucyshyn, J. M., Albin, R. W, & Nixon,

C. D. (1997). Embedding comprehen-sive behavioral support in familyecology: An experimental, single caseanalysis. Journal of Consulting andClinical Psychology, 65, 241-251.

Lucyshyn, J. M., Olson, D., & Homer, R. H.(1995). Building an ecology ofsupport: A case study of one womanwith severe problem behaviors livingin the community. Journal of TheAssociation for Persons With SevereHandicaps, 20, 16-30.

Luiselli, J. K., & Cameron, M. J. (Eds.).(1998). Antecedent control proceduresfor the behavioral support of personswith developmental disabilities.Baltimore: Paul H. Brookes.

Lutzker, J. R., & Campbell, R. (1994).Ecobehavioral family interventions indevelopmental disabilities. PacificGrove, CA: Brooks/Cole.

Mace, F. C. (1994). The significance andfuture of functional analysis method-ologies.Journal of Applied BehaviorAnalysis, 27, 393-400.

*Mace, E C., & Belfiore, P. (1990). Behav-ioral momentum in the treatment ofescape-motivated stereotypy. Journalof Applied Behavior Analysis, 23, 507-

514.

*Mace F C., Browder, D. M., & Lin, Y.

(1987). Analysis of demand condi-tions associated with stereotypy.Journal of Behavior Therapy andExperimental Psychiatry, 18, 25-31.

*Mace, F C., Browder, D. M., & Martin,D. K. (1988). Reduction of stereotypyvia instruction of alternative leisurebehavior. School Psychology Review,

17, 156-165.

Mace, F. C., Hock, M. L., Lalli, J. S., West,

B. J., Belfiore, P., Pinter, E., & Brown, D.

K. (1988). Behavioral momentum inthe treatment of noncompliance.Journal of Applied Behavior Analysis,21,123-141.

*Maguire, K. B., Lange, B., Scherling, M.,& Grow, R. (1996). The use of re-hearsal and positive reinforcement inthe dental treatment of uncooperativepatients with mental retardation.Journal of Developmental and PhysicalDisabilities, 8, 167-177.

1 I_ 099

REFERENCES

Malette, P., Mirenda, P., Kandborg, T.,Jones, P., Bunz, T., & Rogow, S. (1992).Application of a lifestyle developmentprocess for persons with severeintellectual disabilities: A case studyreport. Journal of The Association forPersons With Severe Handicaps, 17,179-191.

Malouf, D. B., & Schiller, E. P. (1995).

Practice and research in specialeducation. Exceptional Children, 61,414-424.

*Marcus, B.A., & Vollmer, T. R. (1996).Combining noncontingent reinforce-ment and differential reinforcementschedules as treatment for aberrantbehavior. Journal of Applied BehaviorAnalysis, 29, 43-51.

Martin, P. L., & Foxx, R. M. (1973). Victimcontrol of the aggression of aninstitutionalized retardate. Journal ofBehavior Therapy and ExperimentalPsychiatry, 4, 161-165.

*Mason, S. A., & Iwata, B. A. (1990).

Artifactual effects of sensory-integrative therapy on self-injuriousbehavior. Journal of Applied BehaviorAnalysis, 23, 361-370.

*Mason, S. A., McGee, G. G., Farmer-Dougan, V., & Risley, T. R. (1989).Apractical strategy for ongoingreinforcer assessment. Journal ofApplied Behavior Analysis, 22, 171-179.

Matson, J. L., & DiLorenzo, T. M. (1984).Punishment and its alternatives: A newperspective for behavior modification.New York: Springer.

111100

*Matson, J. L., & Keyes, J. (1988). Contin-gent reinforcement and contingentrestraint to treat severe aggressionand self-injury in mentally retardedadults. Journal of the Multihandi-capped Person, 1, 141-153.

Matson, J. L., & Taras, M. E. (1989). A 20-year review of punishment andalternative methods to treat problembehaviors in developmentally delayedpersons. Research in DevelopmentalDisabilities, 10, 85-104.

McDowell, J. J. (1982). The importance ofHerrnstein's mathematical statementof the law of effect for behaviortherapy. American Psychologist, 37,771-779.

Metlen, G., Majure, A., & Stroll-Reisler, C.(1996, July). A person-centeredapproach to supporting people withsevere reputations. TASH (TheAssociation for Persons With SevereHandicaps) Newsletter, 22, 7-9.

