measuring response rates in associative skill...

Measuring Response Rates in Associative Skill Development

Carl Binder

Unpublished Draft Doctoral Working Paper Experimental Psychology Department – Harvard University

Cambridge, MA

1979

This unpublished document combines references to research and application seldom cited

in the literature of Precision Teaching or Fluency-based Instruction. Scanned from an original

working draft, it was never published or presented as is, but used in preparation of several

published papers. Its perspective and combination of topics were shaped by the scholarly need to

connect some of our early work in Precision Teaching with basic research in experimental

psychology. Its purpose was to pull together strands of experimental learning research in which

the time dimension –latency, rate of response (frequency), or response duration – was included in

the measurement of learning and performance.

The first chapter covers a diversity of sources under a common term, “associative skills.”

It highlights research and theory from multiple sub-fields that might be relevant to combinations

of motor and verbal skills, including discriminations and combinations, similar to those taught in

Precision Teaching classrooms. It summarizes relevant effects and findings.

The second chapter explores implications of using Skinner’s variable, rate of response, as

a measure of skill proficiency – a term used by Precision Teachers at the time. It draws on

precedents in learning research, and presents some pilot research from our lab classroom.

The third chapter traces the evolution from rate of response in the laboratory to standard

celeration charting and Precision Teaching in the classroom. It introduces the term fluency, and

summarizes our research program in the Behavior Prosthesis Lab within the framework we

developed of “four kinds of ceilings” limiting student performance that emerged as we measured

response rates (i.e., behavior frequencies) in the classroom. We hope that this backgrounder

might be helpful to scholars and practitioners in Precision Teaching and related fields.

NOTE: Please pardon use of the term retarded as in retarded students, or severely

retarded, in chapters II and III, written in an era when that term was accepted practice among

educators and scholars, an advance from previously used terms, idiot, imbecile, and moron.

Contact the author at [email protected]

C. Binder Measuring Response Rates in Associative Skill Development Harvard Department of Psychology 1979 1

€



carlbinder

Typewritten Text

Figure 1

carlbinder

Typewritten Text


carlbinder

Typewritten Text

Figure 2


carlbinder

Typewritten Text

Figure 3

€


Rat,iglule

From lAboralorv to Classroon

After rore than a decade of human operant condilioning research in

the LaboraLory, Lindsley decided to take the rationale and measurement

technol-ogy of an experimental analJrsis of behavior into classroons. His

first paper in Lhe fie}i of educalion (Lindsley, 196l+) began:

ChilCren are nct relardeC. 0n1y their belavior in average envircn-

menls is somelires relarded. In fact, it is nociern sciencets abil-iLy

lc design suilaole environments for these children lhal is relarcied.f t ^\\p. 1.)

laboralory analyses of retarcied behavicr (Barretl & Llndsl-ey, L952), eon-

vinceci Lindsley that the melhodology of behavior analysj-s could be applieci

in classrooms.

A primary ingreciien+- of an experimenlal analysis of behavior is the

dislinction oetween operalions and functions. Operalional definillon des'

cribes a serj-es of evenls, or op€rations, which the experinenler arranges

in lhe hope of affecti-ng lhe behavior under observalion. Through systematic

manipulation of operalionaliy defineci events, in conjunclion wilh behavioral

measurellrcnL, the scientist nay discover functional relalionships between an

individualts behavj-or and environnental events (Skinner, t953a, p, 35). Be-

cause nany educators (anri even behavioral researchers ) have faileci Lo riis-

tinguish between operations and functions, Lindsley (ntt!) expressed rrthe

for:r-componenl operant behavioral equationil in Lwo different forms. The

form coresponding to operational definition is:

EAMAES.

In tenporal sequence, an antecedent event (EA) o"c.,r" before a grovenrenl (M)


on the part of lhe individual. The movenenl is then followed b.y a subse-

quenl event 1nS; *:-tt', a particular arrgneement (A).

In an experinrenlal analysis, the sei"entisL nay verify lhe functions

of these events. Accordingly, the second forn of the eqr:alion is:

S R K C.

Ant,ecedent events may funclion as stimuli (S), controlllng probability of

response. Movements nay function as responsgs (R)roperating on the envir-

onmenl to produce effects. Subsequent events nay function as acceleraiing

or decelerating consequences (C), if Lhe fllanner in which they are arrangeo

to follow movenenLs functi-ons as a ccntinsenc,v (X) for lhe behaver. i'Iith

thi-s enphasis on funcLional analysis, Lindsley proposeci an approach to edu-

cation, baseci on neasurement. Lindsley used the term moverpnt cyc1e, Lo

describe potentia!- resFonses. It nray be coincicienlal thal lhis term also

appears in percept'aa}-nolor learning Ii-leralure (".g., Fitts , 1'951+). There

was a prececient in labora+"ory operanl condiLioning for stutif ing repeaLable

movenenls which coulci be counled by aulornaLic apperatus. Lindsley and his

associ-ates argueci thal such movenent cycl-es, rather than bodi1l' positions

or orientations (",S., rrkeeps hancis in lapt'), constituled the most imporLanl

targels for applied behavior analysis.

UnLike other behavioral psychologisls who have moved from basie research

into application, Lincisley di<i not seek recipes for bel'.avior change. In

faet, this may be the major difference belween hi-s "precisi-on teachingrl

approach and other forns of applied behavior analysls. Emphasizing the in-

dividual organi-sm, as Skinner did (tgSS), Ll-ndsley reasoned that effeeLive

educaLion musl take individual differences lnto account. He argued thal

instead of recipes, educalors need a measureuenl system sensitj-ve to the

effecliveness of individualizeti proceciures (Lindsley, t972). llith such an

approach, teachers would be able to discover which procedures are functional


for which individr:al-s, anci perhaps finci eengrglly effeclive proceciures as

wel1. The method, like Lhal of an experimental analysis of behavior, was

to be strictly inductive.

Measuremgnt as a Cownunication Device

In an analysis of behavior, rleasurerenl functi-ons as a conmunicaLion

device in two ways, First, lt allows the behaver to conmunicate to the

behavior analyst (Barrett, L9?7). Skinner (t948, p. 2S9) argueci that the

t'organisn is always right. t' 0r as Lindsley (nZZ) put it, in the eciucational

context, "The child knows besl. " The poinl is that measurenenl reveals the

funelions for the behaver of events aruangeci by the educator or scientist.

ft allows the scientist or educalor to ciecicie whether or not specific pro-

ceciures are effective, or funclional. In addilion, measurement funclions as

a rnedir:m of conmunlcation among behavior analysts, all lhe more so with

stanriardizeci forms of measurement. The cu,nulative recorci, for exanpJ-e,

provides for exlremely efficienl cornmunj-calion among experimental behavior

analysls. One need only know lhe sLep-size and paper-speed lo be abl-e lo

interpret a cumulative record. Because cf his long experience with cumu-

lalive records, Lindsl"y (L972) deci<ied to design a slanciard behavior charl

for educalors anci other behavior-change agents.

The Slandard Behavior Clart

Although j-t is eurrently in its ninlh revision, the Standard Behavior

Chart has renained essential-Iy the saroe since its invention in !965. Il

a11ows for a standardized display of any countable behavior, thus proviciing

a neans of efficient cornmunication anci a framework for viewing a range of

research questions and behavioral dimensions. The following ls a sulnmary

of its basic riisplay charaeleristics and suffrary qr:antifiers. For a IIDre

thorough discussj-on, see Penrypacker, Koenig, and Lindsley (1972), l{hiLe


anci Haring (tqZS), or White and Liberty (WZq.

Cal-gnslar Base agd Synchronization

Because the Standard Behavior Chart is based on ealendar days, not

sessions, one can observe the effecls of intersession intervals. The Daily

Behavior chart, spanning 1l+0 days, is especially varuable in school app-

licat,icns. (There are also weekly, monlhly, anC yearly versions. ) In

North America, it is convenLional- lo begin the first charl of the year on

the sunciay before labcr Day (wnite and Haring, 1975). Especially within

a given progran, school, or research projecl, such eal-ender synchroniza-

tion is useful for exarnining paltems of behavior across behavers, or within

behavers across behaviors. Oflen it is possible to ciiscover behavioral

eovariation simply cy superimposing two cr more charts on a light-bcx.

Count Per Minule Crdl-nate

?he Daily Behavior Charl reduees all data tc count per nrinule. Lincisley

adcpt-s6 the response rale nieas'Jre which had been so useful in laboratory

operant conditioning (Skinner, 1aio, Iqi3c, 7q56).

Dr:ralions. iatencies. Reccro Floors

It is also possible to record periods of tirne on the Standard Chart.

Sinee rate is the reciprocal of duraLion, iime measurernents may be charieci

as lhe rale corresponding to one response in the specified amount of tirne.

For example, a duration ol t2 seconds woulci appear as a count of 5 per nin-

ule. A duration of 10 minutes woulci appear as .10 per rninule. A latency

of 25O nilliseconcis would appear as 2lro/frnute. The lg|@ or cougUlns

period floor represents the length of the recorciing period. It is inporlanL

to be able to graph this dimensi,on along with counts per ninule. The endur-

ance data in Chapter fI provide one example of how these two measuremenls

t*ig\ be functionally relaled,. It is conventional to chart the record floor

as a srnall dash for each day on which rale, latency, OT ciura$ion ciaLa, are


also charted. The recorci floor enables the charler to make judgmenls abou+-

measuremenl proceciures. A time period may be too short to pernrit observa-

tion of any insLances of comparati-ve1y 1ow-frequency behaviors. Systematic

variation of lhe counling period a1lows calibralion of lhe nieasuremenl pro-

cedure. Comparisons of rates oblained across a range of counting periccis

yielci informalion about appropriate observation or practice ciurat,ions.

