Effects of accommodations on high-stakes testing for students with reading disabilities.

Effects of accommodations on high-stakes testing for students with reading disabilities. The effects of accommodations on the performance of students withdisabilities on accountability and other high-stakes tests have been thetopic of several recent reviews (Chiu & Pearson, 1999; Fuchs, Fuchs,& Capizzi, in press; Sireci, Li, & Scarpati, 2003; Thompson,Blount, & Thurlow, 2002; Tindal & Fuchs, 2000). These reviewslamented la��ment��ed?adj.Mourned for: our late lamented president.la��mented��ly adv. the relative dearth of empirical studies Empirical studies in social sciences are when the research ends are based on evidence and not just theory. This is done to comply with the scientific method that asserts the objective discovery of knowledge based on verifiable facts of evidence. of the effects ofaccommodations, noting that the research base was inconsistent and notadequate to support firm conclusions about the impact of accommodationson the test performance of students with disabilities. The lack ofconsistency across studies reflected the wide range of accommodationsevaluated in research, differences in implementation, and theheterogeneity het��er��o��ge��ne��i��tyn.The quality or state of being heterogeneous.heterogeneitythe state of being heterogeneous. of the students identified as disabled (Sireci et al.). The accommodation with the most support involved the provision ofextended time for students with disabilities. Unfortunately, extendedtime, particularly on timed reading tests, benefited students withdisabilities only slightly more than students without disabilities.Although some studies failed to demonstrate differential effects ofextended time for students with disabilities (Elliot & Marquart,2004), the meta-analysis Of Chiu and Pearson (1999) reported an effectsize of .37 standard deviations for students with disabilities that wasonly slightly higher than for students without disabilities (.30).Accommodations that benefit students with and without disabilities areof concern because when all students benefit from an accommodation andit is provided only to students With disabilities, the accommodationprovides an unfair advantage to those with disabilities. The scores ofstudents with disabilities may be optimal, but optimal scores may not bevalid indicators of the students' level of performance because theymask the true effects of the disability (Fuchs et al., in press). Another commonly utilized accommodation involves Some form of oralpresentation Of the material, including read aloud methods in which thedirections or passages are read to the student, methods in which thestudent reads orally, or methods that use an audiocassette or computerto facilitate read aloud administrations. It is not surprising thatdifferent methods of reading passages aloud to students withdisabilities involving reading improve their performance on readingcomprehension tests (Crawford & Tindal, 2004; Harker & Feldt,1993; McKevitt & Elliott, 2003; Meloy, Deville, & Frisbie,2000). All four of these experimental studies found that students withdisabilities obtained higher scores when the material was read to them,but students without disabilities also obtained higher scores. Althoughthe benefits of accommodations were stronger for students withdisabilities, reading a passage aloud to assess reading abilityinvalidates the test as it no longer assesses the construct of interest,typically reading comprehension. Crawford and Tindal noted specificallythat "reading a test aloud improves the scores of students ...regardless of their educational classification. One interpretation ofour results is that we changed the construct from reading comprehensionto listening comprehension." (p. 101) In contrast, reading the directions and word problems on a mathtest to students who are poor in reading does not invalidate in��val��i��date?tr.v. in��val��i��dat��ed, in��val��i��dat��ing, in��val��i��datesTo make invalid; nullify.in��val the test.Students with reading disabilities show a differential response to suchaccommodations On math assessments (Johnson, 2000; McKevitt &Elliott, 2003; Tindal; Heath, Hollenbeck, Almond almond,name for a small tree (Prunus amygdalus) of the family Rosaceae (rose family) and for the nutlike, edible seed of its drupe fruit. The "nuts" of sweet-almond varieties are eaten raw or roasted and are pressed to obtain almond oil. , & Harniss, 1998;Weston, 2003). Specifically, students Who are poor in reading, butproficient pro��fi��cient?adj.Having or marked by an advanced degree of competence, as in an art, vocation, profession, or branch of learning.n.An expert; an adept. in math, do better when the directions and word problems on amath test are read to them, and the effects are generally much largerthan for students without reading difficulties (Fuchs et al., in press),in this instance, the accommodation Compensates for the effect of thedisability that would otherwise serve as an irrelevant source ofvariance in the performance of students with disabilities because thatsame factor (reading ability) does not contribute to variance in theperformance of nondisabled students on a math test. By eliminating thisirrelevant source of variance in the scores of Students withdisabilities, a valid and optimal assessment Of their performance isobtained. A less frequently utilized accommodation that does not invalidate areading test allows students with reading disabilities to read thepassages aloud. This procedure may enhance comprehension by facilitatingmore efficient use of short-term memory, a problem experienced by manypoor readers (Swanson & Saez, 2003). Fuchs et al. (2000) found noeffects of accommodations involving large print or extended time incomparing Grades 4 to 5 students with and without learning disabilities(LD). In contrast, allowing students to read the passage aloud resultedin larger effects for students with LD than students without LD. Thisfinding was not replicated in a study of middle and high school students(Elbaum, Arguelles, Campbell, & Saleh, 2004), but this study did notspecifically identify the students as reading disabled and involvedreading of passages at a Grade 3 to 5 level because of floor effects onage-level passages. Elbaum et al., like Schulte, Elliott, andKratochwill (2001), observed that students with LD were much morevariable in their response to accommodations than comparison students,with both positive and negative responses that varied in magnitude. Perhaps the most consistent finding reported across studies andreviews examining accommodations for students with disabilities is theobservation that the students are heterogeneous. Simply definingstudents as learning disabled or as "students with adisability" without considering the area of academic disability maydilute di��lutev.To reduce a solution or mixture in concentration, quality, strength, or purity, as by adding water.adj.Thinned or weakened by diluting. the effect of an accommodation. The accommodation should bespecific to the type of academic disability. Thus, reading theinstructions on a math test often helps poor readers improve theirperformance, but should have little effect on good readers with poormathematical knowledge. Combining the performance of such students wouldreduce the effect of the oral administration accommodation on the mathtest (Weston, 2003). In a summary of the research on testingaccommodations, Thompson et al. (2002) reported that 31% of studiesdefined the sample as "learning disabled," 17% with"cognitive disability," 16% as "students withdisabilities," and only 9% as "reading or math disabled."This summary of the research suggests that many studies provideaccommodations without considering the relation of the accommodation tothe area of academic difficulty, which provides little information onthe specificity of the accommodation to the disability. Why would largeprint help students who do not experience diminished vision? The idea that valid accommodations should benefit only those with adisability has been termed the interaction hypothesis (Sireci et al.,2003). An accommodation should eliminate at least one source of variancethat is not fundamental to the underlying construct, affecting themeasurement of the construct in students with a disability, but not instudents without a disability or a different type of disability. Becausethe source of variance is fundamentally irrelevant to the measurement ofthe construct, a valid accommodation will improve performance only forstudents with a disability. Construct irrelevance ir��rel��e��vance?n.1. The quality or state of being unrelated to a matter being considered.2. Something unrelated to a matter being considered.Noun 1. can be demonstratedwhen an accommodation does not alter the construct validity construct validity,n the degree to which an experimentally-determined definition matches the theoretical definition. of the testor inflate inflate - deflate the student's score in a manner that is unfair to thosewho did not receive the accommodation. Providing larger text for astudent with low-level vision or Braille text for a student who is blindare classic examples of accommodations that would interact withdisability status. The present study addressed the interaction hypothesis byevaluating a package of accommodations specifically designed toaccommodate students with word decoding de��code?tr.v. de��cod��ed, de��cod��ing, de��codes1. To convert from code into plain text.2. To convert from a scrambled electronic signal into an interpretable one.3. problems. Students whose primaryarea of difficulty involve word recognition problems are oftenidentified as "dyslexic dys��lex��icor dys��lec��ticadj.Of or relating