Metz, C. D. (1992a). Life quilters: Crafty

cooperation. (Available from the LifeQuitters Project, University AffiliatedCenter for Developmental Disabilities,918 Chestnut Ridge Road, Suite 2,Morgantown, WV 26502)

Metz, C. D. (1992b). Life quilters: Othello

unlimited. (Available from the LifeQuitters Project, University AffiliatedCenter for Developmental Disabilities,918 Chestnut Ridge Road, Suite 2,Morgantown, WV 26502)

Metz, C. D. (1992c). Life quilters: Recycling

independence project. (Available fromthe Life Quitters Project, UniversityAffiliated Center for DevelopmentalDisabilities, 918 Chestnut Ridge Road,Suite 2, Morgantown, WV 26502)

REFERENCES

Meyer, L. H., & Evans, I. M. (1989).Nonaversive intervention for behaviorproblems. Baltimore: Paul H. Brookes.

Meyer, L. H., & Evans, I. M. (1993).Science and practice in behavioralintervention: Meaningful outcomes,research validity, and usable knowl-edge. Journal of The Association forPersons With Severe Handicaps, 18,224-234.

Meyer, L. H., Peck, C.A., & Brown, L.(Eds.). (1991). Critical issues in thelives of people with severe disabilities.Baltimore: Paul H. Brookes.

Michael, J. (1982). Distinguishingbetween discriminative and motiva-tional functions of stimuli. Journal ofthe Experimental Analysis of Behavior,37, 149-155.

*Montgomery, R. W. (1993). The collateraleffect of compliance training onaggression. Behavioral ResidentialTreatment, 8, 9-20.

Morrissey, P. A. (1997). Response tobridging the research-to-practice gap.Exceptional Children, 63, 531-532.

Mulick, J. A., & Kedesdy, J. H. (1988). Self-injurious behavior, its treatment, andnormalization. Mental Retardation, 26,223-229.

Mulick, J. A., & Linscheid, T. R. (1988).[Review of the book Alternatives topunishment: Solving behavior prob-lems with non-aversive strategies].Research in Developmental Disabilities,9, 317-321.

National Institutes of Health. (1991).Treatment of destructive behaviors inpersons with developmental disabilities(NIH Publication No.91-2410).Washington, DC: Author.

*Neufeld, A., & Fantuzzo, J. W. (1987).Treatment of severe self-injuriousbehavior by the mentally retardedusing the bubble helmet and differen-tial reinforcement procedures. Journalof Behavior Therapy and ExperimentalPsychiatry, 18, 127-136.

Nickels, C. (1996). A gift from AlexTheart of belonging: Strategies foracademic and social inclusion. In L. K.Koegel, R. L. Koegel, & G. Dunlap(Eds.), Positive behavioral support:Including people with difficult behaviorin the community (pp. 123-144).Baltimore: Paul H. Brookes.

Nietupski, J., Hamre-Nietupski, S., Curtin,S., & Shrikanth, K. (1997).A review ofcurricular research in severe disabili-ties from 1976 to 1995 in six selectedjournals. The Journal of SpecialEducation, 31, 36-55.

Northup, J., Wacker, D. P., Berg, W. K., Kelly,

L., Sasso, G., & DeRaad, A. (1994).The treatment of severe behaviorproblems in school settings using atechnical assistance model. Journal ofApplied Behavior Analysis, 27, 33-47.

O'Brien, S., & Repp, A. C. (1990).Reinforcement-based reductiveprocedures: A review of 20 years oftheir use with persons with severe orprofound retardation. Journal of TheAssociation for Persons With SevereHandicaps, 15, 148-159.

O'Neill, R. E., Horner, R. H., Albin, R. W.,

Storey, K., & Sprague, J. R. (1997a).Functional assessment and programdevelopment for problem behavior.Pacific Grove, CA: Brooks/Cole.

112101

REFERENCES

O'Neill, R. E., Horner, R. H., Albin, R. W.,Storey, K., & Sprague, J. R. (1997b).Functional analysis of problembehavior: A practical assessment guide.Sycamore, IL: Sycamore.

*Pace, G.M., Iwata, B. A., Cowdery, G. E.,Andree, P. J., & McIntyre, T. (1993).