When an observati-on yields a count of zero for a given counting period,

the charter puts a queslion-mark just below the reeorci floor. This conven-

ti-on aeeentuaies lhe fact that a longer observation period might have yiekied

a counL grealer than zero.

A zero-counl for 5 rdnules is di-fferent from a zero-count for one hour

in lhal a respcnse which occurs, sBX, 10 tines per hour nfght not occur al

a1I iuring a 5-ninute observation. This is similar to lhe properly of per-

centage correct measuremenl whereby lhe number of opporlunities to respond

det,err:n-ines the size of possibie percenLage increnenls. It is inpnssible lo

achieve a score of 25'1 ccrect when lhere are 10 cpportunilies to respond.

While anci Harine 0975) have calleci the incremenl-size on the percen+- scale

a percen+"aqe recorci f lcor.

l,ogaritlnric Tra$;fornration on ihe Orciinale

ialhereas the calendar base is an equal-interval scale, the ordinale of

the Standarci Behavior Charl is logarithmic or equal-ratio. This character-

istic of the charl has several acivantages.

Wide ranqe. The ordinate spans a range fron 10Cr0 novernents per ninute

to .OOO495 novenents per ninute, which is one response per day. Although

such behaviors as silent reading or nasterful banjc-picking rnay occur aL a

higher rate than 1W0 per ninute, mosl skilIs are within the 5-cycle range.

Thus, virlually any huran response rale nay appear on the sare chart, for


convenience a6 we1] as for comparison.

&curacy Ratios. Precision teachers chart rates of eorrect responses,

errors, prompted responses and other categories of skilled perfornance as

separate points on the sarne day-Iine. Rate neasurement allows independenl

analysis of lhe probabilily of each category of response. This is not poss-

ible with percentage neasurenent.

The logarithnic ordinate of the Standard Chart nakes i* possible to

express accuracy as a ratio, Thus, rate of conect responses nay be a

multiple (".g., x 2.0) or a fracLion (e.g., * 1.5) of the error rate. Such

Iaccuracy ratios quanlifyf.ccuracy, independenl of rale (Pennypacker, el. aL.,

It972), Standard. Chart represents a given accuracy ratio as a vertical dis-

tance belween the poinls for eror rate and rate of conect responses on a

particular ciay. For example, a x 3.0 accuracy ratio (75% corcect) appears

as lhe ciistance between 1.0 errors per ninule anci 3.0 correct responses per

minute, 5.A and 15.0, or any other nultiple ofx 3.0, These distances are

Lhe same on lhe logarithnric ordinate.

ff t,here is a zero count for a category cf response, lhe value of the

record floor is used in place of zero in compuling the accuracy ratio. This

is reasonable since a longer observation period with a counl of no erors,

for exanple, verifies a greater ratio between correct responses and errors

than does a shorler period. The effect of this convention is to provide

a scale of accuracy which goes beyond 1001 correct. For even wlth a count

of zero errors,lt is possible either to inerease the rate of correct re-

sponding or to increase the duratj-on of errorless perforrnance. Either

change would increase the accuracy ratio.

Trends or Drogress lines. The logarilhmic ordinate allows for the ex-

pression of trends as ratios. A rate of behavior may multiply (accelerate) or


divide (deeelerate) to over lime. Trends are expressed as rcel.erations

cn the SlanCari Chart, and a given treni (".g., x1.50) appears with the

sane slope anJnlhere on the chart, irrespective of its starting-point.

fn general, users of the charl apply the procedures of resistanl-

fit explcratory data analysis CFnkey, 1977). That is, they use nedians

ralher than means as indieators of central tendeney, guarters ralher than

standard deviatlons as indiealors of dispersion, and so forth. These

technlques noderate the effects of unusuaily hi-gh or low perforrances on

the overall analysis, anci they enacle charters to use simple nechanical

melhods lo surmiarize data. Among these Lechn-iques is lhe quarter-j-ntersecl

raeLhod of drawing rceleralion lj-nes (','ftite, Nole 3: Pennypacker, ei al.,

tq72: itrhite and Haring, I975). Quarler-inlersect lrenci li-nes, when assigneci

rceleralion values (".g. x3,O, *1.50) sunmarize the effecLs of proceciures,

anri inciiviciuals I behavior change over pericds of tine, ft is possible lo

summarize a given project, phase, or course of behavior change by stating

lhe sLarling rale, rceleration, and ending rate.

Variabilit.y as a ralio. Having obtained a rceleralicn 1ine, i-t is

then possible to draw an envelope of two l.j-nes, para1lel to Lhe rceleration

Llne, whieh encompasses all or any desired percentage of the data polnls.

&u"ig (note 4) and Wtrite(Note 5) analyzed the effects of trimning various

percentages of data poinls from the envelope. Their findings were similar

to the general conclusion (e.g., Wainer, 1976) that trimsiing to 5a-75% pro-

vides a stable , yeL still representatj-ve value. Pennypacker et a1. (tgZZ)

reported that on the Standard Behavior Chart hurnan behavj-or frequencies show

a high degree of homogeneity of variance. They wrole:

This observation, combined r+ilh the fact that increases or decreases

ln behavior frequencj-es are accurately described by a slraight celer-


ati,on 1ine, allows us to make surprisingly accurate preciiclions

using these sinple chart procedures. Charting behavior frequen-

cies on anything other than a rati-o scale would require us to re-

sorl to extremely complex malhemalical operations for raking these

sane prediclions. The exlensive historical use of non-ratio charts

in psychology and education probably accounts for the facf that

very little hifherto has been acconplished in the area of predict-

ing hu:nan behavior, despite widespread agreement that such predic-

tion is both possible and desirable (p. 1W).

The envolope of bounce can be summarized as the ratio of lhe upper lo lhe

lower boundaries. We find, for example, thal the normal aciult population

exhlbils a range of beLween x1.50 and x2.50 on mosl well-practiced associ-

ative skills, whereas handicapped populalions rnay vary a great deal more.

Haring (trtote 5) used 5Ofr envelopes of bounce to dislinguish between severely

hanCicapped sludents wilhrrcompliance problemsil and those who did not exhibit

such irregular response to inslructions from day to da,v in inslructional

programs.

Change in rate as a ratio. A given change in procedure night pro-

duce an incrennental- change i-n frequency, or rate. For example, the change

fron lrials procedures to worksheet perfornance in the reading program des-

cribed in Chapter II r*as a frequency multipller of approximately x3.O.

Change in trend as a ra'b:!o. A given change in procedure may aller

the celeration. A tceleration multiplj.er is an index of this change. If

the two tcelerations have the same sign (x or i), one divicies the larger by

the snraller and gives the quotient the sign of the change ("x" for increase

toward positive slope, "i" for movement tor*ard negative slope). If the

signs of the tworceleralions differ, one aultiplies lhe two and gives the


product the sign of the change (pennypacker et a1., rg72),

Change in accuracv as a ralig. A given procedure nay prociuce a change

in aeeuracy. An irnprovement indgx is a rceleration nultiplier of the correcl

rate cereraLion and the eryor celeration (pennypacker et al., tgTz).

fn sununary, the logarithnlc ordinate of Lhe Standard Behavior Chart

offers a suostantial number of practieal acivanlages. The chart is a toolof great versatility and efficiency for the analysis of behavior and forthe conrn'"rnication of results.

fs the Semi-loqarith.nic Charl BesL?

Scientlst-s transform numerical dala when the transformatiotroviciesradvanlages for graphic or stalislical analysis, or when they bel-ieve thal

the chosen lransformation represents a more accurale model of natural phen-

omena. in the ease of the Stanciarii Behavior Chart, it, is clear thal the

logarithnic orciinate anC linear absci-ssa offer importanl practical anci ana-

lyt,ic advanlages. Nonelheless, il is still possible to question whether

the seni-logarithmic riisplay accuraLely models changes in response rate over

tine.

There are a variety of possible objeclions to using a seni-Iog charl

rather than one with a linear ordinale. The mosl obvious is that the raLio

chart obscures snrall differences between comparatively high response rates.

For example, whereas, the difference between 10 and 20 per minute is clear

on the chart, it is nearly impossible to distinguish 520 from 530 per ninute.

fn equating proportional rather than absolute (or interval) changes, the

logarithnric ordinate models the assumption that behavior rates change i-n

ratios ralher than in intervals. This assumption is egpecially evident inthe use of quarter-intercept tceleration Lines to sumnarize trencis.

The relevant statistical questions are whether the logarithnic ord-


inate preserves honogeneily of varianee across changes in 1evel, and

whether exLensions of tceleralion lines, baseci on logarithnically trans-

formed data accuralely predicr the future course of behavior.

Koenigts (gZz) doctoral lhesi-s acidressed jusl lhese quesLions.

