the effects of instruction in an inference strategy … · inference strategy on the reading...
TRANSCRIPT
THE EFFECTS OF INSTRUCTION IN ANINFERENCE STRATEGY ON THE READING
COMPREHENSION SKILLS OF ADOLESCENTSWITH DISABILITIES
Nanette Salim Fritschmann, Donald D, Deshler, ami Jean Bragg Schumaker
Abstract. The purpose of this study was to determine the effectsof teaching eight secondary students with disabilities, includingseven with learning disabilities, a strategy for answering a varietyof inferential questions. A multiple baseUne across-subjects designwas employed. Outcome measures included scores on researcher-devised comprehension quizzes, a standardized test of readingcomprehension, a strategy use test, a strategy knowledge test, anda reading satisfaction measure. Fidelity of implementation,instructional time, and maintenance of skills were also measured.Results suggest that students with disabilities can learn to use astrategy to answer a variety of inferential questions, and masteryof its use can result in improved scores on criterion-based andstandardized measures of reading comprehension. In addition,students' satisfaction with their reading improved.
NANETTE SALIM FRITSCHMANN, Ph.D., Lehigh University.DONALD D. DESHLER, Ph.D., University of Kansas,
lEAN BRAGG SCHUMAKER, Ph.D., Edge Enterprises, Inc., Lawrence, Kansas.
The current educational climate and its calls forincreased skill acquisition and rising performancedemands are requiring students to learn higher-orderreading skills, like inference skills (e.g., AmericanInstitute for Research, 2005; Partnership for 2IstCentury Skills, 2006). With few exceptions, all second-ary students, including students with disahilities(SWD), are required to take rigorous state reading com-petency exams, most of which involve the use of infer-ence skills. At present, 26 states administer exit exams,and 19 of them withhold diplomas based on poor per-formance on the exit exams {Center on EducationPolicy, 2005).
Increased local demands appear to he rising in tan-dem with the level of reading skills evaluated by
national standardized assessment exams. For example,the proposed 2009 National Assessment of EducationalProgress (NAEP) reading framework reflects expansionfrom its current 1992-2007 framework, to include theassessment of broader reading content and deeper cog-nitive processes {American Institute for Research,2005). This framework represents a shift from assessingskills at the literal/word level of reading comprehen-sion to assessing higher-order skills within readingcomprehension that emphasize "interpreting and inte-grating" reading matter, the very skills required tomake inferences.
This climate of increased reading demands in schoolsand on tests poses significant challenges for strugglingadolescent readers. For students who have a disa-
Voliinw iO, full 2007 245
bility, increased demands are especially problematic(Bulgren, Marquis, Deshler, Schumaker, & Lenz, 2006;Schumaker, Deshler, Bui, & Vernon, 2006). Someresearch has shown that students with learning disabil-ities (LD) enter seventh grade reading, on average, atthe fourth-grade level, and they do not make gains inreading achievement as they progress through the sec-ondary grades (Deshler & Schumaker, 2006; Deshler etal., 2006; Warner, Schumaker, Alley &c Deshler, 1980).Further, large proportions of these students are failingtheir state reading competency exams (Heubert, 2002),as well as tests in their required high school courses(Bulgren, Schumaker, & Deshler, 1988; Hughes, Deshler,Ruhl, & Schumaker, 1993; Wagner et al., 2003).
This is understandable, because, although some ofthem have acquired some basic decoding skills (Catts,Fey, Tomblin, & Zhang, 2002), they have not learnedmany of the skills associated with reading comprehen-sion, including inference skills (Gersten, Fuchs,Williams, & Baker, 2001). Fhe combination of moredemanding academic requirements and their inade-quate reading skills contributes to poor academic out-comes for students with LD (Leach, Scarborough, &Rescorla, 2003; Mastropieri, Scruggs, & Graetz, 2003;Wagner et aL, 2003).
Further complicating matters is the fact that higher-order skills, such as those involved in reading compre-hension, in general, and making inferences, inparticular, can be much more difficult to teach studentswith LD to a point of proficiency than lower-orderprocesses (Fisher, Schumaker, & Deshler, 2002;Swanson, Hoskyns, & Lee, 1999). Making inferences hasbeen defined as the ability to "construct the text baseand the mental models that go beyond the informationdirectly articulated in the text" (Snow, 2002, p. 108).The ability to generate inferences is typically tested byasking questions like, "Why did the boy take action likehe did?" or "What is the main message of this passage?"
The answers to such question do not appear directly inthe text. Instead, the reader is expected to integrate cluesin the text with prior knowledge to create an answer.Thus, although the ability to generate inferences may becritical to text comprehension, the generation of aninference in reading is essentially the result of the in-dividual reader's response to the ideas presented in thetext; this is somewhat dependent on his/her ability toconnect or bridge those ideas with some prior knowl-edge and with clues provided in text (Pressley, 2000).
To provide a theoretical framework for what happensduring inference generation, Kintsch (1998) suggestedthat during the comprehension process, mental repre-sentations are constructed or formed about the infor-mation being read in text. When the reader thinks,talks, or writes about these representations, they
"undergo integration, which results in a well-struc-tured" (Kintsch, 1998, p. 95) understanding of the text.In the case of drawing inferences from text, Kintsch pos-tulated that various factors (e.g., text features, languageskills, and domain knowledge) contribute to compre-hension and assist the reader in integrating informationinto a meaningful structure. When such integrationoccurs, the reader is able to draw a successful inferencerelated to the text. Also, according to Kintsch, thisprocess may be either automatic (unconscious) or con-trolled (conscious and strategic).
Research conducted with elementary-level studentslends support to Kintsch's theory. For example, somestudies have shown that the scores of poor comprehen-ders on inferential comprehension questions improvewhen they receive prompts to attend to integrative fac-tors like those highlighted by Kintsch (1998), such astext features and background knowledge (Cain &Oakhill, 1999), and when they are given integrativestimuli (e.g., a descriptive title) along with the passage(Yuill & Joscelyne, 1988). Other research studies inwhich poor comprehenders have been explicitly taughthow to make inferences have focused on teaching stu-dents to attend to integrative factors similar to thosespecified by Kintsch. In each study, students weretaught one or some combination of the following skills;activating their background knowledge, making predic-tions, asking and answering questions, looking for cluesin the text, making connections between prior knowl-edge and information in the passage, and attending totext structure.