to dyslexia.n.A person affected by dyslexia. ," representing the commonly observedlink of word recognition difficulties with phonological pho��nol��o��gy?n. pl. pho��nol��o��gies1. The study of speech sounds in language or a language with reference to their distribution and patterning and to tacit rules governing pronunciation.2. processingproblems that eventuates in impairment Impairment1. A reduction in a company's stated capital.2. The total capital that is less than the par value of the company's capital stock.Notes:1. This is usually reduced because of poorly estimated losses or gains.2. in reading fluency flu��ent?adj.1. a. Able to express oneself readily and effortlessly: a fluent speaker; fluent in three languages.b. andcomprehension (Shaywitz, 1996). Although many students identified withdyslexia dyslexia(dĭslĕk`sēə), in psychology, a developmental disability in reading or spelling, generally becoming evident in early schooling. To a dyslexic, letters and words may appear reversed, e.g. have problems with other academic and cognitive skills, and inother domains of function (e.g., attention), the impact of wordrecognition difficulties on reading comprehension is well-establishedand affects many students with disabilities. Most high-stakesassessments attempt to evaluate reading comprehension because the goalof reading instruction is to develop the ability to understand text.Teaching students accurate and fluent fluent/flu��ent/ (floo��int) flowing effortlessly; said of speech. word recognition facilitatesaccess to the meaning of the passage. Although decoding problems clearlyinterfere with reading comprehension, it may be possible to comprehendwritten language adequately even when word recognition and fluencyskills are weak, so that accommodations for word recognition problemsmay be reasonable on a comprehension assessment. We hypothesized thataccommodations specifically designed to minimize the impact of wordrecognition difficulties on a high-stakes reading comprehension testwould improve the performance of students with word decoding problems,but would not improve the performance of students with average worddecoding ability. To evaluate this hypothesis, we compared the performance of Grade 3students identified by the schools with dyslexia who had clear evidenceof word decoding problems and average decoders on an accommodated andstandard administration of a version of the Texas Assessment ofKnowledge and Skills (TAKS TAKS Texas Assessment of Knowledge and Skills (statewide student assessment as of Spring 2003)) reading assessment, the Texas mandatedaccountability assessment. We recognized that the accommodations may bemore effective for students with stronger language skills or less severereading difficulties. Therefore, we hypothesized that the degree towhich a student benefited from accommodations would be more apparent inthose with stronger vocabulary development and higher levels of wordrecognition ability, utilizing measures of these skills as covariates inthe experimental design. METHOD PARTICIPANTS The students were third graders who were identified and served inpublic school programs specifically designed for students with dyslexia(n = 91) or were average readers (n = 91) from the same classroom as thestudents identified with dyslexia. As in most states, Grade 3 marks theinitial administration of the Texas state accountability assessment inreading and math, which is tied to statutory provisions designed toprevent social promotion. In Texas, school districts are required toimplement procedures for identifying and serving students with dyslexia.These students receive their primary reading instruction in the corereading-language arts block as well as supplementary reading assistancebased on their eligibility under Section 504 of the Rehabilitation rehabilitation:see physical therapy. Code.The guidelines guidelines,n.pl a set of standards, criteria, or specifications to be used or followed in the performance of certain tasks. are based on the 1994 research definition of dyslexiadeveloped by the International Dyslexia Association The International Dyslexia Association (IDA) is a non-profit education and advocacy organization dedicated to issues surrounding dyslexia.The International Dyslexia Association serves individuals with dyslexia, their families, and professionals in the field. that emphasizes theco-occurrence of word recognition and phonological problems, and theabsence of evidence for mental retardation mental retardation,below average level of intellectual functioning, usually defined by an IQ of below 70 to 75, combined with limitations in the skills necessary for daily living. and sensory disorders(Shaywitz, 1996). Districts vary widely in how the identificationprocedures and interventions are implemented, so we required that theparticipating students clearly demonstrate word recognitiondifficulties. We