From published reporls anci fron a compulerized I'bankrr of hunan response

rate infornation, he gathereci 11186 sets of data. Each sel included be-

tween 20 anci 29 days of ciaLa with an average chan6re of al least 1Of per

week (*l,b o*;1.10 fceleration). He founci that trenci lines drawn lhrough

these dala sets, by eilher qr:arter-inLersect or least squares regression

nelhods, bisected gOfr of lhe poinls in more Lhan 90% of the cases. He

found thal changes in range between firsl and fourth quarters were less

Lhan IO% in more Lhan 50% of the cases. And 5 Lo I riay projectj-ons of

variance enrie]6pss from lhe firsl 10 to 14 ciays of data eneompasses 7O,f,

of the points in more lhan hal-f the eases. These ciala suggest thal lhe

logarithmic lransforrnalion is appropriale for hurnan response rate ciala.

Morecver, Koening (8lZ) found lhat lhe quarter-inlersecl method was as

aeeurate j-n sunmarizing the dala as the leasl sguares regression technique.

One of the impii.cations of Lhe logarithnic transformalion of re-

sponse rates is that a given absolute increase represenls a greaLer change

from a lower rate lhan from a higher rate. This is a reasonable assump-

tion. A student who progresses frour five to len problems completeci per

minute seems, inluitively, to have nacie greater progress than one who

rnoves fron 50 to 55 per rainute. The first is a 1o0S (or x2.o0) l-ncrease,

r+hereas the latter is merely a tO% (or x1.10) inprovement.

0n the other slde of the transfornation issue there is the possi--

bilily that a power, or 1og-1og chart may fit human rate ciata better than

Lhe semi-logarithuric chart. Snoddy (tgZ+) originally suggesteci thaL the

typical shape of the learning curve for highly coherenl speed skills such


as typing, telegraphy, and i-ndustriar assembly is the power function,

accoriiing to which logarithmic incremenls in anount of praclice result inlogarilhnic gains in speeci. Blackburn (1935) came to a sirnilar conclusion

about the effeets of praetice on the rate of solving menlal arilh,qetic

problems. Klennrer (tqtZ) reporled dala showing lhat reaction times insinple perceptr:al-nrotor tasks fo11ow the same function. Anci, interestinglyenough, some researchers in the precision Leaching fielci have suggesteci

that the 1og-1og scale rnay actually fit classroom data betLer than the

log-inlerval SLandard Behavior Chart (White & Billi-ngs1ey, Note 8). They

have found that il is possible tc rake somewhat nore accurate preciictions

and decisions about performances in lhe proficiency-buil.ding sLage of

learning from dala plolted on 1og-1og charts.

ft rnay be thal the iog-Iog charlrs advantage appears only toward

the highest leveIs of proficieney, where physiological llnits, or limitsimposed by component skill deficirs, creale an asymptote. At that, point

lhe greater curve-straighLening ability of the 1og-1og scale erases such

a natural ceiling effect.

For the behavior anal;'sl interesteci primarily in pragmatics, the

l-cg-i-nlerval seale has decided advanlages. ltlhereas on the interval calend.ar

base il makes no ciifference where fhe course of behavlor change begins, on

a ratio scale lt makes a great deal of difference. In practice, then, 1og,

1og charts could not be calendar-sJmchronized across behavers and behaviors.

Although the power function on the 1og-1og scale allows for a sinple trath-

ematical model, the power chart lacks neny of the useful ratio properties

of the Stanciard Behavior Chart. If al sone point it becomes obvious that

the power charl is a better fi-t, then perhaps it will be possible to resolve

the practieal difficulties. In t,he rnean time, collecting response rate dala


and charting thern on a mechanically powerful char+, is a giant step aheaci

in the sl'ody of associative skilIs, Besides, it should be possible to

discover nany funclional relations, even if the fit is not optimal at allpoints along the course of skil1 developmenl,

Agsessrpnt. fnstruction. and C'.rricu1u8

Preeisj-on teaching has been largely an oral traciition with rela-

lively few publications in acadenic Journals or books. Most people in lhe

fielci are teachers or teacher-Lraj-ners, of whom few have any motlvaticn to

publish. There have been a large number of unpublished rnanuscripls, theses,

and working papers, but even the most reeent cf then do nol reflect the

state of knowledge in the fielci. Lindsley +-rained a mi,nbers of Ph. D.ts

at the universily of Kansas, and Penn.ypacker has done ihe same al the

Universj-ty of Florida. irlhite, Liberty, Haring, and their associaLes carry

on an aetive program of research at the Universily of Washington. But, for

the nosl parf, fieldworkers have been lhe main conlributors. School dis-

tricls in washingtcn, oregon, Kansas, Florida, Monlana, and onLario are

among the proni-nent centers of precision teaching activity. What follows

is an attempt to outline the rnajor methocis and finding+f these various.,groups of people r*ho cornmunicale largely by mouth anci with dala on the

Slandarci Behavior Chart. Being strictly functionalist in their orienta-

tion, they generally express their findings and results as empirical gen-

eralizations. Nevertheless, teachers and habilllators of any theoretical

persuasi-on can and do use the Standard Behevior Chart.

Experimenlal Methori

Precision teachers generally use sonn form of plal gheet (Kunzel-

rlann, Cohen, Hulten, I,lartin, & Mingo, 1970; Wirite and Haring, 1975) which

descri,bes the sequenee of antecedents, moverents, arrangements, and subse-


quent events that appear in a parlicular training proced.ure. Data-based

decisions about individual progress leaci lo changes in the plan. Thus,

precision leaching incorporales the basic ingredients of experinental

nethod. fhe plans and charts are often organized into a standard fil-e

systen for a given classroom or child (Binder, Note 9).

Channels Termj-nology

Many precl-sion teachers use a ternainology of ehannels to categor-

ize tasks. They classify lnput channels into such categories as t?hearrr

'tseertr rrtouchr tr lrmarkr " "dortt rrthinkr,, anto forth. For example, oralI

reading is a see/say task, as is naning cbjecls, numbers, cr eolors. si-lent rea'ding is see/tirink. Spontaneous speeeh is think/say, whereas an-

swering queslions nrigh+- by see/say or hear /"^y, ciepending on whether the

questions uere writlen or spokan. These categories are very useful fordiagncslic anal5r5is of skills since specific defici+"s nay appear in eilherinpu',, output, or bcth. They are also useful in the development of skil1taxonornies. (",g., Kovaks, Note 10).

Basie Measurement Techniqueg

Precision leaching generally involves daily measurenent. Because

ehildren as young as five years o1d can learn to counL, lime, and chart

their own performanees (tsates & Bales, 797I), nost non-handicappeci children

measure their own behavior in cooperation with peers and under the super-

vision of teachers. Most skill assessnents are probes ranging fron 15

seconds to two minutes in driralion, depending on the skj-I1. Usr:al1y there

are instructional and practice periods in addition lo the neasuremenl tines,

although nany skills show satisfactory growLh wilh only a few ninutes of

practice per day. Assessment periods are fj:<ed durations from day to day,

nhenever possible. This practice ensures a nore rellable assessment.


As much as possible, teaciring and rneasurenenl procedures are oesigned

so lhat lhey do not l-npose ceilings on performance. For example, spelling

tests are given in a irfree dictationt' mode (Haughton, 1972). Rather than read-

ing a word and then waiting until all children have completeci their alLenpts

to spel1 it, the teacher simply reads the words at a norrnal speaking frequency

and allows children to seleel words, write them ciown, and lhen lj-slen for

others. In thi-s way, all stucients have enough to keep them busy, anci they

can each demonstrale their own, self-paced proficiency. Many skills are

seefwrt'-e tasks. Worksheels allow students io move at lheir own pace and to

work in whal sequence lhe;r choose. In thi-s way lhe precision teaching class-

room conbines the best aspecls of so-calledtrfree" school-s anci Lhe rigor of

data-based i-nstruclion. The teacher becores more of an adivsor to individuals,

involving them in the day'to-day decision-rnaking proeess, than a group l-eclurer.

Sludenls learn self-rnanagernent, and are moLivaled cy their own pro$u". Timers

anci counlers of all kinds abound in such classrooms.

Early Dlscoveries

Rela'"ive unimFnrlance of consequenges. Slnce Lincisley and many of

his early assoeiates were operanl conriilioners, lhey originally emphasized con-

sequences are necessary bul nol sufficienl conciilions for skill developmenl

(Haughton, L972). Alfhough precise use of functional consequenees can produce

acquisition of skilI, the early *orker$ounci lhat students would often conlinueI

to exhj-bit rather low rates of performance in spite of fixed ratlos of rein-

forcement anci other arrangernents. They came to mueh the same conclusion as

Llnderwood (1955) who said, i-n sunnming up a sJrmposium on ski.Il acquisilion:

No participant in the conference seerned inpelled io defend the prop-

osition that reinforcement in the classical sense should enler as a

major Lheoretical notion into theorles of ski11 learning. The anal-


ogy between the animal learning laboratory and the skills laboralory

can sometines beeonre quite close, as for example, in the use of in-

formation feedback in sinple positioning rnovements. Nevertheless,

the consensus seems to be that sueh feedback is best eonceived as a

reans of defining to the subject the nature of the response to be

learned rather than as a reinforcing event in the classical sense.

At least with nonhandieapped children, consequences seered to be primarily a

way of naintaining praclice ag a task. Sone sinple meLhods of ciifferential

reinforcement of high rates boosted childrenrs proficiencies, but nol nearly

so mueh as careful sequencing of anlecedenls and lhe developnent of prerequis-

ile mcvemenl c.ycles.