Stimulus (instructional) fadingduring extinction of self-injuriousescape behavior. Journal of AppliedBehavior Analysis, 26, 205-212.

*Pace, G.M., Iwata, B. A., Edwards, G. L., &McCosh, K. C. (1986). Stimulus fadingand transfer in the treatment of self-restraint and self-injurious behavior.Journal of Applied Behavior Analysis,19,381-389 .

Paisey, T. J. H., Whitney, R. B., & Hislop,P. M. (1990). Client characteristics andtreatment selection: Legitimateinfluences and misleading influences.In A. C. Repp & N. N. Singh (Eds.),Perspectives on the use of nonaversiveand aversive interventions for personswith developmental disabilities (pp.175-197). Sycamore, IL: Sycamore.

*Paisey, T. J., Whitney, R. B., & Moore, J.

(1989). Person-treatment inter-actions across nonaversive response-deceleration procedures for self-injury: A case study of effects and sideeffects. Behavioral ResidentialTreatment, 4, 69-88.

Patterson, G. R. (1982). Coercive familyprocess. Eugene, OR: Castalia.

Pearman, E. L., Huang, A. M., & Mellblom,C. I. (1997). The inclusion of allstudents: Concerns and incentives ofeducators. Education and Training inMental Retardation and Developmen-tal Disabilities, 32, 11-20.

Poling, A., & Ryan, C. (1982). Differentialreinforcement of other behaviorschedules: Therapeutic applications.Behavior Modifications, 6,3-21.

Prosser, G. (1988). Vibratory reinforcementin the field of mental handiCap: A review.Mental Handicap Research, 1, 152-166.

Ratey, J. J., Sovner, R., Parks, A., & Rogentine,K. (1991). Busiprone treatment ofaggression and anxiety in mentallyretarded patients: A multiple-baseline,placebo lead-in study. Journal of ClinicalPsychiatry, 52, 159-162.

*Realon, R. E., Favell, J. E., & Cacace, S.

(1995). An economical, humane, andeffective method for short-termsuppression of hand mouthing. Behav-ioral Interventions, 10, 141-147.

Reichle, J. (1997). Communication interven-tion with persons who have severedisabilities. The Journal of SpecialEducation, 31, 110-134.

Reichle, J., & Wacker, D. P. (1993). Communi-cative alternatives to challengingbehavior. Baltimore: Paul H. Brookes.

Reiss, S., & Aman, M. G. (Eds.). (1998).

Psychotropic medications and develop-mental disabilities: The internationalconsensus handbook. Columbus, OH:The Ohio State University Press.

Reiss, S., & Havercamp, S. M. (1997).Sensitivity theory and mental retarda-tion: Why functional analysis is notenough. American Journal on MentalRetardation, 101, 553-566.

Reiss, S., & Rojahn, J. (1993). Joint occur-rence of depression and aggression inchildren and adults with mentalretardation. American Journal on MentalRetardation, 37,287-294.

102 113

REFERENCES

*Repp, A. C., & Karsh, K. G. (1994).Hypothesis-based interventions fortantrum behaviors of persons withdevelopmental disabilities in schoolsettings. Journal of Applied BehaviorAnalysis, 27, 21-31.

Repp, A. C., & Singh, N. N. (Eds.). (1990).Perspectives on the use of nonaversiveand aversive interventions for personswith developmental disabilities.Sycamore, IL: Sycamore.

Rincover, A., Cook, R., Peoples, A., &Packard, D. (1979). Sensory extinctionand sensory reinforcement principlesfor programming multiple adaptivebehavior change. Journal of AppliedBehavior Analysis, 12, 221-233.

Risley, T. (1996). Get a life! In L. K. Koegel,R. L. Koegel, & G. Dunlap (Eds.),Positive behavioral support: Includingpeople with difficult behavior in thecommunity (pp. 425-437). Baltimore:Paul H. Brookes.

*Roberts, M. L., Mace, E C., & Daggett, J.

A. (1995). Preliminary comparisons oftwo negative reinforcement schedulesto reduce self-injury. Journal ofApplied Behavior Analysis, 28,

579-580.

Rolider, A., & Van Houten, R. (1990). Therole of reinforcement in reducinginappropriate behavior: Some mythsand misconceptions. In A. C. Repp &N. N. Singh (Eds.), Perspectives on theuse of nonaversive and aversiveinterventions for persons with develop-mental disabilities (pp. 119-127).Sycamore, IL: Sycamore.