Although the results of the studies in this area aresomewhat mixed, three studies show positive treatmenteffects for poor comprehenders (e.g., Dewitz, Carr, &Patberg, 1987; Hansen & Pearson, 1983; Yuill &Jocelyne, 1988). However, none of these studies focusedon students with disabilities or on secondary students.The poor comprehenders' average posttest scores onsome of the criterion-based tests hovered below thepassing range. Furthermore, none of these studies usedstandardized measures of reading comprehension.
With regard to secondary students, researchers focus-ing on the use of conscious strategic reading compre-hension processes have reported that students with LDcan learn to use comprehension strategies that conceiv-ably could contribute to inference generation (Gerstenet al., 2001; Swanson et al., 1999). Examples of suchcomprehension strategies include summarization(Gajria & Salvia, 1992), activating background knowl-edge and prediction (Afflerbach, 1990), and clarifying(Simmonds, 1992). Other research has shown that sec-ondary students with LD can learn complex readingstrategies and that their scores on criterion-based read-ing comprehension measures can increase as a result of
Learning Disability Quarterly 246
01/5 e;
UJ
O\
oa^ II cO .2
aSoura C•o oc Sifl 1/5
75
OOX
acUJ
I Iu'ca.
u'CV
cflu'C
• 4 - 1
X!H
00
oc
1_2S>oII
Valiunc 30, Fall 2007 247
the strategy instruction (see Schumaker & Deshler,2(X)6, for a review).
Each of the studies in this area has focused on onereading strategy (e.g., self-questioning, visual imagery)that might be related to inference generation. None hasaddressed inference generation as an outcome measureor has employed a standardized reading test to measurechanges in comprehension. I urther, no study to datehas investigated teaching a comprehensive package ofstrategies that might be used for generating severaltypes of inferences.
Thus, the purpose of this study was to develop andtest the effects ofan instructional program designed toteach an inference strategy to secondary students withdisabilities. Specifically, the study was designed toexamine the effects of explicit instruction in a multi-component inference reading comprehension strategyby assessing (a) student knowledge of the strategy, (b)student use of the strategy while reading narrative pas-sages, (c) student ability to answer four types of infer-ential questions as well as literal comprehensionquestions, (d) student scores on a standardized measureof reading comprehension, (e) student reading andstrategy satisfaction, and (f) required instructional timefor students with disabilities in a secondary setting.
METHODParticipants
Participants were eight ninth-graders with disabilitieswhose parents had given consent for their participa-tion. Students were enrolled in learning-supportedEnglish/language arts classes. A standardized readingtest, the Group Reading Assessment and DiagnosticEvaluation (GRADE) (Williams, 2001), indicated thatthe students' reading scores fell at least five grade levelsbelow their current grade placement. Each student hadbeen designated as a student with a disability and hadbeen placed in a resource program for a minimum of180 minutes per day on the recommendation of a mul-tidisciplinary special services team with documentedparent/guardian approval (see Table 1 for demographicand test data on the students). Of the eight partici-pants, seven had been classified by their school districtas having a learning disability; one had been classifiedas having mental retardation. The participating districtfollowed the IQ-achievement discrepancy model foridentification of learning disabilities, requiring a mini-mum discrepancy of 18 points between a student's IQand achievement scores 0- Harrington, personal com-munication, July 21, 2007). Further, the state withinwhich the participating district is located is one of fivestates in this country that rely on the professional judg-ment of a team in the determination process (Reschly,Hosp, & Schmied, 2005).
SettingIhe school district is located in an urban midwestern
community with a population of approximately124,000. Instruction and testing took place in a class-room in the high school and was conducted with twogroups of four students each during different class peri-ods.The Inference Strategy
The Inference Strategy is a reading comprehensionstrategy designed to help students create meaning fromclues provided in text and respond to a variety of infer-ence questions. Standardized reading tests were ana-lyzed to determine the types of inference questions thatstudents were expected to be able to answer. Four maintypes emerged: purpose, main idea/summarizing, pre-dicting, and clarifying (see Figure 1 for examples).
The Inference Strategy taught to the students in thisstudy consists of five steps. During Step 1, "Interact withthe passage and the questions," students first previewthe passage, paying particular attention to the title andthe length of the passage. Then they read the questionsand mentally identify two main categories of questions:factual questions and think-and-seek (inferential) ques-tions. Next, the students further classify the think-and-seek questions into four types; purpose, mainidea/summarization, prediction, and clarification ques-tions. During the second step, "Note what you know,"students activate any background knowledge or experi-ences they may have related to the topic and questions,underline any key words in the questions that indicatewhat information to look for in the passage, and notecode letters next to each question to indicate the cate-gory of the question and the question type. During thethird step, "Find the clues," students carefully read thepassage and find and underline clues that are directlyrelated to key words in the questions. They then createtentative answers to the questions mentally. The fourthstep, "Explore more details," prompts students to lookfor any additional clues in the passage that support thetentative answers they have selected. The final step,"Return to the question," calls for the students to goback to each question and make sure that an answer hasbeen selected and marked.
Thus, these steps were designed to cue students toattend to their prior knowledge, to attend to the type ofinference they were being asked to make, to attend tokey words in the questions that would lielp them searchfor clues in the text, to searcli for those clues, and toproblem solve once they had gathered the clues to makeinferences ahout the information in the passage. Theuse of the generic strategy steps was constant regardlessof the type of question the student was addressing.However, students looked for different types of clues.
Disability Quarterly 248
Figure 1. Example inferential questions.
Purpose Questions:1. What is the author's reason tor writing this passage?
2. The author's purpose for writing this passage is to ...
i. What is the most likely reason the author wrote this passage?
Main Idea/Summarizing Questions:
1. Which of the following sentences best summarizes this passage?
2. This passage is mostly about...
3. The main topic of this passage is ...
Predicting Questions:
t. What is the next most likely event to happen after the end of this passage?
2. What would likely happen if the Olympics were held in Germany again?
3. Based on this passage, in the future, Joe will probably ...
Clarifying Questions:
1. The man at the newsstand toid Darren a dollar was a lot of money because
2. Why was Wilbur afraid?
3. What caused Charlie to lose his glasses?
depending on the type of think-and-seek question. (Seethe Procedures section for a description.)
The mnemonic device "INFER" was created from thefirst letters of the steps to help students to remember thesteps' names and their order in the strategy. Initially,the strategy steps are to be used in the "1" to "R" order;however, thereafter, the steps may be used in a recursiveand flexible manner to allow students to cycle back toany step if necessary.Instruments and Measures
Fidelity checklists. A checklist was used to measureteacher adherence to the instructional sequence for thelessons, it listed several teacher behaviors; provide anadvance organizer, discuss the purpose of the lesson andprovide rationales for the lesson, state expectations for
student behavior, describe a step of the strategy or howto use the strategy, model the strategy, provide practiceopportunities with feedback, and provide a post organ-izer.