Resqgnse:response relationships. Haughton (t972) recounts how early precision

teachers firsl became aware of the importance of high levels of proficiency in

tool skills, basic movemenl cycles which conslitute lowest conmon denomi-nators

across acadenic ski11s. He and his associates underslooci the basic principles

of task analysis, and thus had oegun to carefully sequence prerequisiLe skil1s

in order lo naxinize positive transfer. But they had not yet learned the im-

portance of high levels of proficiency in the basics. It became clear, for

example, Lhal students shoulci be able lo think/write lrs and Ofs very profi-

cienlly in orcier to be able to progress smoothly fhrough a printing curricu-

Ium. They began to see that reading words quickly -- any words -- was an

important prerequisite for moving on to new material. Proficient speaki-ng,

think/say sounds, is prerequisite to proficient reading. Arui students shoulci

be able to see/say numbers and thinkft;:i-Le numbers quickly in order to progress

smoothly through an arithnetic sequence. Thus, even though they had been

measuring response rates from the very beginning, the early precision teachers

took sorne time to ful-1y appreciate Lhe inporLance of high rates of responding.


Even thcugh lhey were neasuring rates, they were slill thinking in terms of

absence/presence of behavior, or percenlages. But 1n lhe early 1970,s they

becane aware of what have more recenlly been lermed deficit-irnposed ceilines

(Binrier, Note 11).

There are at leasl four finds of ceilings which inpose linits on lhe

growlh of proficiency. ft is necessary lo remove lhe measuremenl-imposgci eeil-

in€ in order to inspecl a studentts skill proficiency beyonci rere acquisition.

Changes in procedure remove teacher-impogeci ceil-i-nFsrallowing sLudenls to per-

form at lheir cwn pace. It then becare obvious that ihere are rieficil-iroposed

ceiLLnFs, created by lack of profici-ency in importanL prerequisile rnovement cy-

cyles. For example, stucients who cannol proficiently fill a conlainer with

objecls will experience difficulty in sorling lhe objecls int,o ciifferenl con-

lainers. And filling containers nay be constrained by cieficits in reaching,

grasping, placing, or rel-easin,q objecls. Simi-1arly, 1ow rates in writing cr

prinling letters will constrain lhe developmeni of proflciency in free-diclation

spel-Iing or any other wri',i-ng lask, With physically or menlally hanciicapped

ehildren, it may be lhal cerlain irrenediable lirnils, called stucienl-defined

ceil-inss (Binder, Note 1l-), will impose permanenL li-rnils on the growlh of

certain skj-lls. In that eventrit will be necessary to provide prosthetic

training, or proslhetic environments (linastey, 1951+),

In short, just as perceptual-notor ski11 researchers (Bitoaeau &

Bilodeau, 1951+i Gagne, Baker & Foster, 1950) had founci thal irnproving the per-

fornance of a less proflci-ent component, even at the expense of practice on a

more proficient one, could boost over-all task proficlency, so the precision

teaehers discovered the iniportance of proficiency in response-response rela-

lionships. This is also consistent with Barreltrs (llote t) f:-nAing that sore

severely retarcied subjects could learn discrinlnations as thoroughly as any


normal aciult, but, that they were deficlent in response rate and Lhus took

l-onger to acquire the discriminaLions.

Fluenc.y Aims

Precislon teaehers fornalized their findings about the need for high

leveIs of component ski1l proficiency in terms of fluenc-y aims, or response

rale cri-teria (Haughlon, L9?2). They began systematically to determine the

rates at which students nust perform component skills to progress snroothly

from aequisition t,o proficieney, or fluency, in subsequently learneci ski-l1s.

Often charls wouli indicate that sludents were having difficulty in ,Cevelop-

ing fluency in particular ski1ls. Slepping back Lo prerequisite skills anci

boosling proficiency often eLiminated. this difficully on relurn to lhe sub-

sequenl step in the sequence. Thus, teachers began to discover nininum f1u-

ency aims for each step i-n well-ordered skil1 seguences. They then were able

tc cievel-op assessmenl stralegies based on these aims.

St,arlin (tgZZ), for example, devei-oped assessment invenLories fororal reading, basic computalional skills, and spelling. Her slrategy was

baseC on three levels of assessment. An inveltory leve=1 would consisl of a

relatively large sLiee of the material, for exarnple, addition sufits fron 0-9.

The scregning level incluCeC smaller slices within a given inventory, for

example, sums fron 0 Lo 5. Fina11y, the item Ievel included cumulative sets

of specific llems within a screeni-ng, for example, sums involving only the

addition of 1, 2, ar:'ri j, In reading.and spelIing, the leve1s would be deler-

rdned by phonetic regularities, length of words, and so forth. Starlin fearned

that certain ranges of performance could guide the teacher j-n selecting the

appropriate leve1 of instruction. fn oral reading, she found thaL a perfor-

mance of fewer than 2O words per ninuLe indicates the neeci to sU-ce back_to

a sraller amount of material, perhaps from the inventory level lo a screening


Ievel. And a perforuance of 50 to 1CO words per ninute indicaled that students

could practice independently and continue to improve.

Minimu:ro ains for application or transfer. Thus, teachers discovered

the need for establishing mlninun rate ains to ensure snooth application of

component skills to subsequenl composite skills. If we think of response rate

as a measure of probabilily, this flnding sinply elucidates the requirenenl that

skills occur with high probabillty under appropriate eonditions in order for

combinalions of those skills to occur with high probability. This is perhaps

the nost basj-c finciing of precision teaching. Anci it is a principle with

broad generality, applicable to a range of issues in ti:efstudy of associaliveI

skills. It is founded on the use of response rate as a rpasure of such

skiIIs.

Endurgnce and retention. High levels of fluency also appear to be

necessary for endurance, or nainlenance of fluency over comparatively long

periods of perfornance (Haughton, Note 12). After students develop high

levels of proficiency in short assessmenL periods, they becone more capable

of pracLicing on their own for longer periods of line. Haughton has estinated

that, to be capable of pracliclng skills w'ithoul supervision, students shoulti

be able to perform those skil1s at at least one half the fluency stanriard.

Otherwise, they will decrease their performance rates in the absence of super-

vj-slon. Endurance and the rnalntenance of frattentionri to task appear elosely

related, aspugg""ted ln Chapter II. Haughton (Uote 12) clalns that retention

is also reLated to fluency, and }'lhlte and Harine Og16) suggest that failure

to naintain skills may be due to lnadeqr:ate levels of proflciency, There may

be two dimensions to these related observations. First, high probabilities of

performance nay be related to retention in the traditional sense. ThaL is,

skilts which are well established in the first place are less likely to be for-


gotten during a period of non-use.

High levels of fluency probably rnake their greatesl contrioution lo

retention by enabling studenls to naintain ski1ls through application in higher

skills. BarretL (L977) suggested that in a well-designeci ski11 sequence, rrsince

every skil1 in the sequence is a component of or a contributor to the skitl

above it, skills learned early in the sequence are, presurnably, being used re-

peatedly and therefore shoulci not di-sappear fron the behaverts reperloire.r'

Norgs. fluency. anci mastgrv. Cert,ain levels of perform€Lnce are nec-

essary for efficient, effeclive use in appropriate appl-cations. People who

read aceuralely, but slowly, experience difficully when they have to use read-

ing skills in educational or occupational- settings. This is a particularly

good reason for moving beyond mere accuracy measurement and for removing leach-

er-imposed ceilings. Furlhermore, slow readers do nol generally read for rec-

realion and lherefore do nol naintain whatever ski11 they once attained. Sim-

i1arly, retardeci Frsons who are capable of making change accwately, but very

slowly, are unlikely to nai-ntain the ski1l. Instead, friendly or impatienb

shopkeepers will probably do it for then. A defini-lion of fIuenc.v, then, is

the rate at which skills are effeetive and relatlvely efficienl. Dysfluency

is defineci as a rate aL r*hi-ch ski-lls are i-neffeclive.

One way of determining minimr:m flueney standards is to assess people

assurned to be nornal in their perfornance of ski1ls. Thus, precision teachers

have gathered data to establish norms (Barett, L979). Adult norns are the

usual standards of fluency, although teachers often use peer norms as ways of

estabLishing age-appropriate ains (Cnifa Demonstration Prograns, Note 2; While &

Haring, 7976).

Because norng seldom represenl levels of ski1l nastery, another approach

(Haughton, Note l-2) seeks to riiscover the levels at which truly exceptional


lndividuals perform skilIs. With these as aims, it nay be possible to develop

true maslery in large numbers of individuals. In fact, whereas Haughton (nlZ)

previ-ously set 1O0 words per rninute with fewer than 2 errors as an airn for be-

ginning readers, he now suggesls an aim of 200 to 3OO per minute as an appro-

priale level (Haughton, Note 12). Alt,hough it may be dangerous to set aims

too high, it is certainly ciangerous to set airns too low.

Decision Rules

The Standard Behavior Chart forms a basis for educational- decision-

naking. But il does nol nake the decisions. Teachers musl do so. And in

orcier to increase the frequency with i*hich teachers make data-based instruc-

tional decisions, a number of precision Leachers, especially Kat,hleen Liberty

anci Owen White, have been workingr to develop a series of decision rules based

on palterns of data points on the chart.