Ruef, M. B. (1997). The perspectives of sixstakeholder groups on the challengingbehavior of individuals with mentalretardation and/or autism. Unpub-lished doctoral dissertation, Univer-sity of Kansas, Lawrence.

S. Rep. No. 102-357, 102d Cong., 2nd Sess.3712 et seq. (1992).

Sailor, W. (1996). New structures andsystems change for comprehensivepositive behavioral support. In L. K.Koegel, R. L. Koegel, & G. Dunlap(Eds.), Positive behavioral support:Including people with difficult behaviorin the community (pp. 163-206).Baltimore: Paul H. Brookes.

*Sanders, M. R., & Plant, K. (1989).Programming for generalization tohigh and low risk parenting situationsin families with oppositional develop-mentally disabled preschoolers.Behavior Modification, 13, 283-305.

Sands, D. J., Kozleski, E. B., & Goodwin, L.D. (1991). Whose needs are wemeeting? Results of a consumersatisfaction survey of persons withdevelopmental disabilities in Colo-rado. Research in DevelopmentalDisabilities, 12, 297-314.

*Santarcangelo, S., Dyer, K., & Luce, S. C.

(1987). Generalized reduction ofdisruptive behavior in unsupervisedsettings through specific toy training.Journal of The Association for PersonsWith Severe Handicaps, 12, 38-44.

Santelli, B., Turnbull, A. P., Marquis, J. G.,

& Lerner, E. P. (1993). Parent to Parentprograms: Ongoing support forparents of young adults with specialneeds. Journal of Vocational Rehabili-tation, 3, 25-37.

`11 4 103

REFERENCES

Santelli, B., Turnbull, A. P., Marquis, J. G., &Lerner, E. P. (1995). Parent to Parentprograms: A unique form of mutualsupport. Infants and Young Children, 8,48-57.

*Sasso, G. M., Reimers, T. M., Cooper, L. J. ,

Wacker, D., Berg, W., Steege, M., Kelly, L.,

& Allaire, A. (1992). Use of descriptiveand experimental analyses to identifythe functional properties of aberrantbehavior in school settings. Journal ofApplied Behavior Analysis, 25, 809-821.

*Saunders, R. R., Saunders, M. D., Brewer,A., & Roach, T. (1996). The reduction ofself-injury in two adolescents withprofound retardation by the establish-ment of a supported routine. BehavioralInterventions, 11, 59-86.

Schaal, D. W., & Hackenberg, T. (1994).

Toward a functional analysis of drugtreatment for behavior problems ofpeople with developmental disabilities.American Journal on Mental Retarda-tion, 99, 123-140.

Schalock, R. L. (Ed). (1996). Quality of life:Perspectives and issues. Washington, DC:American Association on MentalRetardation.

Schalock, R. L. (Ed). (1990). Quality of life:Vol. 1. Conceptualization and measure-ment. Washington, DC: AmericanAssociation on Mental Retardation.

Scheerenberger, R. (1990). Public residentialservices for the mentally retarded: FY1988-1990. Fairfax, VA: NationalAssociation of Superintendents ofPublic Residential Facilities for theMentally Retarded.

115104

*Schloss, P. J., Smith, M., Santora, C., &Bryant, R. (1989). A respondentconditioning approach to reducinganger responses of a dually diagnosedman with mild mental retardation.Behavior Therapy, 20, 459-464.

Schroeder, S. R., & Tessel, R. (1994).Dopaminergic and serotonergicmechanisms in self-injury andaggression. In T. Thompson & D. Gray(Eds.), Treatment of destructivebehavior in developmental disabilities(pp. 198-212). Newbury Park, CA:Sage.

Scotti, J. R., Evans, I. M., Meyer, L. H., &Walker, P. (1991). A meta-analysis ofintervention research with problembehavior: Treatment validity andstandards of practice. AmericanJournal on Mental Retardation, 96,233-256.

Scotti, J. R., Ujcich, K. J., Weigle, K. L.,

Holland, C. M., & Kirk, K. S. (1996).Interventions with challengingbehavior of persons with developmen-tal disabilities: A review of currentresearch practices. Journal of TheAssociation for Persons With SevereHandicaps, 21, 123-134.