The number of these behaviors varied for each les-son's checklist (from 7 to 11), depending on the contentof the lesson. For example, if one type of inferentialquestion was to be introduced in a given lesson, therewere seven items on the checklist, corresponding tothose listed above. If two question types were to beintroduced in a given lesson, then the strategy wasdescribed and modeled twice, once for each questiontype, and nine items were listed on the checklist.
1 he delivery of each of the instructional lessons wasrecorded using a tape-recorder and audiotapes. Thechecklist was filled out by a scorer while listening to
Voltmic JO, lull 2mi7 24')
audiotapes of the lessons. If the scorer heard the teacheremit one of the teacher behaviors listed on the check-list, one point was awarded for that behavior. Zeropoints were awarded if the behavior was not emitted. Apercentage score was calculated for the percentage ofteacher behaviors in which the teacher engaged acrossthe lessons.
Strategy use test. To obtain repeated measures ofstudents' use of the Inference Strategy, a pool of 30ninth-grade-level narrative passages from the JamestownReaders - Timed Readings in Literature Series (Spargo,1989) was created; passages were randomly selected andsequenced for each student from this pool. This level ofpassages {i.e., ninth grade) was chosen to provide infor-mation on performance at the students' current gradelevel because this is the level at which they are expectedto perform in general education classes. Narrative pas-sages were selected to control for the type of passage aswell as to fulfill requirements for the English class inwhich the students were enrolled.
For each probe test, the student was asked to readthree of the 400-word passages silently and use theInference Strategy in relation to the passage and the fivefactual and inferential questions that followed the pas-sage (see the next section for more about these ques-tions). To measure use of the strategy, students wereawarded one point for each of the following strategicbehaviors: underlining key words in a question, under-lining clue words in the passage related to the question,recording a code letter identifying the category of ques-tion (i.e., factual versus think-and-seek), and recordinga code letter identifying the type of think-and-seek ques-tion {i.e., purpose, main idea/summarization, predic-tion, or clarification).
Scorers used an answer key for awarding a total of fourpoints per think-and-seek question and three points perfactual question. (Students classified think-and-seekquestions by type. They did not have to classify the fac-tual questions by type, so they could not earn a fourthpoint for these questions.) A total of 19 points wereavailable per passage for the strategy use score or 57points per probe test (since three passages were read perprobe test).
Criterion-based comprehension test. As mentioned,each passage the students read was followed by fivemultiple-choice questions written by a researcher specif-ically for a given passage. The five questions consistedof one factual question and one question for each of thefour types of think-and-seek questions: purpose, mainidea/summarization, predicting, and clarifying. Studentanswers to these questions were the criterion-basedcomprehension measure used in this study. Studentswere awarded 1 point per correct answer (based on ananswer key) for a total of 5 points per passage and a total
of 15 points per probe test (since there were three pas-sages per probe test).
Strategy knowledge test. The Strategy KnowledgeTest was used to measure the students' knowledge ofthe steps of the Inference Strategy as a pretest andposttest measure. It included five short-answer ques-tions, which required students to list and explain thestrategy steps and their uses. An answer key specifiedthe parameters for correct answers. For example.Question #1 was, "What is the first step a reader takeswhen using the INFER Strategy?" {The correct answerwas "Interact with the questions and the passage.") Foreach answer, a student could earn either zero points{no information), 5 points {partial information [e.g.,the student wrote "Interact with the questions"]), or 10points (completely correct information [e.g., the stu-dent wrote the whole name of the .step]). A total of 50points was available on the test.
Standardized reading test. Two subtests of theGRADE (Williams, 2001) were administered to thestudents prior to and after instruction: SentenceCompletion and Passage Comprehension. The com-bined scores from these subtests comprise the GRADEcomprehension composite score, which was the scoreused in this study. Forms A and B of the GRADE wereused for pre- and posttesting, respectively. Reliabilitycoefficients for the alternate forms and test-retest are inthe .90 range. In addition, correlations between scoreson the GRADE and on the Gates-MacGinitie ReadingTest "range from .81 to .94, with half of the coefficientsbeing .89 or higher" (Williams, 2001, p. 85).
Student satisfaction survey. The Student Satisfac-tion Survey consisted of 10 questions, each formattedusing a 7-point Likert-type scale, ranging from "\"{"Totally wrong") to "7" ("Totally right"). This instru-ment was administered to obtain a pretest and a posttestmeasure. On survey items, the students were asked torate their attitudes toward reading in school and learn-ing and toward using the Inference Strategy for thepurpose of reading passages and responding to compre-hension questions. Example survey items are "I am agood reader in school," "I know what steps I can take tomake meaning from what I read," and "I feel that I canuse the Inference Strateg>' to help me understand whatI read in class."
Time required for instruction, instructional time wasrecorded in a journal kept by the researcher. Start andstop times, including hours and minutes, were recorded,along with dates of instruction. Teacher time beganwhen the teacher started {or restarted) the lesson withthe students. It ended when an interruption occurred(e.g., a phone call), or when the students began practic-ing the strategy. Student time began when the teacherstarted {or restarted) the lesson with the students. It
Learning Disahitity Quarterly 250
ended when an interruption occurred, or when thestudents stopped practicing and handed in tbeir work.Thus, several start and stop times were potentiallyrecorded for each lesson.
Reliability. For the fidelity checklists, two scorersindependently scored 40% of the lessons, and theirscores were compared item by item. An agreement wastallied if both scorers had recorded the same score on anitem. The percentage of agreement was calculated bydividing the number of agreements by the number ofagreements plus disagreements and multiplying by 100.The scorers agreed on 38 out of 39 possible agreementsfor a total percentage of agreement of 97% (range = 91%to 100%).
Similar reliability checks were completed for theStrategy Use Test, the criterion-based comprehensiontests, and the Strategy Knowledge Test for 4O'Mi of thetests taken during baseline and post-intervention condi-tions. Scorers were blind as to which tests were takenduring baseline and after instruction, respectively. Withregard to the Strategy Use Test, the scorers agreed on949 out of 969 possible agreements, for a total percent-age of agreement of 98'M> (range = 93% to 100.0%). Onthe criterion-based comprehension tests, the scorersagreed on lOO'M) of the 255 items scored. Finally, on theStrategy Knowledge Test, the scorers agreed on 28 itemsout of a possible 30, for a total percentage of agreementof 93'M. (range = 80% to 100%).