Freschits (f9?4) di-scussion of precision teaching represents the

least formal approach to decision-making. Teachers simply inspect charled data,

noting whelher rales of eorrect and ineorrecl responses are changing in the

desj-red ciirections, or nol. ff they are, the teacher continues the curenl

plan. If nol, the plan is changed. Although daily data collection and data-

based decisions in this form represenl a major acivance over typical educational

procedures, nany precision leachers nonetheless make very few declsions, allow-

ing procedures to continue wlthout change even r+hen ineffective.

Noting ihis low frequency of decision-making, Liberly (tlote f3) in'

vented d-vnanic ai4g, or progress aims, This procedure involves specification

of a fceleralion, or rate of change, according to which teaehers can Judge

whether adeqr:ate progress is occurring. Liberty analyzed data on several hun-

dred charts including all kinds of skiIls and children of aIl ages. She found

that of those students whose progralns produced @ progress, 5O16 shor*ed growth


of x1 .33 or better on acceleration targets anci :1 .l+6 for deceleration targets.

Aboul 53% progressed at x1.25 for acceleraLion, 56% aL +t.Zj for deceleralion.

She suggested 1.25 as a nlnimum tceleration. After collecting three days of

data and finding the nidpoint,, teachers using ciynamic ains draw a tceleration

Ii-ne fron that point toward the fluency alm on the chart. If on more than

three days in a row, the sludentts rate fa1ls below (or above, in the case

of ciecelerations) tne dynamic aim, the teacher chan6ies the p1an. Teachers

rnay learn thal sorne chlldren are able to change faster than 25% per week, and

thus establish indiviciualized progress aims for such children. But the xL.25

or i.1.25 rule is a useful first choice for rni-ninurn fceleralion.

The use of dynamic aims substantially increased the frequeney of de-

cislon-makine (Haring & liberty, Note 14), but it still does not suggest what

ehanges are l-ikely lo produce the desired effect. Should the teacher alter

the consequences, slice back on the content, step back lo an easier skill,

or alter the instruclicnal procedure? A five-year research project, curenlly

uncier way at the University of Washington, is designed to sol,ve this problen

(Haring, Note 15). With Lioerty as its coordinator, this project has speci-

fied and repeatedly revised a seL of rules according to which teachers can

make the change mosl likely to succeed. Afler drawing splil-nriddle fcelera-

tion U-nes lhrough patterns of correcl and error rates on the chart, teachers

refer to these rules, whi-ch suggest the appropriate change for a glven pattern

of accelerations and/or decelerations of corrects and errors (Haring, Note 15).

0ver the several years of the proJect, the researchers have collected data on

the successes and fai-Iures of each suecessive rerrision of the ru1es. The most

recent seL of rules predicted the effects of plan changes 5-n an average of 68%

of the cases, across the various categories of change.

Although the University of Washington group has also attempted to de-


rive such rules on the basis of percenlage eorrect data, lhey have been unable

to cio so.

Diasnostig lPre s :rlptiveJs s es sments

Establishnent of fluency ains and the developlent of decision-making

rules has allowed precision teachers to carry out precise diagnostic/prescrip-

tive assessnents. The sinplest form of assessnent involves a few daysr co1l-

ection of data, and comparison with fluency ains and/cr peer ranges of perfor-

nance across a selection of skills. Even one-day samples ("snapshotst') are

useful in this regard (".g., Haughton, Note 17). fl is possible to apply this

form of assessnent within a classroom, in an entire school, or across several

school districts (Ctr:-fa Service Demonstralion Prograrns, Note 2).

However, it is far rnore useful to select a set of componentl pr€r€q-

uisite, and conposite sk1l1s in a given currieulum area, and then assess stu-

dents for 5 to L0 days on those skills. In that r*ay, not only levels of pro-

ficiency, but also tcelerations over the duralion of the assessment period,

al1ow teachers to decide where to place students in a given curriculum and what

cieficits to remediate (Wfrite & Haring, 1976). This approach represents a sub-

stantial improvemenl over traditional forns of stanciardized assessrnent. Il

allows students to demonslrate profi.cieney Ievels on speclfic skills, with re-

peated opportuni-ties, over a reasonable period of tine, Anri it compares their

actual performances to objective norns of performance on those skills raLher

than to very general norms, establlshed according to psychonetric principles.

Instruetional Imp]icatlons of Fate Megsurenent: An bcanple

In most elementary schooJs reaciing is either irnplicitly or explicitly

assesseci on a percentage correct scale. Accordingly, students firsf learn to

read a story or IisL of words, anci then when they hav$xhioited the ability to"tdo so, pass on to the nexl story or set of words. fn fact, since most educa-


tional methods do nol even involve criterion'referenceci assessment, students

pass on Lo lhe next story whether or not lhey were able to accuralely read lhe

previous one. But in any case, teachers do not generally advance students on

the basis of fluency aims. As a result, firsl-graciers nay pass through a list

of books, still reading at between 30 and 80 words per ntinuLe. Doehring (t97t)'

for example, found that average first graders read about 33 s.vllables per nin-

ute of randon words on worksheets, and about 46 syllables per mi-nute in text.

These relatively low rates of reading nay create severe problens in relention

and comprehension,

A precision teacher would require stu<ients to reach relatively high

fluency aims on their first selection, between 200 anci 3OO words per ntinute

(l,Ienninga anci Brough, Note 18). Having reached this rale on a given passage,

the stucient nay be reciting it from memory. Bul the student is nonetheless

reading it, responding wilh appropriate movement cyles to textlr:al slimuli.I

And as successive passages increase vocabulary, it wilf oecomerimpossible to

memorize them.

With each change to a new passage, there will be a rate decrement and

an increase j-n eryors. Bul relatively quickly, the performance will return to

fluency. Thus, from the very beginning, a childrs oral reading will be main-

tained at near adult leve]s of fluency. Precision teaching enphasi-zes fluency

as a prerequisite for increasing conlent. The traditional approach emphasizes

eontent, often at the expense of fluency.

PrecigloLTeaghine wlth the Rgtardg!

Although it was work wlth the retarded that originally inspired the

development of the Standard Behavior Chart (Llnastey, 1954), and nany of the

first precisj-on teachers worked with the retarded (GaUoway, 1972), most of

the inportant advances in precision teaching untit recenlly have occurreci in


elassrooms for regular and rrildly handicapped students (-Wtrite & Haring, 1976).

There are severaf reasons for thls.

First, severely retarded studenLs generally have not learned to chart

their own behavior. Uslne the Standard Behavior Chart in cLassroorns for such

students therefore involves significantly nore work for teachers than does work

wlt,h students who do their own tinlng, counting, and chartlng.

Second, it is more difficult to remove ieacher-imposed ceili-ngs from

instructional procedures for the severely retarded. Whereas regular children

can work wilh worksheets and other self-presented naterials from the very be-

ginning, painstaking procedures are often needed to move severely retarded

students toward self-paced procedures (Pease and George, Note 19). Therefore,

teachers are not readily convinced that profi-ciency is possible for such stu-

dents.

Third, severely retarded and o+,herwise handi-capped stucients often ex-

hibit substantial day-to-day variability in performance (Haring, Note 6). This

nay result fron various neurological impairments, physlcal illness, or the fail-

ure of teachers to oblain instructional compliance.

Finally, only recently have precision teachers begun to look carefully

at critical tool skills for many of the fine anci gross motor perforrnances Lhat

constitute the curricula in most cLassroons for the severely retarcied. Haughton

(trtote 12) fras observed that deficiis in reaching, pointing, touching, grasping,

placing and releasing lnpose ceilings on most performances of Lhese students.

He has suggested that teachers return to these baslc notor components, along

with basic body control rnovements, in f,heir progranrning for the severely re-

tarded. Haughton anci hls associates (Kovacs, Note 10) have begun to establish

flueney aims for these skiIls.


Precision Teaching as Functional Analysis

Above all, preci-sion teachers are functionalists. From its concep-

tion (Llncisley, 1972) precision teaching has been neasurenpnt-based and atheo-

retical, following the tradition of funeLional analysis established by Skinner

(nSA). The establishment of fluency aims, curriculum sequences, and decision

rules have developed directly from ennpirical data, with no intervening leveIs

of hypotheses or constructs. fn this sense, precision teaching differs from

all other approaehes to education. In fact, Lindsley (tgZt) tras denied lhat

it is anrrapproach.r' Rather, he says, it is a precise way of moni-toring the

effeets of whaLever proceciures a teacher night choose. It is simply the daily

measurement of response rates in the classroom. In lhis sense, it also differs

frorn a grea+- deal- of lhe theory-based research thal eciucalors and psychologisls

have condueled on various aspecLs of learni-ng and skil1 acquisition.

Ia,Berge and Samuels (tgZLn), for example, present a theory of I'auto-

matic information processing in reading.r' The following quotation contrasls

sharply with the stricLly functional analysis of precision teaching:

In the present model it is assumed that all well-learned stimuli are

processed upon presentation inlo an internal representation, or code,

regardless of where attention is dj-rected at the tire . . Il is

assurned that we can only attend to one thing at a time, so long as

no nore than one requires aLtentlon.... Our criterion for deciding

r,rthen a ski11 or sub-skll1 is autornatic ls that it can complete its

processing while attenti-on is directed elsewhere. (p. 29t+-295)

In various flowcharts, the authors construct an elaborate system of eociers,

deteetors, and processors. Like Fitts (fq54) they use a eomputer nrodel of

rrprograms and sub-routinestr to aecounl for observed functional relationships.