recruited from six suburban districts in southeast Texas thatmaintain programs for students with dyslexia. In these six districts, atotal of 48 schools and 113 teachers participated. Based on a poweranalysis designed to detect small effects of accommodations, oursampling plan was to recruit 100 students with word decoding problemsand 100 average decoders. Nine students with poor word decoding did notcomplete the study because the parents declined permission (n = 2),illness precluded participation (n = 1), the student transferred betweenscreening and the TAKS assessment (n = 4), or the school declinedpermission because it was felt the student was too impaired to take thetest (n = 2). The latter issue was raised more frequently, but we wereusually successful in suggesting that students at all levels of readingability should be included and that precautions would be taken to ensurethat the student did not find the experience humiliating hu��mil��i��ate?tr.v. hu��mil��i��at��ed, hu��mil��i��at��ing, hu��mil��i��atesTo lower the pride, dignity, or self-respect of. See Synonyms at degrade. . As the 91remaining students with poor decoding were matched with an averagedecoder A hardware device or software that converts coded data back into its original form. See decode and MPEG decoder. from the same classroom, we only needed 91 comparison students. The students were administered the Letter-Word Identification andWord Attack subtests of the Woodcock-Johnson III Tests of Achievement(WJ; Woodcock woodcock:see snipe. woodcockAny of five species (family Scolopacidae) of plump, sharp-billed migratory birds of damp, dense woodlands in North America, Europe, and Asia. , McGrew, & Mather, 2001). The administration of theseWJ subtests, which measure reading of real words and pseudowords inisolation, was undertaken to ensure that the student was impaired inreading decoding skills, a necessary criterion because of the variationin identification procedures and because some students showedsignificant improvement in reading skills after they began the readingprograms. Using grade norms, students were accepted into the group withpoor decoding only if they read below the 26th percentile on the BasicReading Skills cluster score of the WJ, a composite of the two subtestswith an internal consistency In statistics and research, internal consistency is a measure based on the correlations between different items on the same test (or the same subscale on a larger test). It measures whether several items that propose to measure the same general construct produce similar scores. reliability of .97. If a student hadreading decoding scores from school-based assessments that indicatedthat the student read above the 25th percentile either at the point ofentry into the program or after a period of intervention, that studentwas not further evaluated. Across districts, a total of 184 studentsidentified with dyslexia were eligible for screening; 100 met theeligibility criterion, with most ineligible in��el��i��gi��ble?adj.1. Disqualified by law, rule, or provision: ineligible to run for office; ineligible for health benefits.2. because of scores thatexceeded the WJ cut point. We were not able to further evaluate thestudents to determine whether they met research definitions of dyslexia,focusing instead on the presence of poor decoding skills. Students in the comparison group were nominated nom��i��nate?tr.v. nom��i��nat��ed, nom��i��nat��ing, nom��i��nates1. To propose by name as a candidate, especially for election.2. To designate or appoint to an office, responsibility, or honor. by teachers asaverage readers from the same classes as those identified with dyslexia.They had not been identified for special education or related services.These students had WJ reading scores that were in the average range:above the 39th percentile, but below the 76th percentile. A total of 219students were referred, and 131 obtained scores within the averagerange. Students selected were the first identified and closest in genderand ethnicity ethnicityVox populi Racial status–ie, African American, Asian, Caucasian, Hispanic . Table 1 presents demographic variables by group and condition. Thegroups were comparable in age and demographic characteristics with goodrepresentation of gender and different ethnicities. About half of eachgroup qualified for a free/reduced lunch. The groups with poor decodingwere about 6 months older than the groups of average decoders becausesome students had been retained. Virtually all poor decoders were servedthrough Section 504, with less than 27% (n = 24) with an additionalspecial education placement. We observed that districts were reluctantto identify students who were also identified for special education,expressing concern that the TAKS was not appropriate for students inspecial education. TEXAS ASSESSMENT OF KNOWLEDGE AND SKILLS The study was