Reliability for teacher and student instructional timewas determined for three of the instructional sessionsduring which two scorers recorded the times separately.Agreement was registered each time the two scorers'times agreed on a start or stop time to the minute. Thetotal percentage of agreement on teacher time waslOO'K), and on student time it was 93'M). The scorers'times agreed on 10 out of 10 possible agreements forteacher times and on 15 out of 16 possible agreementsfor student times.Procedures
Instructor. The instruction was completed by the firstauthor, who is a certified special education teacher withfive years' experience teaching students with disabili-ties. She also is a certified strategic instruction modelprofessional developer with the University of KansasCenter for Research on Learning.
Instructional materials. An instructional protocol(Fritschmann et al., in prep.) was written to ensure thatinstruction was standardized across the two classes. Itwas comprised of scripted step-by-step instructions foreach lesson plus visual devices to be used during theinstruction.
Two types of practice materials were constructed. Forthe first type, nine short, one-paragraph passages were
written by a researcher followed hy two to four multi-ple-choice questions corresponding to the type of ques-tion(s) covered in each of the lessons. These shortpassages were used for practice activities immediatelyafter students had been introduced to how to use thestrategy for a new question type.
For the second type of practice material, narrative pas-sages from the fourth-, sixth- and eighth-grade levelsof the Jamestown Readers - Timed Readings in Literatureseries (Spargo, 1989) were selected, so students couldstart practicing the strategy with relatively easy passages(the fourth-grade passages). Gradually, across practiceattempts, they progressed to more difficult passages(sixth-grade passages), and then to passages written onegrade level below their current grade (eighth-grade pas-sages). The length of passages ranged from 200 to 400words, increasing with the reading level of the passage.For each passage, five multiple-choice questions (onefactual and four inferential questions, eacb correspon-ding to one of the four types of think-and-seek ques-tions) were written by a researcher.
These long passages and accompanying questionswere used after all the question types had been intro-duced to the students so that they could practice usingthe strategy with a mixture of questions. Student per-formance on these activities was scored using an answerkey and the guidelines described under the Measuressection for the Strategy Use Test and the criterion-basedcomprehension test.
Pretest procedures. Students completed a ReadingSatisfaction Survey and a minimum of three probe testscontaining a total of nine Strategy Use Tests and ninecorresponding five-item criterion-based comprehensiontests over a one-week period. For each probe test, aStrategy Use Score and a Comprehension Test Scorewere determined. When the third baseline probe wascompleted, four students whose baselines were stable(hereafter referred to as Cohort A) began instruction inthe Inference Strategy. Once the Cohort A studentsshowed an increase in their use of the strategy, the fourother students (Cohort B) completed at least one addi-tional baseline probe until their baselines were stable.Then they began the instruction.
Intervention procedures. The students receivedinstruction in the Inference Strategy in sessions rangingin length from 60 to 75 minutes, depending on theschool schedule. Instruction was based on a validatedinstructional methodology for teaching learning strate-gies to students with disabilities (Schumaker & Deshler,2006).
In the first instructional session, the students wereasked to make a commitment to actively learn and usethe Inference Strategy. Also, in the first class session, theinstructor explained and described in detail the steps of
Volume 30, Fall 2007 251
the Inference Strategy. This was followed by instructionthat included how to identify and code the two maincategories of questions, (a) factual questions (i.e., ques-tions with answers that are "right there" in the passage)and (b) "think-and-seek" questions (i.e., questions thatrequire the reader to really think about and seek outclues so that they can infer an answer). Additionally,the students were instructed on how to identify andmark each type of think-and-seek question with codeletters.
The students were then taught through descriptionand demonstration how to look for key words embed-ded in the question to code and respond to factual ques-tions. At the end of the first session, the studentscompleted practice activities for coding question cate-gories and answering factual questions for short pas-sages, and they received feedback on their work. If theymet mastery (i.e., earned 80% or more of the points),they proceeded to the next instructional session. If theydid not reach mastery on the practice activity, theycompleted additional practice activities until they metmastery.
During the second instructional session, the studentswere provided detailed information on purpose ques-tions and main idea/summarizing questions, and theywere taught how to identify and mark those questionswith code letters. They were also taught how to look forkey words embedded in the questions and clues imbed-ded in the text and how to correctly respond to thesetypes of questions. For example, for purpose questions,students were taught that authors may have three mainpurposes as they write a passage: to entertain, toinform, and to persuade. Each type was defined, andstudents were taught to look for the key words"author's purpose" or "author's reason" in the ques-tion. Then as they read the passage, they were taught toask themselves "Why do 1 think the author wrote this?"and attend to certain types of clues that would helpthem determine the author's purpose. For example, forinformative passages, they were taught to look for largeamounts of facts and details like they might see in atextbook. For entertaining passages, they were taughtto look for sections that made them happy or fearful.After instruction, discussion, and modeling, studentswere provided with short practice passages followed bya mixture of factual, purpose, and main idea questions,and were given feedback on their efforts. Again, mas-tery was required before students proceeded.
The third and fourth instructional sessions focused oninstruction of and practice with predicting and clarify-ing questions, respectively. As in the previous sessions,the researcher explained in detail how to identify thenew type of question and search for clues in the passageto support a correct response to that type of question.
This process was modeled for the students prior to thepractice activities with the short passages. The instruc-tor provided group and individual feedback during andafter each practice activity, and mastery was required.
In subsequent sessions, students were provided prac-tice activities with the longer fourth-grade-level read-ing passages, in which they were required to use all thesteps of the strategy in response to all question types. Ifthe student earned a score at or above the mastery lev-els (80% on the Comprehension Test, 70% on theStrategy Use Test), they were moved up to the nextreading level (sixth-grade level, then eighth-gradelevel). They continued to practice and receive individ-ual feedback until they reached mastery on a passagewritten at the eighth-grade level. (See Fritschmann,2006, for more details on the instruction.)
Posttest procedures. Upon completion of instruc-tion in the Inference Strategy and reaching the masterycriteria on one eighth-grade practice activity, studentstook a posttest containing three 400-word, ninth-gradepassages. After reading each passage and using thestrategy, students were asked to complete a five-question criterion-based comprehension test. Each stu-dent completed a minimum of one such posttest (con-taining a total of 3 passages and 15 questions).
Following these tests, the students were administered,as a group, the two selected subtests included in 1-orm Bof the GRADE (Williams, 2001). They were also admin-istered the Student Satisfaction Survey and StrategyKnowledge Test in a group setting. The students com-pleted each instrument independently, and there wasno time limit for test completion. This posttestingoccurred during the last weeks of the school year.