They come to the satre conclusions as precision teachers as to the


difference between accuracy and proficieney at different stages in the learn-

ing process, and they use tine as a neasurerent dimension. r?Further lraining

beyond Lhe accuracy eriterion musL be provlded'r, they say, 'rif the associalion

is to occu-r without attention,... lalency serves as the critical j-ndicator of

autonatic assocj-ative processing'r (p. 3O?). In this fype of theorizing, hypo-

thetical processes appear nore imporlant than data. Thus they speak of meas-

urernenl as indicaline an underlying state of automati-city, rather than as

sinply one elerent in a functional relationship.

Like precision teachers, laBerge & Samuels (fqZ4) find that practice

beyond accuracy is imporlant. And they remark thal:

The finding that unfarnllj-ar letters improved wi-th practice more than

did familiar letters offers supporL for the hypothesis that sonrething

is being learned aoout Lhe unfamiliar letters over the days of train-

ing.... It is evidenl lhat lhe lalency difference of a narning response

to the new and old letlers is quite large at firsl and converges over

days of training....when aecuracy appears to be stationary (p. 3O3).

The finding that learning continues beyond mere accuracy 1ed them to 'r?he main

eonclusion.,..that what is being improved with practice is automaticity" (p. 304).

But what is I'automaticity"? As Skinner has frequenLly pointed out (1950,1974,

l97S), theoretieians often move the environnent inside the organism, accor:nling

for an observation - - in this caae a short latency response which is relatively

inmune to distraction - - in terrns of an lnternalj"zeri synonyrn for the behavior.

In thls context, we night ask whether such stove creates no obvious practical

predictive advantage.

IaBerge & Samuels (tgl4) agree with precision teachers on the inpor-

tance of feedback:

Since the important growth of automaticity takes place after the sub-

ject has achieved aceuraey, overt feedback for correct and incorrect


responses nay be redundant because at this stage of learning the sub-

ject knows when he is correct or not. However, there is another lype

of feedback which may affecl the rate of automaticity learning. 'yihile

learni-ng proeeeds toward the autornatic leve1, it rrighl be appropri.ate

to inforrn the subject of the tire it took to execute his response.

G. 3t7)

This, of course, is exactly the kind of feedback provlded to slucients who count,

time, and charl their own behavior. LaBerge and Samuels go on to say:

0f course, latency feedback for a response to a partlcular stimulus

will nol be meaningful by iLself, bul musl be related to some crit,erion

baseU-ne. For example, lhe tine it takes to identify a new word should

be compared to the tine il lakes to identify a word lhat is already al

the automatic ]evel. (p. 3f?)

This, of course, if the fultion of fluency aims i-n precision teaching.

Relating proficiency lo skill-sequencing, laBerge and Samuels note

thai "the fluenl reader has presumably nastered each of the subskills aL the

aulonralic levelr' (p. 3tg). Continuing in this vein, they say:

We are tempted to generalize to classroom routines in elementary

schools in which letter naming is directly taught and tested only up

to the accuracy level. A child nay be quite accurate in naming or

sounding the letters of the alphabet, but we nay not know how mueh

attention it costs hin to do lt. This kind of infornation could be

helpful in predlcting how easily he can nanage new learn-ing ski1ls

which build on associations he has already learned.... One shoulci take

into account the amount of attention required by these subskills as

a part of the readiness eriterion,

We night ask, why not sinply examine how fluently a student must be able to


F€rform a subskil-l to progress smoolhly through the next curriculum l-evel?

Does the concept of I'atLentionrr add anything here?

fn discussing rrlevels of processing, " laBerge and Sanuels invoke lhe

notion of chunking to aceount for increased fluency. They observe thal t'lo

encourage chunking, we may have to relax the demand for accuracyt' (p. 319).

This speculation is strikingly consistent wlth Haughtonts (tgZZ) finding that

until students reach a reading raie of about 50 words per ninute, attenpts to

decelerale emors are largely ineffective. After that point, it is relatively

easy to reduce error rates.

Finally, IaBerge and Samuefs (i974) rernark:

Throughout this paperr we have stressed the importance of autornaticity

in perforrnance of fluent readi-ng. Now we lurn to a consicieration of

ways to train reading subskllls to automatic 1evels. Unforiunately,

very 1ilt1e systernalic r?l-search has been ciirected specifically to

lhis advanced stage of learning. (p. 314)

Doehring (tqZt), writing in a sinilar vein, asserts that rra deeper understand-

ing of the learning process is necessary before reading inslruction can be irn-

provedlr (p.t*Z). He says thal among the greatest dj,fficulties are I'lhe prob-

lens of describing the accuracy and speeci of responses in the same qr:antila-

tive terns." (p. 42)

As this review of precision teachi-ng indicates, there has been a gooci

deal of work on the developrent of fluency in reading, although 1ittle has been

published. The work has focused, not on the I'underlying proeessesrrt but on

the discovery of functional relationships. It has useri neasures of response

rates and aceuracy ratios, whlch solve the problerns of sinultaneously quanti-

fyrng accuracy and speed and which are also more convenient and less costly

than latency neasures.


fn the continuous feedback loop between basic and appli-ed research,

it night serve the interests of bolh researchers and educators to ernulale the

functi-ona1 approach of precision teaehers. Their work has uncovered a number

of inportant phenonena lhat deserve closer inspection from basic researchers.

0n Conpelene.y Testing

There is cunently great concern about the failure of schools lo teach

students reading, writing, and computational skilLs. l4any have suggested that

all students should take competency lests at the end of high school - - a prop-

osition thal nighl only result in proof of failure long after it is possible

to do anything aboul it. Precision teaching, incorporating daily, data-based

decisions aboul instructi-ona1 effectiveness, would seem a more appropriate anti-

dole. The alternalive seens to be coniinueci wave afler wave of lheoretical

explanations aboul how learning occurs, and why lt doesnrt. Without an empir-

ical approach, based on daily E€asurement, it seems likely that functional i\\ -

ileracy, now estirnaled to affllct at least 20% of the adult populaLion in this

country, will continue to increase.

Surunarv and Conclusions

Precision teaching is based on functional analysis of behavior, orig-

inally developed in the free operant condj-tioning laboratory. ft involves the

measuremenl and charting of response rales on the Standard Behavior Chart, a

powerful communication device .

The Standard Behavior Chart is a slx-cyc1e senilogarithnic graph.

Its ordinate is logarlthrrlc whlle its 140-day calendar base is llnear. ft ls

possible to plot durations and latencies on the chart as reciprocals, enabllng

the charter to inspect relations between fluency and session length, arong

other things. The ordinate spans a range of response rates from one per day

to 1W0 per minute. Its logarithmic scale nakes it possible to exani-ne accu-


racy, variability, and trenci lndependent of rate. It also enables the charter

to sumrnarize data in terns of a nuncer of ratios or nultipliers.

Precision teaehing ls prirnarily a rpasurenent system. 0n the basis

of that system, teaehers have dj-scovered a variety of enpirical regularities.

fn partlcular, they have discovered the importance of developing high leve1s

of fluency in conponent skills. Fluency is related to transfer, retention, or

maintenance, and to endurance of skills. These findings lead to the establlsh-

ment of fluency airns and decision-naklng rules and to diagnostic/prescriptive

skill analysis.

fn eontrast to an inforrnation processi-ng approaeh, precision teaching

has achieved a great deal in classrooms, wlthout ornate theoreti-cal struclures.

As a basic methodology, precision teaching seems eninently applicable to lab-

oratory study of associative skilIs.


References

Abrams, S. G., & Zuber, B. L. Some Characteristics of information processing during reading. Reading

Research Quarterly, 1972, 8(1), 40-51.

Abrams, Jack A. Motor behavior. In Melvin Marx (Ed.), Learning Processes. New York: The Macmillan

Co, Inc., 1972.

Atkinson, R.C., and Shiffrin, R.M. Human memory: A proposed system and its control processes. In K.W.

Spence & J.T. Spence (Eds.), The Psychology of Learning and Motivation. New York: Academic

Press, 1968.

Barrett, B. H. Behavior analysis. In J. Wortis (Ed.), Mental Retardation and Developmental Disabilities

Vol. 9. New York: Brunner/Mazel, 1977.

Barrett, Beatrice H., & Lindsley, Ogden R. Deficits in acquisition of operant discrimination and

differentiation shown by institutionalized retarded children. American Journal of Mental

Deficiency, 1962, 67(3), 424_436.

Barrett, Beatrice H. Acquisition of operant differentiation and discrimination in institutionalized retarded

children. American Journal of Orthopsychiatry, 1965, 35(5), 862-885.

Barrett, Beatrice H. Communitization and measured message of normal behavior. In R. York & E. Edgar

(Eds.), Teaching the Severely Handicapped Vol. 4. Columbus: Special Press, in press.

Bates, S., & Bates, D. F. “… and a child shall lead them”: Stephanie’s chart story. Teaching Exceptional

Children, 1971, 3, 111-113.

Beck , C. H. M. Paced and self-paced serial simple reaction time. Canadian Journal of Psychology, 1963,

17(1), 90-97.

Bilodeau, Edward A., & Bilodeau, Ina McD. The contribution of two component activities to the total

psychomotor task. Journal of Experimental Psychology, 1954, 47(1), 37-46.

Blackburn, J. M. Acquisition of Skill: An analysis of learning curves. Great Britain, I.H.R.B. Report No.