conducted independently of the state's mandatedassessment with a practice form of the Grade 3 TAKS developed in fieldtesting. No modifications were made of the TAKS booklets; the onlymodifications were in the instructions provided by the examiners. TheGrade 3 TAKS reading assessment involves a practice story and threestories of increasing difficulty. For each story, the student reads aparagraph that typically has a title and illustrative il��lus��tra��tive?adj.Acting or serving as an illustration.il��lustra��tive��ly adv.Adj. 1. pictures. Mostresponses are multiple choices, but are designed to access the literal In programming, any data typed in by the programmer that remains unchanged when translated into machine language. Examples are a constant value used for calculation purposes as well as text messages displayed on screen. In the following lines of code, the literals are 1 and VALUE IS ONE. meaning of the passage, vocabulary, and different aspects of criticalreasoning about the material in the paragraph. Both expository andnarrative materials are included. The TAKS is not timed, and studentsare typically allowed as much time as they need to complete theassessment. Like all TAKS tests, the Grade 3 reading comprehension assessmentis a criterion-referenced test A criterion-referenced test is one that provides for translating the test score into a statement about the behavior to be expected of a person with that score or their relationship to a specified subject matter. aligned with grade-based standards fromthe Texas Essential Knowledge and Skills (Texas Education Agency, 2004).Development was based on a large, interactive item development processthat included several thousand individuals involved in Texas education.Proposed items were field-tested and evaluated for reliability,validity, and bias by the contractor, Pearson Educational Measurement,through a process that involves embedding 1. (mathematics) embedding - One instance of some mathematical object contained with in another instance, e.g. a group which is a subgroup.2. (theory) embedding - (domain theory) A complete partial order F in [X -> Y] is an embedding if of future items in live testforms in random placement and spiraled across the state so that nostudent sees more than a fraction of the future items or knows whichitems are "live" and which items are under evaluation. In thisway, a new form of the test is developed each year under live conditionsfor use in the subsequent year. This process is necessitated by therequirement that the test be released each year (seewww.tea.state.tx.us/student.assessment/resources/release/taks/index.html for the actual 2003-2004 TAKS tests).The Grade 3 reading assessment is designed to assess basic understandingof content, application of knowledge and literary elements, use ofstrategies to analyze reading material, and critical thinking skills.The internal consistency (coefficient coefficient/co��ef��fi��cient/ (ko?ah-fish��int)1. an expression of the change or effect produced by variation in certain factors, or of the ratio between two different quantities.2. alpha) of the Grade 3 test is .89(Texas Education Agency). Validity focused on the content and thealignment of items to standards. Criterion-related validity studies arein progress as TAKS was first used in 2003-2004. However, the format forassessing reading comprehension in the TAKS is like that used in manyassessments with established construct validity. NATURE OF THE ACCOMMODATIONS A working group convened by the Texas Education Agency, whichincluded representatives of the state's three reading researchcenters, the curriculum division, the student assessment division, thestate dyslexia advisory committee, and the publisher, assessed theliterature on accommodations. The group agreed that accommodations thatcannot be readily implemented within a school would not be considered.The critical consideration for students with dyslexia was to minimizethe impact of their difficulties with word decoding. This groupdetermined that three accommodations might potentially reduce the impactof decoding problems on an alternative administration withoutinvalidating in��val��i��date?tr.v. in��val��i��dat��ed, in��val��i��dat��ing, in��val��i��datesTo make invalid; nullify.in��val the TAKS as an assessment of reading: (a) extending thetesting session into two blocks to complete the TAKS; (b) reading ofproper nouns to students; and (c) reading of stems and possibleresponses from the comprehension questions to students after they haveread the passage independently. The rationale behind these accommodations was as follows: When decoding words is difficult, students with readingdifficulties commonly experience fatigue and diminished interest duringthe task. Consequently, extending the testing session into two blocksmay be beneficial to counteract effects of fatigue and loss of interestbecause