Maintenance procedures. Two maintenance testswere administered during the next academic year. Thefirst was administered eight months after the postteststo six of the eight original subjects who were presenton the day the test was administered. The students didnot receive a review of the Inference Strategy prior totaking the first maintenance test, nor had the strategybeen reviewed with them since they had completed thestudy eight months earlier. 1 here was no time limit onthe test.
A second maintenance test was administered to fourstudents who were still enrolled in the school 12months after the posttests. In this case, the studentstook part in a 45-minute review session where theypracticed using the Inference Strategy the day beforetaking the maintenance test.Research Design
A multiple-probe-across-subjects design (Horner &Baer, 1978) was employed to determine the effects ofinstruction on students' strategy use and reading com-
Leaming Disability Quarterly 2S2
prehension performance. All students were given atleast three reading probes before instruction began. Thestudents who served as the second tier in the design(Students 2, 4, 6, and H) had at least four probes beforeinstruction began. Their instruction commenced onlyafter the scores of students in the first tier of thedesign had improved. All students' baselines were sta-ble before they began the instructional program. Addi-tionally, a pretest-posttesl analysis was employed tocompare the standardized reading test scores earnedbefore and after instruction on the GRADE.
RESULTSFidelity of hupiemeutation Results
I he instructor emitted 86 ot 88 listed behaviors onthe fidelity checklists for a total percentage of imple-mentation of 98%.Strategy Use Test and Criterion-BasedCoinprehension Test Results
Figures 2-4 display the percentage of points earnedby each of the eight participants on each Strategy UseTest (diamond symbols) as well as the percentage
Figure 2. Percentage ot points earned for strategy use and percentage of comprehension questionsanswered correctly by Students 1 and 2.
STUDENT 1Baseline
STUDENT 2
I
COI
cent
age
Pen
100908070-60-50403020100 • -
Instruction
4 4 4 6 8Reading Passage Grade Level
Generalization
9 4 6 8 9Reading Passage Grade Level
strategy Use — Comprehension
JO, I'iill 2007 2SJ
Figure 3. Percentage of points earned for strategy use and percentage of comprehension questionsanswered correctly by Students 3, 4, and 5.
STUDENT 3
o£oo
2couwa.
.{QQ_ Baseline9 0 -80-70-60-5040-
•301 y~~^~'-m20 ' _ - ' ' '10-0 - — • • •
STUDENT 41009080706050
*S 40fl) 30
u4)
O
o0)u 20 '
10-0 -
STUDENT 51009080706050403020100
9
Instruction Generalization
4 4 4 6 8Reading Passage Grade Level
4 6 8 9 9Reading Passage Grade Level
£ooVoim"co9Q.
4 4 6 8 9Reading Passage Grade Level
strategy Use - - • » - - - Comprehension
Leariiitig Disubility Qiuirterly 254
Figure 4. Percentage of points earned for strategy use and percentage of comprehension questionsanswered correctly by Students 1 and 2.
STUDENT 6
• • - •uo0
CJuroBc0)o
•IQQ Baseline
9080-70-605040 • - , .3020100
• • ^
• • •
STUDENT 7
£oo0)O)(0
• » - '
cu«a.
10090-80-70-60-5040302010-
•
\ \ r4 4 4 6 8Reading Passage Grade Level
Generalization
4 4 6 8 9Reading Passage Grade Level
Strategy Use - - » - - Comprehension
of comprehension questions answered correctly (squaresymbols) on each criterion-based comprehension test.(Each symbol represents performance on three passagesand three sets of questions.) The grade level at whicheach reading passage was written is shown along theX-axis of each graph. Student 3's data were graphedwith those of Students 4 and 5 because his graph part-ner moved from the school and could not continuewith the study. Likewise, Student 6's data were graphedwith those of Students 7 and 8 because his graph part-ner was excluded midway from the study due to a highabsentee rate.
The figures show that the percentage of comprehen-sion questions answered correctly increased with theonset of instruction and maintained a positive trendthrough the remainder of instruction and posttesting.During baseline, the students answered an average of31.74% of the comprehension questions correctly.During instruction, they answered an average of77.39'M) of the questions correctly; during the posttestcondition, they answered an average of 82% of thequestions correctly. Similar results were evidenced withregard to the Strategy Use Test. During baseline, thestudents earned an average of 0% of the points on the
Volume 30, Fail 2007 2SS
Strategy Use Test. During instruction, they earned anaverage score of 66.39%; during the posttest condition,they earned an average score of 81.94% on the StrategyUse Test.
Table 2 displays the results from the maintenancetests. On the left-hand side are listed the comprehen-sion (M = 41%, SD= 11.73) and strategy use percentagescores {M = 19%, SD = 14.19) earned by the six partici-pants who took a maintenance test eight months afterinstruction was completed without participating in areview (Maintenance Test 1). On the right-hand sideare the scores earned by the four participants who tookthe test 12 months after instruction was completed andafter a brief review (Maintenance Test 2). The scoresearned by the latter group of students suggest that witha brief review and practice, students were able to applythe steps of the strategy and answer questions correctly
at levels that were higher than during baseline and thatapproximated their posttest performance more closelythan when they took the test without a review.
A Friedman Test was conducted to evaluate differ-ences between the median for the Strategy Use Testscores during baseline (median = 0%), posttest (median= 82%), and maintenance without review (median = 19%) conditions. (Data from the maintenance with-review test were not included because of the low num-ber of subjects.) Significant differences were found, x^,{2,N = 6) = 11.565,/?<.01, and the Kendall coefficientof concordance (effect size index) of .964 indicatedstrong differences among the three median scores.
Follow-up pairwise comparisons were conductedusing a Wilcoxon Signed-Ranks Test (Siegel & Castellan,1998); the LSD procedure was used to control Type Ierrors across these comparisons at the .05 level. The
Iable 2Percentage Scores on the
student
1
2
3
4
5
6
7
8
Mean (SD)
Maintenance Tests
Maintenance Test 1Comprehension Score Strategy-Use Score
Without Review^ Without Review^
40
n/a
26
60
47
33
40
n/a
41 (11.3)
n/a=not available for testing; ^"-f>,''"-'
0
n/a
15
22
40
29
10
n/a
19 (14.19)
MaintenanceComprehension Score
With Review''
60
67
53
n/a
n/a
n/a
n/a
80
65 (11.52)
Test 2Strategy-Use Score
With Review*'
79
84
77
n/a
n/a
n/a
n/a
88
82 (4.97)
Learning Disability Quarterly 256
lable 3Percentage Scores on the Strategy
Student
1
2
3
4
5
6
7
8
Mean (SD)
•Significanl differt'nce found.