Seegmiller, J. E. (1972). Lesch-Nyhansyndrome and the X-linked uricacidurias. Hospital Practice, 7, 79-90.

*Shulda, S., & Albin, R. W. (1996). Effectsof extinction alone and extinctionplus functional communicationtraining on covariation of problembehaviors. Journal of Applied BehaviorAnalysis, 29, 565-568.

*Sigafoos, J., & Kerr, M. (1994). Provisionof leisure activities for the reductionof challenging behavior. BehavioralInterventions, 9, 43-53.

REFERENCES

*Sigafoos, J., & Meikle, B. (1996). Func-tional communication training for thetreatment of multiply determinedchallenging behavior in two boys withautism. Behavior Modification, 20,60-84.

*Sigafoos, J., Penned, D., & Versluis, J.(1996). Naturalistic assessmentleading to effective treatment of self-injury in a young boy with multipledisabilities. Education and Treatmentof Children, 19, 101-123.

Singer, G. H. S., Singer, J., & Horner, R. H.(1987). Using pretask requests toincrease the probability of compliancefor students with severe disabilities.Journal of The Association for PersonsWith Severe Handicaps, 12, 287-291.

Skinner, B.F. (1974). About behaviorism.New York: Vintage.

*Slifer, K. J., Ivancic, M. T., Parrish, J. M.,Page, T. J., & Burgio, L. D. (1986).Assessment and treatment of multiplebehavior problems exhibited by aprofoundly retarded adolescent.Journal of Behavior Therapy andExperimental Psychiatry, 17, 203-213.

Smith, M.D. (1985). Managing theaggressive and self-injurious behaviorof adults disabled by autism. Journalof The Association for Persons WithSevere Handicaps, 10, 228-232.

Smith, M.D. (1986). Use of similarsensory stimuli in the community-based treatment of self-stimulatorybehavior in an adult disabled byautism. Journal of Behavior Therapyand Experimental Psychiatry, 17,121-125.

Smith, M.D. (1990). Autism and life in thecommunity. Baltimore: Paul H.Brookes.

*Smith, M.D., & Coleman, D. (1986).Managing the behavior of adults withautism in the job setting. Journal ofAutism and Developmental Disorders,16, 145-154.

*Smith, R. G., Iwata, B. A., Vollmer, T. R., &Zarcone, J. R. (1993). Experimentalanalysis and treatment of multiplycontrolled self-injury. Journalof Applied Behavior Analysis, 26,

183-196.

Snell, M. E. (Ed.). (1993). Instruction ofstudents with disabilities (4th ed.).New York: Merrill.

Sovner, R. (1989). The use of valproate inthe treatment of mentally retardedpersons with typical and atypicalbipolar disorders. Journal of ClinicalPsychiatry, 50 (Suppl. 3), 40-43.

*Sprague, J., Holland, K., & Thomas, K.(1997). The effect of noncontingentsensory reinforcement, contingentsensory reinforcement, and responseblocking on stereotypical and self-injurious behavior. Research inDevelopmental Disabilities, 18, 61-77.

*Sprague, J. R., & Horner, R. H. (1992).Covariation within functionalresponse classes: Implications fortreatment of severe problem behavior.Journal of Applied Behavior Analysis,25, 735-745.

116105

REFERENCES

*Steed, S. E., Bigelow, K. M., Huynen, K. B.,& Lutzker, J. R. (1995). The effects ofplanned activities training, lowdemand schedule, and reinforcementsampling on adults with developmen-tal disabilities who exhibit challengingbehaviors. Journal of Developmentaland Physical Disabilities, 7,303-316.

*Steege, M. W., Wacker, D. P., Berg, W. K.,

Cigrand, K. K., & Cooper, L. J. (1989).The use of behavioral assessment toprescribe and evaluate treatments forseverely handicapped children.Journal of Applied Behavior Analysis,22,23-33.

*Steege, M. W., Wacker, D. P., Cigrand,K. C., Berg, W. K., Novak, C. G.,

Reimers, T. M., Sasso, G. M., &DeRaad, A. (1990). Use of negativereinforcement in the treatment of self-injurious behavior. Journal of AppliedBehavior Analysis, 23, 459-467.

Stokes, T. F., & Baer, D. M. (1977). Animplicit technology of generalization.Journal of Applied Behavior Analysis,10,349-367 .