of word decoding difficulties. Proper nouns are often idiosyncratic id��i��o��syn��cra��sy?n. pl. id��i��o��syn��cra��sies1. A structural or behavioral characteristic peculiar to an individual or group.2. A physiological or temperamental peculiarity.3. , excessively frustrating frus��trate?tr.v. frus��trat��ed, frus��trat��ing, frus��trates1. a. To prevent from accomplishing a purpose or fulfilling a desire; thwart: topoor decoders, and minimally associated with understanding of themeaning of a passage. Reading proper nouns to students may reducereliance on decoding skills as well as re ducing frustration and fatigueand preventing loss of motivation without invalidating the test. The TAKS reading test is a comprehension assessment that determineshow well the student understands the passage that he or she reads. Tothe extent that the TAKS is intended to assess students' ability tounderstand the passages that have been read by asking them to respond tospecific comprehension questions, the determination of comprehension isnot dependent on the student's ability to read the comprehensionquestions. Although read aloud accommodations of the actual TAKSpassages were not allowed, read aloud accommodations of thecomprehension stems were allowed, thus reducing demands on word decodingskills. Although it would be interesting to determine which of theaccommodations were most effective, much of the literature does notsuggest that accommodations have any major effects on studentperformance. Therefore, the working group recommended that all threeaccommodations be provided in order to maximize the power of anaccommodated administration. PR0CEDURES Students within the poor decoding and average decoding groups wererandomly assigned to the accommodated and standard versions of the TAKS.To ensure that group size did not influence the results, all studentswithin a particular school and condition were tested in small groups bya trained examiner in a room provided by the school. The group sizesranged from 2 to 8 students and always included students of comparablenumbers from the poor decoding and average decoding groups, thuscontrolling for group size. For the accommodated administration, thepublisher of the TAKS modified the instructions and format so that thestandard version of the TAKS could be administered in two sessions. Inthe first session, the student completed the practice story and thefirst two stories in the test; in the second session, the third, mostdifficult, story was completed. Booklets and scoring sheets wereprovided and scored by the publisher. A data file was provided withappropriate student identifiers. These data were merged with demographicdata and performance data collected in the schools. For both groups, the TAKS was administered by trained examiners.Although we attempted to blind examiners to group identity, someexaminers had evaluated students for the study. However, the groups weremixed and administration was highly scripted. The administrationdirections were read exactly as they appeared in the manual. For thestandard administration, the examiner followed the manual, reading thedirections and monitoring the students, but not otherwise interactingwith the group. For the accommodated administration, the administratorfollowed guidelines developed for oral administration of the TAKS Grade7 mathematics test, reading only the proper nouns, comprehension stems,and possible answers. Students also had proper nouns booklets to readand follow along with the examiner. Students in both groups responded inthe standard manner by bubbling their responses in the TAKS bookletscantron, as oral responses would be disruptive in a small groupsetting. In addition to the administration of the TAKS, each studentreceived an individual assessment of oral language vocabulary using theWoodcock Language Proficiency Language proficiency or linguistic proficiency is the ability of an individual to speak or perform in an acquired language. As theories vary among pedagogues as to what constitutes proficiency[1], there is little consistency as to how different organisations Battery-Revised Picture Vocabulary subtest(Woodcock, 1991). This well-standardized assessment (internalconsistency reliability of .85) measures vocabulary by asking studentsto identify pictures and provide the name. The vocabulary assessmenthelped determine whether accommodations differentially affected TAKSreading performance depending on the students' level of generallanguage ability. In a series of confirmatory factor analytic Adj. 1. factor analytic - of or relating to or the product of factor analysisfactor analytical studies ofstudents in Grades 2 to 3, many with reading difficulties, Fletcher etal. (1996) found evidence for a general language factor that was bestindicated by assessments of vocabulary, including measures involvingconfrontation naming and receptive