Knowledge Pretest
Pretest
0
0
0
0
0
0
0
0
0
and Posttest
Posttest
100
96
80
100
96
92
80
90
91.75* (8.031)
median score for the Strategy Use posttests was signifi-cantly higher than the median score for the baselinetests, p = .012, and it was also significantly higher thanthe median score for the maintenance without-reviewtests, p = .028. The median score for the maintenancewithout-review test was significantly higher than themedian baseline score, p = .043.
A Friedman Test was also conducted to evaluate thedifferences in the medians for the percentage of com-prehension questions answered correctly during thebaseline (median = 33.51 %), posttest (median =78.42%), and maintenance-without-review conditions(median = 41.00%). Significant differences were found,XH2,N = 7) = 12.00, p < .01, and the Kendall coefficientof concordance (effect size index) of 1.00 indicatedstrong differences among the three median scores.HolU)w-iip pairwise comparisons were conducted usinga Wilcoxon Signed-Ranks test (Siegel & Castellan, 1998);the l.Sn procedure was used to control for Type I errorsacross these comparisons at the .05 level. The medianscore on the posttest comprehension tests was signifi-
cantly higher than the median score for the baselinecomprehension tests, p = .012, and it was also signifi-cantly higher than the median score for the mainte-nance test, p = .028. The median maintenance score wassignificantly higher than the median baseline score,p = .027.Strategy Knowledge Test Results
1 able 3 lists the percentage scores earned by the eightstudents on the Strategy Knowledge pretest andposttest. As illustrated, each student earned a score of0% on the pretest (M = 0.00%, .SD = 0.00). Percentagescores on the posttest ranged from SO'Ki to lOO'Mi {M =91.75%, SD = 8.031). The Wilcoxon Signed-Ranks Test(Siegel & Castellan, 1988) was used to determinewhether there were significant gains from pretest toposttest on this measure; the LSD procedure was used tocontrol Type I errors across these comparisons at the .05level. A significant difference was revealed between thepretest and posttest scores, z = -2.530, p = .00. The effectsize for this gain was r - 0.99, representing a large gainaccording to Cohen (1988).
Volume 30, Fall 2007 257
Standardized Reading Test ResultsTable 4 lists the mean standard scores and grade-level
scores earned by students on the pretest and posttest onthe GRADE. The Wilcoxon Signed-Ranks Test revealed asignificant difference between the pretest and postteststandard scores, z = -2.521, p = .012. The effect size forthis gain was r = 0.91, a large gain according to Cohen(1988). This gain represents an average increase of 2.82grade levels (range = 1.4 to 3.6 grade levels) in readingcomprehension.
Satisfaction ResultsThe mean pretest and posttest reading satisfaction
ratings are reported in Table 5. As illustrated, signifi-cant differences were found between the pretest andposttest mean scores for the 7-point ReadingSatisfaction Questionnaire using a Wilcoxon Test, z =-2.524, p = 0.12. The effect size for this difference wasr = 0.95, a large gain according to Cohen (1988).
Table 6 displays the mean pretest and posttest ratingsand the standard deviations for each item on thequestionnaire. The items that received the highestposttest ratings related to enjoying the exercises andpassages while learning the strategy (M = 6.62) andusing the strategy to aid understanding in class(M = 6.37).Time Required for Instruction
Instructor time. The total amount of instructor timerequired to deliver the initial Inference Strategy instruc-tion ranged from 280 min. to 350 min. (M= 300 min.,or approximately 5 hours).
Student time. The total amount of student timeincluded time spent working with the instructor (seethe time reported above) and working independentlyon practice activities with reading passages. It rangedfrom 770 min. to 1040 min. (M = 905 min., approxi-mately 15 hours).
lable 4The GRADE
Student
1
2
3
4
5
6
7
8
Mean (SD)
Comprehension Standard Scores and Grade-Level Equivalents
GRADEPretest SS
55
74
55
72
66
55
58
55
61.25 (8.17)
•Sianjfii aiil difference fnunii.
GRADEPosttest SS
88
88
87
94
83
91
89
89
88.63' (3
Grade-LevelEquivalent Pre
2.5
3.9
2.3
3.8
3.3
2.7
2.9
2.5
.16) 2.99 (0.52)
Grade-LevelEquivalent Post
5.6
5.6
5.3
7.2
4.7
6.3
5.9
5.9
5.81 (0.73)
teaming Disability Quarterly 258
lable 5OveraU Mean Ratings for All Students on
Student
1
2
3
4
5
6
7
8
Mean (SD)
'Significant difference found.
the Reading
Pretest
1.1
1.6
1.5
2.3
1.5
1.8
1.3
1.3
1.55 (3.75)
Satisfaction Questionnaire
Posttest
5.8
4.9
5.1
5.5
4.4
6.4
6.5
6.6
5.65* (.815)
DISCUSSIONSeveral conclusions may be drawn from the results of
this study. First, instruction in the Inference Strategyincreased students' use of strategic skills to a masterylevel. Second, the instruction produced a positivechange in the students' ability to respond to inference-type questions on criterion-based tests. Increases instrategy use and ability to answer inferential questionswere found only after students participated in theInference Strategy instruction as demonstrated throughthe multiple-baseline design. Third, students' posttestscores on the GRADE were significantly higher thantheir pretest scores following instruction in theInference Strategy. Moreover, the GRADE scores indi-cated that, on average, the students made a 2.8 grade-level gain in reading comprehension within 15 hoursof instruction. The effect sizes related to the gains asso-ciated with all three of these major outcome measures
were large. During the instruction, the students gradu-ally worked up to reading passages written at theirgrade level and responded to associated inference com-prehension questions at mastery levels. They alsoearned scores above 90% on a test of their strategyknowledge at the end of the study.
ln addition, the results of the Student SatisfactionQuestionnaire suggest that students who participatedin the Inference Strategy instruction were more satis-fied at the end of the study than at the beginning of thestudy with how they felt about reading and differentcomprehension processes. On the one question relatedto their reaction to the instruction, the students indi-cated that they enjoyed the instruction (M = 6.62 on a7-point scale). Finally, instruction of the InferenceStrategy required five hours of initial instructionaldelivery and then another 10 hours of supervising prac-tice activities and providing feedback.
Volume 30, Fall 2007 259
Table 6Mean Pretest and Posttest Ratings, Standard Deviations, and Ranges for AllEach Item on the Reading and Strategy Satisfaction Questionnaire
Fill in the box that hest describes how you feel...
Reading in school is boring.