Sturmey, P. (1994). Assessing the func-tions of aberrant behaviors: A reviewof psychometric instruments. Journalof Autism and Developmental Disor-ders, 24, 293-304.

*Taylor, C. R., & Chamove, A. S. (1986).Vibratory and visual stimulationreduces self-injury. Australia and NewZealand Journal of DevelopmentalDisabilities, 12, 243-248.

Taylor, J. C., & Carr, E.G. (1992). Severeproblem behaviors related to socialinteraction. I: Attention seeking andsocial avoidance. Behavior Modifica-tion, 16, 305-335.

106 117

Taylor-Greene, S., Brown, D., Nelson, L.,Longton, J., Gassman, T., Cohen, J.,Swartz, J., Horner, R. H., Sugai, G., &Hall, S. (1997). School-wide behav-ioral support: Starting the year offright. Journal of Behavioral Education;Z 99-112.

Thompson, T., Hackenberg, T., & Schaal,D. (1991). Pharmacological treatmentsfor behavior problems in developmen-tal disabilities. In U.S. Department ofHealth and Human Services, Treat-ment of destructive behaviors inpersons with developmental disabilities(NIH Publication No. 91-2410, pp.343-510). Washington, DC: NationalInstitutes of Health.

Tifft, S. (1996, March). True life testimony.Paper presented at the PositiveSupport Planning Workshop, Eugene,OR.

Touchette, P. E., MacDonald, R. F., &Langer, S. N. (1985).A scatter plot foridentifying stimulus control ofproblem behavior. Journal of AppliedBehavior Analysis, 18, 343-351.

Turnbull, A. P. (1988). The challenge ofproviding comprehensive support tofamilies. Education and Training inMental Retardation, 23, 261-272.

Turnbull, A. P., Friesen, B. J., & Ramirez, C.(1998). Participatory Action Researchas a model of conducting familyresearch. Journal of The Association forPersons With Severe Handicaps, 23,178-188.

Turnbull, A. P., & Ruef, M. (1996). Familyperspectives on problem behavior.Mental Retardation, 34, 280-293.

REFERENCES

Turnbull, A. P., & Ruef, M. (1997). Familyperspectives on inclusive lifestyle issuesfor people with problem behavior.Exceptional Children, 63, 211-227.

Turnbull, A. P., & Turnbull, H. R. (1996).Group action planning as a strategy forproviding comprehensive familysupport. In L. K. Koegel, R. L. Koegel, &G. Dunlap (Eds.), Positive behaviorsupport: Including people with difficultbehavior in the community (pp. 99-114).Baltimore: Paul H. Brookes.

*Umbreit, J., & Blair, K. S. (1996). Theeffects of preference, choice, andattention on problem behavior at school.Education and Training in MentalRetardation, 31, 151-161.

*Underwood, L. A., Figueroa, R. G., Thyer,B. A., & Nzeocha, A. (1989). Interruptionand DRI in the treatment of self-injurious behavior among mentallyretarded and autistic self-restrainers.Behavior Modification, 13, 471-481.

Vandercook, T., York, J., & Forest, M. (1989).The McGill action planning systems(MAPS): A strategy for building thevision. Journal of The Association forPersons With Severe Handicaps, 14, 205-215.

Vaughn, B., & Horner, R. H. (1997). Identify-ing instructional, tasks that occasionproblem behaviors and assessing theeffects of student versus teacher choiceamong these tasks. Journal of AppliedBehavior Analysis, 30, 299-312.

Virginia Institute of DevelopmentalDisabilities. (1996). Why Jason behavesthat way and what to do about it: Adialogue about problem behaviors andpositive support for family members andfriends [Booklet]. Richmond, VA:Author.

Vollmer, T. R., & Iwata, B.A. (1992).Differential reinforcement as treat-ment for behavior disorders: Proce-dural and functional variations.Research in Developmental Disabilities,13, 393-417.

*Vollmer, T. R., Iwata, B.A., Smith, R. G., &Rodgers, T. A. (1992). Reduction ofmultiple aberrant behaviors andconcurrent development of self-careskills with differential reinforcement.Research in Developmental Disabilities,13, 287-299.

*Vollmer, T. R., Iwata, B.A., Zarcone, J. R.,Smith, R. G., & Mazaleski, J. L. (1993).The role of attention in the treatmentof attention-maintained self-injuriousbehavior: Noncontingent reinforce-ment and differential reinforcement ofother behavior. Journal of AppliedBehavior Analysis, 26, 9-21.