receptive/re��cep��tive/ (re-cep��tiv) capable of receiving or of responding to a stimulus. vocabulary. Morris et al. (1998)found subtypes of poor readers who varied in overall languagedevelopment. Some individuals with poor reading had a relativelyspecific reading disability, whereas others had word decoding problemsas part of a pervasive disturbance of oral language best indicated bylevel of vocabulary development. STATISTICAL ANALYSIS Statistical analyses were framed in order to estimate the effect ofaccommodations for students with decoding difficulties as well as toestimate the magnitude of the difference in the effect of accommodationsfor poor decoders compared with average decoders. The expectation thatvalid accommodations will eliminate a source of irrelevance in theperformance of students with disabilities presumes that the effect ofaccommodations will differ for groups of poor decoders and averagedecoders, that is, that group and administration condition willinteract. Thus, our primary analysis involved testing the group byaccommodation interaction. The other question asked whether the effect of accommodationsinteracted with covariates involving vocabulary and decoding ability.Because students were not selected on the basis of vocabulary scores,the interaction of vocabulary with accommodations was tested as both atwo-way interaction with accommodation and as a three-way interactionwith accommodation and group. In contrast, decoding was highlycorrelated cor��re��late?v. cor��re��lat��ed, cor��re��lat��ing, cor��re��latesv.tr.1. To put or bring into causal, complementary, parallel, or reciprocal relation.2. with group membership (point biserial correlation Noun 1. biserial correlation - a correlation coefficient in which one variable is many-valued and the other is dichotomousbiserial correlation coefficient of .89) asword reading ability served as one basis for defining group membership.Consequently, we examined the interaction of decoding ability withaccommodations only within the group with poor decoding. All analyses were run as mixed models using SAS (1) (SAS Institute Inc., Cary, NC, www.sas.com) A software company that specializes in data warehousing and decision support software based on the SAS System. Founded in 1976, SAS is one of the world's largest privately held software companies. See SAS System. PROC (language) PROC - The job control language used in the Pick operating system.["Exploring the Pick Operating System", J.E. Sisk et al, Hayden 1986]. MIXED. A mixedmodel is one that has both fixed effects and random effects Random effects can refer to: Random effects estimator Random effect model . A fixedeffect is estimated to be the same for all members of a group, whereas arandom effect permits members of a group to have varying values. In oursample, we had data at several levels of an organizational hierarchy, sothat students were nested within classrooms, within schools, withindistricts. Some of the variance in students' scores could beattributed to the placement of students in the same classroom, or thesame school, or the same district. Current statistical models can takeinto account the dependence of observations at multiple levels of such ahierarchy. Because our sample was comprised of only 1 to 6 students perteacher and only 2 to 12 students per school, we clustered schoolswithin districts and calculated variance due to dependence ofobservations in the same schools and in the same districts. Thus, themodel ignored clustering at the teacher level, but included clusteringat the school and district level in order to obtain more accuratestandard errors for estimates of the effects of accommodation. By usinga mixed model instead of a standard regression model, we could allowdifferent values for estimates at the school level, taking into accountthe fact that some schools are clustered together in the same district.Had clustering been ignored in the analysis, it is likely that standarderrors would have been underestimated, resulting in an inaccurate (i.e.,too liberal) test of the effects of accommodation. An analysis of covariance CovarianceA measure of the degree to which returns on two risky assets move in tandem. A positive covariance means that asset returns move together. A negative covariance means returns vary inversely. model was fit that predicted TAKSStandard scores from accommodation status, decoding status, vocabulary,and their interaction, allowing mean TAKS standard scores to vary acrossschools within districts. To allow for a dearer interpretation of themodel parameters, vocabulary scores were centered at the grand mean,which forces the differences among groups to be estimated at the grandmean value of the Covariate. On the basis of each student's TAKS' continuous score, itwas possible to determine whether the student had met the state'saccountability standard. After converting each student's