73, 1936.

Braam, Leonards S., & Berger, Allen. Effectiveness of four methods of increasing reading rate,

comprehension and flexibility. Journal of reading, 1968, 11(5), 346-352.


Briggs, George E. Transfer of training. In Bilodeau, E. A., & Bilodeau, Ina McD. (Eds.), Principles of

Skill Acquisition. New York: Academic Press, 1969.

Butler, D.C., & Peterson, D.E. Learning during “extinction” with paired associates. Journal of Verbal

Learning and Verbal Behavior, 1965, 4, 103-106.

Cattell, J. McKeen. The time it takes to see and name objects. Mind, 1886, 11, 63-65.

Connolly, Kevin, Brown, Kathleen, & Bassett, Eryl. Developmental changes in some components of a

motor skill. British Journal of Psychology, 1968, 59(3), 305-314.

Cofer, Charles N. Verbal Learning. In Melvin Marx (Ed.), Learning Processes. Mew York: The

Macmillan Co., Inc., 1972.

Demckla, Martha Bridge, & Rudel, Rita. Rapid “automatized” naming of pictures, objects, colors, letters,

and numbers by normal children. Cortex, 1974, 10(2), 186-202.

Doehring, Donald G. Acquisition of rapid reading responses. Monographs of the Society for research in

Child Development, 1976, 41(2, Serial No. 165).

Dresslar, F. B. Some Influences which affect the rapidity of voluntary movements. American Journal of

Psychology, 1892, 4, 514-527.

Duncan, Ann Dell. Behavior rates of gifted and regular elementary school children. (Doctoral dissertation

University of Kansas, 1969). Dissertation Abstracts International, 1969, 29(10-A), 3520.

Eaton, Ira E. The relationship between perceptual-motor ability and reading success. (Doctoral

dissertation, St. Louis University, 1975). Dissertation abstracts, 1975, 6, 3562-A.

Ebbinghaus, H. Memory: A Contribution to Experimental Psychology. (H. A. Ruger & C.E. Bussemius,

trans.). New York: Teachers College, Columbia University, 1885.

Edwards, D.D., & Edwards, J. S. Fetal Movement: development and time course. Science, 1970, 169, 95-

97.

Eimas, P. D., & Zeaman, D. Response speed change in an Estes’ paired associate “miniature” experiment.

Journal of Verbal Learning and Verbal Behavior, 1963, 1, 384-388.

Ellis, Henry C. Transfer and Retention. In Melvin Marx (ed.), Learning Processes. New York: The

Macmillan Co., Inc., 1972.


Estes, W. K., & Skinner, B. F. Some quantitative properties of anxiety. Journal of Experimental

Psychology, 1941, 29, 390-400.

Eysenck, S. B. G. Retention of a well-developed motor skill after one year. Journal of Genetic

Psychology, 1960, 63, 267-273.

Ferster, C. B., & Skinner, B. F. Schedules of Reinforcement. New York: Appleton-Century Crofts, 1957.

Fitts, Paul M. Perceptual-motor skill learning. In Arthur W. Melton (Ed.), Categories of Human Learning.

New York: Academic Press, 1964.

Fitts, P. M., & Peterson, J. R. The Information capacity of discrete motor responses. Journal of

Experimental Psychology, 1964.

Fitts, Paul M. The information capacity of the human motor system in controlling the amplitude of

movement. Journal of Experimental Psychology, 1954, 47(6), 381-391.

Fleishman, Edwin A. Comments on Professor Jones’ paper. In Bilodeau, Edward A. (Ed.), Acquisition

of Skill. New York: Academic Press, 1966.

Freschi, D. F. Where we are. Where we are going. How we’re getting there. Teaching Exceptional

Children, 1974, 6, 89-97.

Gagne, Robert M. The Conditions of Learning. New York: Holt, Rinehart & Winston, Inc., 1970.

Gagne, R. M., & Foster, Harriet. Transfer of Training from practice on components in a motor skill.

Journal of Experimental Psychology, 1949, 39, 47-68.

Gagne, R. M., Baker, K. E., & Foster, H. Transfer of discrimination training to a motor task. Journal of the

Experimental Psychology, 1950, 40, 314-328.

Gagne R. M. Task analysis:-- its relation to content analysis. Journal of Educational Psychology, 1974, 11,

11-18.

Gagne, R. M., & Briggs, L. J. Principles of Instructional Design. New York: Holt, Rinehart & Winston,

1974.

Galloway, Charles. Precision parents and the development of retarded behavior. In June B. Jordan, &

Lynn S. Robbins (Eds.), Let’s Try Doing Something Else Kind of Thing. Arlington, Virginia:

Council for Exceptional Children, 1972.


Goldiamond, Israel. Machine definition of ongoing silent and oral reading rate. Journal of the

Experimental Analysis of Behavior, 1962, 5(3), 363-367.

Goss, A.E. Manifest strengthening of correct responses of paired-associate under post criterion zero

percent occurrence of response members. Journal of Genetic Psychology, 1965, 72, 135-144.

Hall, John F. Verbal Learning and Retention. Philadelphia: J.B. Lippincott Co., 1971.

Hall, R. F., & Wenderoth, P. M. Effects of number of responses and recall strategies on parameter values

of a paired-associate learning model. Journal of Verbal Learning and Verbal Behavior, 1972, 11,

29-37.

Haughton, Eric. Aims: Growing and Sharing, In June B. Jordan, & Lynn S. Robbins (Eds.), Let’s Try

Doing Something Else Kind of Thing. Arlington, Virginia: Council for Exceptional Children,

1972.

Heinz, Robert D., & Eckerman, David A. Latency and frequency of responding under discrete-trial fixed-

interval schedules of reinforcement. Journal of the Experimental Analysis of Behavior, 1974,

21(2), 341-355.

Herrnstein, Richard J., & Boring, Edwin G. (Eds.). A Source Book in the History of Psychology.

Cambridge: Harvard University Press, 1965.

Hull, C. L. Principles of Behavior. New York: Appleton-Century, 1943.

Hull, C. L. The place of innate individual and species differences in a natural-science theory of behavior.

Psychological Review, 1945, 52, 55-60.

Horton, David L., & Kjeldergaard, Paul M. An experimental analysis of associative factors in mediated

generalization. Psychological Monographs: General and Applied, 1961, 75(11, Whole No. 515).

Irion, Arthur L. A Brief history of research on the acquisition of skill. In Edwards A. Bilodeau (Ed.),

Acquisition of Skill. New York: Academic Press, 1966.

Irion, Arthur L. Historical introduction. In E. A. Bilodeau & Ina McD. Bilodeau (Eds.), Principles of Skill

Acquisitions. New York: Academic Press, 1969.

Jenkins, James J. Mediated Associations: Paradigms and situations. In Charles N. Cofer & Barbara S.

Musgrave (Eds.), Verbal Behavior and Learning. New York: McGraw-Hill Book Co., Inc., 1963.


Johnston, J. M., & Pennypacker, H. S. A behavioral approach to college teaching. American Psychologist,

1971, 26, 219-244.

Judd, Wilson A., & Glaser, Robert. Psychonomic Bulletin, 1967, 1(2), 11.

Judd, Wilson A., & Glaser, Robert. Response latency as a function of training method, information level,

acquisition, and overlearning. Journal of Educational Psychology, 1969, 60(4, Pt. 2), 30.

Keller, F. S. The Phantom plateau. Journal of the Experimental Analysis of Behavior, 1958, 1(1), 1-13.

Keller, L., Thomson, W. J., & Tweedy, J. R. Efforts of reinforcement intervals in pair-associate learning.

Journal of Experimental Psychology, 1967, 73(2), 268-277.

Klemmer, E. T. Communication and human performance. Human Factors Journal, 1962, 4, 75-79.

Kunzelmann, H. P., Cohen, M., Hulten, W., Martin, G., & Mingo, A. Precision Teaching: An Initial

Training Sequence. Seattle: Special Child Publications, 1970.

LaBerge, David, & Samuels, S. Jay. Toward a theory of chronic psychotics as revealed by free operant

conditioning methods. Diseases of the Nervous System Monograph, 1960, 21, 66-78.

Lenneberg, Eric H. Speech as a motor skill with special reference to nonaphasic disorders. In Ursula

Bellugi & Roger Brown (Eds.), The acquisition of language. Monographs of the Society for

Research in Child Development, 1964, 29(1, Serial No. 92).

Leonard, J. A., & Conrad, R. Maintenance of high accuracy without augmented feedback. Nature, 1963,

199, 512-513.

Lindsley, Ogden R. Direct measurement and prosthesis of retarded behavior. Journal of Education, 1964,

147, 62-81.

Lindsley, Ogden R. From Skinner to precision teaching: The child knows best. In June B. Jordan, & Lynn

S. Robbins (Eds.), Let’s Try Doing Something Else Kind of Thing. Arlington, Virginia: Council

on Exceptional Children, 1972.

Maliphant, R., Supramaniam, S., & Saraga, E. Acquiring skill in reading: a review of experimental

research. Journal of Child Psychology and Psychiatry, 1974, 15(3), 175-185.

Marx, Melvin. Learning Processes. New York: The Macmillan Co., Inc., 1972.

McGeoch, J.A. The acquisition of skill. Psychological Bulletin, 1927, 24, 437-466.


Milward, R. Latency in a modified paired-associate learning experiment. Journal of Verbal Learning and

Verbal Behavior, 1964, 3, 309-316.