*Vollmer, T. R., Marcus, B. A., & LeBlanc,L. (1994). Treatment of self-injury andhand mouthing following inconclusivefunctional analyses. Journal of AppliedBehavior Analysis, 2Z 331-344.

*Vollmer, T. R., Marcus, B. A., & Ringdahl,J. E. (1995). Noncontingent escape astreatment for self-injurious behaviormaintained by negative reinforcement.Journal of Applied Behavior Analysis,28, 15-26.

Wacker, D. P., Steege, M. W., Northup, J.,Sasso, G., Berg, W., Reimers, T., Cooper,L., Cigrand, K., & Donn, L. (1990). Acomponent analysis of functionalcommunication training across threetopographies of severe behaviorproblems. Journal of Applied BehaviorAnalysis, 23, 417-429.

118 107

REFERENCES

Wheeler, J. J. (1996). The use of interactivefocus groups to aid in the identifica-tion of perceived service and supportdelivery needs of persons withdevelopmental disabilities and theirfamilies. Education and Training inMental Retardation and Developmen-tal Disabilities, 31, 294-303.

White, C., Lakin, K., Bruininks, R., & Li, X.(1991). Persons with mental retarda-tion and related conditions in state-operated residential facilities: Yearending June 30, 1989 with longitudinaltrends from 1950-1989. Minneapolis:University of Minnesota, Center forResidential and Community Services,Institute on Community Integration.

Wiesler, N. A., Hanson, R. H., Chamber-lain, T. P., & Thompson, T. (1985).Functional taxonomy of stereotypicand self-injurious behavior. MentalRetardation, 23, 230-234.

Willemsen-Swinkels, S. H. N., Buitelaar, J.K., Nijhof, G. J., & van Engeland, H.(1995). Failure of naltrexone hydro-chloride to reduce self-injurious andautistic behavior in mentally retardedadults. Archives of General Psychiatry,52, 766-773.

*Winterling, V., Dunlap, G., & O'Neill, R. E.(1987). The influence of task variationon the aberrant behaviors of autisticstudents. Education and Treatment ofChildren, 10, 105-119.

119108

Wolery, M., & Ezell, H. K. (1993). Subjectdescriptions and single subjectresearch. Journal of Learning Disabili-ties, 26, 642-647.

Wolf, M. M. (1978). Social validity: Thecase for subjective measurement, orhow applied behavior analysis isfinding its heart. Journal of AppliedBehavior Analysis, 11, 203-214.

*Zarcone, J. R., Iwata, B.A., Mazaleski, J.L., & Smith, R. G. (1994). Momentumand extinction effects on self-injurious escape behavior andnoncompliance. Journal of AppliedBehavior Analysis, 27, 649-658.

*Zarcone, J. R., Iwata, B.A., Smith, R. G.,Mazaleski, J. L., & Lerman, D. C.(1994). Reemergence and extinctionof self-injurious escape behaviorduring stimulus (instructional)fading. Journal of Applied BehaviorAnalysis, 27, 307-316.

*Zarcone, J. R., Iwata, B. A., Vollmer, T. R.,

Jagtiani, S., Smith, R. G., & Mazaleski,J. L. (1993). Extinction of self-injurious escape behavior with andwithout instructional fading. Journalof Applied Behavior Analysis, 26,353-360.

Zeiler, M. D. (1970). Other behavior:Consequences of reinforcing notresponding. Journal of Psychology, 74,149-155.

U.S. Department of EducationOffice of Educational Research and Improvement (OERI)

National Library of Education (NLE)

Educational Resources Information Center (ERIC)

NOTICE

Reproduction Basis

O

ERIC

This document is covered by a signed "Reproduction Release(Blanket)" form (on file within the ERIC system), encompassing allor classes of documents from its source organization and, therefore,does not require a "Specific Document" Release form.

This document is Federally-funded, or carries its own permission toreproduce, or is otherwise in the public domain and, therefore, maybe reproduced by ERIC without a signed Reproduction Release form(either "Specific Document" or "Blanket").

EFF-089 (3/2000)

reproductions supplied by edrs are the best that can be made · christopher e. smith, phd kaarin...

Documents