[ am a good reader in schooi.
Reading assignments in schooi confuse me.
I know what an inference in reading is.
Coming to a conciusion from what i read in classcan be difficult for me.
I know what steps 1 can take to malce meaningfrom what I read.
i enjoy reading in class.
I felt comfortable using the Inference Strategywhile 1 read in class.
1 feel that 1 can use the Inference Strategy to heipme understand what 1 read in class.
1 enjoyed the exercises and passages that wereused while learning the Inference Strategy.
* Represents a mean gain of at least one poinl.
MeanSD
Range
Mean.SD
Range
MeanSD
Range
MeanSD
Range
MeanSD
Range
MeanSD
Range
MeanSD
Range
MeanSD
Range
MeanSD
Range
MeanSD
Range
Pretest
6.120.835-7
2.001.071-4
6.371.064-7
1.370.52
1-2
6.500.765-7
2.060.64
1-3
2.251.281-4
1.0001
].(K)01
1.0001
Students on
Posttest
4.3752.39
1-7
3.63*2,64
1-6
6.001.191-7
4.75*2.25
1-7
5.371.603-7
4.56*1.922-7
3.75*2.12
1-6
5.87*1.553-7
6.37*0.526-7
6.62*0.526-7
Lcarniitg Disability Qiuirterly 260
Several limitations apply to this study. First, althoughthe sample size is adequate for the experimental designused, it was smail. Expanding the research design toinclude more classrooms would increase the sample sizeand, therefore, eliminate the problem of committing aType 11 error and making generalizations that are notbased on a sufficiently large sample of students.
Another possible limitation relates to the characteris-tics of the participants. They represent students withsevere reading comprehension deficits, beyond thosetypically evidenced in the population of students withlearning disabilities at large. How much instructionaltime would be required and the kinds of gains thatmight be made by students exhibiting lesser deficitsstill need to be determined. Another limitation is thattest scores on the students' decoding skills were notavailable. Thus, it is unknown whether they had higherlevels of decoding skills than comprehension skills,which enabled them to tackle the higher reading-levelpassages.
An additional concern is that the instruction wasprovided by a researcher. Whether other teachers canproduce the same types of reading gains is unknown.Further, expository passages were not included in theinstructional materials or the criterion-liased tests.Although they were included on the GR.ADE, how thestudents performed on those particular passages is notknown. Whether the strategy can be used successfullywitli expository passages remains to be determined.Ihis is an important consideration because expositorypassages represent a considerable portion of the mate-rials used in secondary content-area classes and areincluded within state and national assessments.Another concern is that the results were not disaggre-gated according to the type of question on the tests.Thus, at this time, whether the students respondedmore successfully to some types of questions than oth-ers is not known.
A final limitation relates to the maintenance data col-lected eight months after the initial intervention.While these data were collected on a relatively smallsubset of the entire sample, some trends require anexplanation. Specifically, performances on both strat-egy use and reading comprehension probes were rela-tively low. Educational research has shown that themaintenance of a targeted behavior as a result of anintervention has largely been an unrealized goal ofbehavioral interventions (McConnachi ii Carr, 1997).While features of the intervention itself need to he con-sidered as possible reasons for the low performance,other factors may be responsible. In this study, the stu-dents received no additional instruction in the use ofthe strategy nor prompts to use it after the study wascompleted. These maintenance data seem to under-
score the importance of continued practice and peri-odic maintenance probes if students are expected tocontinue to use a given strategy.
On the one hand, these data indicate that compre-hension strategies learned by students with disabilitieswill not continue to be an active part of their repertoireof reading behaviors unless students are given opportu-nities to practice the strategies over a sustained periodof time. On the other hand, these data are encouraging,because they clearly demonstrate that proficiency withthe targeted behavior can be quickly regained when arelatively modest review procedure is used. Clearly, theinstructional dynamics surrounding maintenance andsustained use of complex strategies such as theInference Strategy require considerably more study.
Research is needed to determine the effects ofInference Strategy instruction with larger, more di-verse, and more carefully described groups of students,and with teachers, as opposed to researchers, providingthe instruction. Maintenance procedures need to beexplored in more depth and detail. In addition, furtheranalysis of the data focusing on student performanceon each of the question types may provide helpfulinformation. Last, research is needed that focuses onuse of the strategy with expository as well as narrativepassages.Implications for Practice
The instruction described here for the InferenceStrategy has the potential of impacting education morebroadly in the following ways. First, the instruction canpotentially serve as a vehicle through which smallgroups of students with disabilities can be taught theinference skills required on today's standardized read-ing tests. As a result, schools will be better equipped tomeet increasing requirements and students' needs.Second, teachers may be able to combine this type ofinstruction with other forms of explicit reading com-prehension strategy instruction (e.g., see Schumaker &Deshler, 2006, for a review) because of similar instruc-tional design features. Third, the Inference Strategyinstruction might be used with student populationsrepresenting low socio-economic and ethnically diversegroups, as well as those who have severe readingdeficits.
REFERENCESAfflerbach. P. P. (1990). The influt'iice of prior knowledge on
expert readers' main idea construction strategies. ReiulhisRt'scdrch iluarttrly, 25(1), ;il-46.
.'\merican Institute for Research. (2005, Spring). Ri'iui'mg friimeworkfor the 2009 Natioiiiil Asse.ssnii'nt of Eiliuatioiuil Progress, pre-publication vtlition. Washington, DC: Author. Retrieved March . ,2006, from http://www.nagb.org/pubs/iiubs.hlml
Bulgren. J. A., Marquis, J. G.. Deshler. IX 1)., Schumaker, J, B., i tLenz. B. K. (2006). Instructional C{)nlcxt of inclusive secondary
Volume .?0, Fait 2007 261
general education classes: Teachers' instructional roles and prac-tices, curricular demands, and research-based practices and stan-dards . Learning Disahilities: A Contemporary journal, 4{\), 39-66.
Bulgren, J. A., Schumaker, J. B., & Deshler, D. D. (1988). Effective-ness of a concept teaching routine in enhancing the perform-ance of LD students in secondary-level mainstream classes.Learning Disability Quarterly, 11{\), 3-17.
Cain, K., SrOakhill,J. (1999). Inference making ability and its rela-tion to comprehension failure in young children. Reailiiig andWriting: All hitertlisdplinary loiimal, / / ( 5 ) , 489-503 .
Catts, H. W.. Fey. M. E., Tomblin, J. B., & Zhang, X. (2002). A lon-gitudinal investigation of reading outcomes in children withlanguage impairments. Journal of Speech. Language, and HearingResearch, 45, 1142-1157.