Morton, John The effects of context upon speed of reading, eye movements and eye-voice span. Quarterly

Journal of Experimental Psychology, 1964, 16(4), 340-354.

Naylor, James C., & Briggs, George E. Long-term Retention of Learned Skills: A Review of the

Literature. United States Air Force ASD Technical Report (No. 61-390), 1961.

Osgood, Charles E. Meaningful similarity and interference in learning. Journal of Experimental

Psychology, 1946, 36(4), 277-301.

Otto, W., & Fredericks, R.C. Relationship of reactive inhibition and reading attainment. Journal of

Educational Psychology. 1963, 54, 227-230.

Pennypacker, H. S., Koenig, C. H., & Lindsley, O. R. Handbook of the Standard Behavior Chart. Kansas

City, Kansas: Precision Media, 1972.

Peterson, Lloyd R. Paired-associate latencies after the last error. Psychonomic Science, 1965, 2, 167-168

Postman, Leo. Transfer, interference and forgetting. In Kling, J. W., & Riggs, Lorrin A. (Eds.),

Experimental Psychology. New York: Holt, Rinehart & Winston, Inc., 1971.

Richardson, J., & Gropper, Mitsi S. Learning during recall trials. Psychological Reports, 1964, 15, 551-

560.

Roberts, M., Bondy, A., Mira, M., & Cairns, G. Continuous tracking of behavioral tracking in infants.

Journal of Genetic Psychology, 1978, 132(2d Half), 225-60.

Runquist, W. N. Verbal behavior. In J. B. Sidowski (Ed.), Experimental Methods and Instrumentation in

Psychology. New York: McGraw-Hill, 1966.

Shackel, B. Pilot study in electro-oculography. British Journal of Ophthalmology, 44, 89-113.

Schmidt, Bernard. Changing patterns of eye movement. Journal of Reading, 1966, 9(6), 379-385.

Skinner, B. F. The Behavior of Organisms. New York: Appleton-Century Crofts, 1938.

Skinner, B. F. Science and Human Behavior. New York: The MacMillan Co., 1953.

Skinner, B. F. The Technology of Teaching.

Skinner, B. F. Walden Two. New York: Macmillan, 1948.

Skinner, B. F. Are theories of learning necessary? Psychological Review, 1950, 57, 193-216.


Skinner, B. F. The analysis of behavior. American Psychologist, 1953c, 8, 69-79.

Skinner, B. F. A case history in scientific method. American Psychology, 1956, 11, 221-223

Skinner, B. F. About Behaviorism. New York: Knopf, 1974.

Skinner, B. F. Why I am not a cognitive Psychologist. In B. F. Skinner, Reflections on Behaviorism and

Society. New York: Prentice Hall, 1978.

Shoddy, G. S. Learning and stability. Journal of Applied Psychology, 1926, 10, 1-36.

Spring, Carl. Encoding speed and memory span in dyslexic children. Journal of Special Education, 1976,

10, 35-40.

Spring, Carl. Naming speed as a correlate of reading ability and sex. Perceptual and Motor Skills, 1975,

41(1), 134.

Starlin, Ann. Sharing a message about curriculum with my teacher friends. In June B. Jordan, & Lynn S.

Robbins (Eds.), Let’s Try Doing Something Else Kind of Thing. Arlington, Virginia: Council on

Exceptional Children, 1972.

Stennett, R. G., Smythe, P. C., Pinkney, June, & Fairbairn, Ada. The relationship of eye movement

measures to psychomotor skills and other elemental skills in learning to read. Journal of Reading

Behavior, 1972-73, 15(1), 1-13.

Stevens, Joseph P. Applications of power functions to perceptual-motor learning. Journal of Experimental

Psychology, 1964, 68(6), 614-616.

Stevens, Joseph C., & Savin, Harris B. On the form of learning curves. Journal of the Experimental

Analysis of Behavior, 1962, 5(1), 15-18.

Taylor, S.E., Frackenpohl, H., & Pattee, J.L. Grade level norms for the components of the fundamental

reading skill. Bulletin No. 3, New York: Huntington, Educational Development Laboratories,

1960.

Taylor, Stanford E. Eye movements in reading: facts and fallacies. American Educational Research

Journal, 1965, 2(4), 187-202.

Thalberg, Staton P. Reading rate and immediate versus delayed retention. Journal of Educational

Psychology, 1967, 58(6, Pt. 1), 373-378.


Theios, John. Reaction time measurements in the study of memory processes: Theory and data. In G. H.

Bower (Ed.), The Psychology of Learning and Motivation, Vol. 7. New York: Academic Press,

1973.

Tinker, Miles A. Devices to improve speed of reading. Reading Teacher, 1967, 20(7), 605-609.

Underwood, Benton J. Motor skills learning and verbal learning. In Edward A. Bilodeau (Ed.),

Acquisition of Skill. New York: Academic Press, 1966.

Wainer, Howard. Robust statistics. Journal of Educational Statistics, 1976, 1, 258-312.

White, Owen R., & Harring, Norris G. Exceptional Teaching. Columbus: Charles Merrill, 1976.

White, Owen R., & Liberty, Kathleen A. Behavioral assessment and precise educational measurement. In

Norris G. Haring, & Richard L. Schielfelbusch (Eds)., Teaching Special Children, 1976.

White, Sheldon H., Spiker, Charles C., & Holton, Ruth. Associative transfer as shown by response speeds

in motor paired-associate learning. Child Development, 1960, 31, 609-616.

Wolfe, Dael. Training. In S. S. Stevens (Ed.), Handbook of Experimental Psychology. New York: John

Wiley & Sons, 1951.

Wickens, C. D. Temporal Limits of human information processing: a developmental study. Psychological

Bulletin, 1974, 81, 739-755.

Williams, Joanna P. Learning to read: A review of theories and models. Reading Research Quarterly,

1973, 8(2), 121-126.

Wright, Logan, & Willis, Carol. Reminiscence in normal and defectives. American Journal of Mental

Deficiency, 1969, 73(5), 700-702.

Reference Notes

1. Barrett, Beatrice H. Annual Report 1972-1973. Waltham, Massachusetts: Behavior Department, Walter

E. Fernald State School, 1973.

2. Child Service Demonstration Programs Progress Report IV. Tacoma, Washington: Intermediate School

District N. 111, 1974.

3. White, O. R. The “split-middle”: A “quickie” method of trend estimation. Working Paper No. 1,

University of Oregon: regional Center for Handicapped Children, 1971.


4. Koenig, C. H. Charting the Course of Future Behavior. Unpublished doctoral dissertation, University of

Kansas, 1972.

5. White, O. R. The prediction of human performances in the single case: An examination of four

techniques. Working Paper No. 15, University of Oregon: Regional Center for Handicapped Children,

1972.

6. Haring, Norris G. An investigation of phases of learning and facilitating instructional events for the

severely handicapped, Annual Progress Report, University Of Washington, Experimental Education Unit,

1976-77.

7. Haughton, Eric. Personal communication, October 27, 1978.

8. White, Owen, Billingsley, Feliz, & Munson, Robin. Evaluating the daily progress of severely and

profoundly handicapped pupils. Paper presented at the meeting of the American Association for the

Education of the Severely and Profoundly Handicapped, Baltimore, October, 1978.

9. Binder, C. V. Guide to a Standard File System. Waltham, Massachusetts: Behavior Prosthesis

Laboratory, Walter E. Fernald State School, 1979.

10. Kovacs, Mary (Ed.), A Practical Taxonomy of Normal Body Control and Beginning Skills. Ontario:

Hastings County Board of Education, 1978.

11. Binder, C. V. Four kinds of Ceilings. In C. V. Binder (Ed.), Data-Sharing (Waltham, Massachusetts:

Behavior Prosthesis Laboratory, Walter E. Fernald State School), October, 1978.

12. Haughton, Eric. Untitled contribution. In C. Binder (Chair), More parameters of pupil freedom.

Symposium presented at the meeting of the American Association for the Education of the Severely and

Profoundly Handicapped, Baltimore, October, 1978.

13. Liberty, K. A. Data Decision rules. Unpublished working paper, Experimental Education Unit, University

of Washington, 1975.

14. Haring, Norris, & Liberty, Kathleen. “What do I do when?” Stages of Learning and facilitating

instructional events. Paper presented at the meeting of the American Association for the Education of the

Severely and Profoundly Handicapped, Baltimore, October, 1978.


15. Haring, Norris G. Field initiated research studies: An investigation of learning and instructional

hierarchies in the severely/profoundly handicapped -- Renewal application for the fifth year. University

of Washington, Experimental Education Unit, 1979.

16. Haring Norris G. An Investigation of phases of learning and facilitating instructional events for the

severely handicapped, Annual Progress Report. University of Washington, Experimental Education Unit,

October, 1978.

17. Haughton, Eric (Ed.). Snapshot Notes. Ontario: Hastings County Board of Education, 1977.

18. Menninga, Rita, & Brough, Mary. Screening grade one – three. Ontario: Hastings County Board of

Education, June, 1978.

19. Pease, D., & George, F. Flashcards to worksheets: Transitional training in normalization of academic

behavior. In Instructional Development and Behavior Analysis Program, Annual Report. Waltham,

Massachusetts: Behavior Prosthesis Laboratory, Walter E. Fernald State School, 1975.


measuring response rates in associative skill...

Documents