Center on Educational Policy. (2(K)5). High school exit exams. Policybrief 1, December. Retrieved April 1, 2006, from http://www.cep-dc.org/hlghschoolexit
Deshler, D. D., & Schumaker, J. B. (2006). Teaching adolescents withdisabilities: Accessing the general education curriculum. New York:Corwin Press.
Dewitz, P., Carr, E., is Patberg, J. (1987). Effects of inference train-ing on comprehension and comprehension monitoring. ReadingResearch Quarterly, 22(\), 99-121.
Fisher, J. B,, Schumaker, J. B., & Deshler, D. D. (2002). Improvingthe reading comprehension of at-rlsk adolescents, ln C. C. Block& M. Pressley (Eds.), Comprehension instnictioti: Research-basedbest practices {pp. 351-364), New York: Guilford.
Fritschmann, N. S. (2006). The effects of instruction in un inferencestrategy on the reading comprehension .skills of adolescents withteaming disabilities. Unpublished doctoral dissertation. Univer-sity of Kansas, Lawrence.
Fritschmann, N. S., Schumaker, J. B., & Deshler, D. D. (in prepara-tion). INFER: The inference strategy instnictors manual.
Gajria, M., & Salvia, J. (1992). The effects of summarizationinstructions on text comprehension of students. ExceptionalChildren, S8{6), 508-520.
Gersten, R., Fuchs, L. S., Williams, J. P., & Baker, S. (2001).Teaching reading comprehension strategies to students withlearning disabilities: A review of research. Review of EducationalResearch. 7/(2), 279-320.
Hansen, J., & Pearson, P. D. (1983). An instructional study:Improving the inferential comprehension ot good and poorfourth-grade readers, foumal of Educational Psychology, 75(6),821-829.
Heubert, J. P. (2002). Disability, race and high-stakes testing.Wakefield, MA: National Center on Accessing the GeneralCurriculum. Retrieved March 30, 2006, from http://www.cast.org/pubiications/ncac/ncac_disabnity.html
Horner, R. D., & Baer, D. M. (1978). Multiple-probe technique: Avariation of the multiple baseline. Journal of Applied BehaviorAnalysis, 11(1). 189-196.
Hughes, C. A., Deshler, D. D., Ruhl, K. L., & Schumaker, J. B.(1993). Test-taking strategy instruction for adolescents withbehavior disorders. Journal of Emotional and BehavioralDisorders, /(3), 188-199.
Kintsch, W. (1998). Comprehension: A paradigm for cognition. NewYork: Cambridge University Press.
Leach, J. M., Scarborough, H. S., & Rescoria, L. (2003). Late-emerg-ing reading disabilities. Journal of Educational Psychology, 95(2),211-224.
Mastropieri, M. A., Scmggs, T, E., & Graetz, J. E. (2003). Readingcomprehension instruction for secondary students: Challenges
for struggling students and teachers. Learning DisabilityQuarterly, 26(2), 103-116.
McConnachi, G., & Carr, E. G. (1997). The effects of child behav-ior problems on maintenance of intervention fidelity. BehaviorModification, 97(2), 123-158.
Partnership for 2Ist Century Skills. (2006). Learning for the 21stcentury. A report and mile guide for 21st century skills. RetrievedMarch 29, 2(X16, from littp://www.21stcentury skills.org.
Pressley, M. (2(K)0).What should comprehension instruction bethe instruction of? In M. L. Kamil, P. B. Mosenthal, P. D.Pearson, is R. Barr (Eds.), Handbook of reading reseurch, Volume U(pp. 545-562). Mahwah, NJ: LEA.
Reschly, D. J., Hosp, J. L., & Schmied, C. M. (2005). And miles togo ....• State SLD requirements and authoritative recommendations.Retrieved July 31, 2007, from www.nrcld.org/research/states/index.shtml
Schumaker, J. B., & Deshier, D. D. (2006). Teaching adolescents tobe strategic learners. In D. D. Deshler fet J. B. Schumaker (Eds.),Teaching adolescents with disabilities: Accessing the general educa-tion cuniculum (pp. 121-156). New York: Corwin Press.
Schumaker, J. B., Deshler, D. D., Bui, Y., & Vemon, S. (2006). Highschools and adolescents with disabilities: Challenges at everyturn. In D. D. Deshler *s[ J. B. Schumaker (Eds.), Teaching adoles-cents with disahilities: Accessing the general education curriculum(pp. 1-34). New York: Corwin Press.
Siegel, S. & Castellan, J. N. (1988). Nonpimimetric statistics for thebehavioral sciences (2nd ed.). Boston: McGraw-Hill.
Simmonds, E.P.M. (1992). The effects of teacher training andimplementation of two methods of improving the comprehen-sion skills of students with learning disabilities. LearningDisabilities Research ami Practice. 7(4), 194-198.
Snow, C. E. (2002). Reading for understanding: Toward a research anddevelopment program in reading comprehension. Santa Monica,CA: Rand.
Spargo, E. (1998). Timed readings in literature - Jamestown Series.Providence, RI: Glencoe/McGraw-Hili of McGraw-Hill Educa-tion.
Swanson, H. I.. Hoskyn, M., 8t Lee, C. (1999). Interwntions for stu-dents with learning di.sabilitie.'i: A weta-analysis of treatment out-comes. New York: Guilford.
Wagner, M., Marder, C, Blackorby, J., Camento, R., Newman, L,Levine, P., & Davis-Mercier, E. (with Chorost, M., Garza, N.,Guzman, A., & Sumi, C). (2003). The achievements of youth withdisabilities during secondary school. A report from the NationalLongiUuiinai Transition Study-2 (1^LTS2). MenIo Park, CA: SRIInternational. Retrieved April 3, 2006, from http://www.nlts2.org
Warner, M. M., Schumaker, J. B., Alley, G. R., & Deshler, D. D.(1980). Learning disabled adolescents in the public schools: Arethey different from other low achievers? Exceptional EducationQuarterly, 1{2). 27-35.
Williams, K. T. (2001). GRj\DE: Group Reading Assessment andDiagtiostic Evaluation. Shoreview, MN: Pearson AGS Globe.
Yuill, N., & Josceiyne, T. (1988). F.ffect of organizational cues andstrategies on good and poor comprehendcrs' story understand-ing. Journal of Educational Psychology, 80(2), 152-158.
Please address correspondence to: Nanette S. Fritschmann,School of Education, Lehigh University, Bethlehem, PA 18015;nfritschCnieh igh.edu
Imming Disability Quarterly 262