continuousscore on the TAKS to Pass/Fail on the basis of the established cutpoint, we also conducted a log-linear analysis to evaluate the effectsof decoding status, accommodations, and their interaction on TAKS passrates. In this analysis, we modeled the natural logarithm Natural logarithmLogarithm to the base e (approximately 2.7183). of the odds ofpassing the test as a function of Vocabulary, Accommodation (Yes, No),Group (Poor decoding, Average decoding), and the interaction ofAccommodation and Group. This analysis was conducted using a standardlog-linear model rather than a generalized gen��er��al��izedadj.1. Involving an entire organ, as when an epileptic seizure involves all parts of the brain.2. Not specifically adapted to a particular environment or function; not specialized.3. linear mixed model (Breslow& Clayton, 1993; McCullagh & Nelder, 1989), the unifyingmodeling framework that extends mixed models to outcomes with non-normaldistributions, including rates and proportions. Because of the limitednumber of students per group, per condition, and per school on which toestimate a school-level odds ratio, we could not conduct this analysisto account for possible effects of nonindependence because of clusteringof students within the same school. RESULTS Table 2 provides means, standard deviations, and ranges by groupand administration condition for the Grade 3 TAKS reading test that arenot adjusted for the vocabulary covariate. As expected, the groups withpoor decoding had word reading scores that were at the 12th to 13thpercentile on average and well below the comparison groups, who wereslightly above average. Although Students were not selected on the basisof vocabulary development, the scores were comparable across groups andin the average range. The mixed model analysis of covariance showed that the Group XAccommodations interaction was statistically significant, F(1, 155) =12.04, p = .0007, d= .86. Regardless of Group, higher vocabulary scoressignificantly predicted higher TAKS performance (b = 4.17, SE = 0.76,F(1,173) = 29.87, p < .0001), but did not significantly interact withGroup or Accommodation, F(1,172) < 1. The results clearly showed thatstudents with poor decoding skills who received accommodations hadsignificantly higher TAKS reading scale scores (least squares M = 2059;SE = 19.1) than those with poor decoding who did not receiveaccommodations (least squares M = 1919; SE = 18.3), t(176) = 5.43, p< .0001, d = .91. In contrast, there was no significant differencebetween the performance of average decoders who received accommodations(least squares M = 2179; SE= 19.2) and those who took standard (leastsquares M = 2162; SE = 18.3), t(176) = 0.68, p = .50, d= .15)administrations of the TAKS reading test. Please note that the leastsquares means were adjusted for the vocabulary covariate and will notcorrespond to Table 2, which is based on actual student scores. Whether one takes the direct effect size calculation of theinteraction of d= :86, which is based on a pooled within-group standarddeviation using all four groups, or the difference in effect sizes forstudents with and without dyslexia (.91 - .15 = .76), which standardizesthe mean difference in each group based on the pooled within-groupstandard deviation for each group, the effect of accommodations forstudents with poor decoding relative to students with average decodingwas large. Both of these estimates of the effect sizes were computedwithout taking into account the clustering of students within schools inthe estimate of the standard deviation that goes into the denominator denominatorthe bottom line of a fraction; the base population on which population rates such as birth and death rates are calculated.denominatorofthe effect size estimate. If clustering effects were removed from thepooled within-groups standard deviation, the effect sizes would belarger, but would be less comparable to the standard effect sizesreported in the literature. Although the models were fit allowing for variability in meanperformance at the school level within districts, variance in interceptswas small relative to variability at the student level ([[tau].sup.2]schools = 1,052, p < .08; [[sigma].sup.2] students within schools =14,092; ICC ICCSee: International Chamber of Commerce = .07). Fitting the model under various alternativecovariance Structures (e.g., random effects at the school and districtlevel) did not alter conclusions. In each case, accommodations improvedperformance for children with poor decoding and had no effect onperformance for students with average decoding. Thus, statisticalconclusions about the effects of accommodation were unaffected byassumptions about the random effects structure of the data. To test for the possible moderating effect of decoding ability onaccommodations, we examined the relation of decoding ability andaccommodations within the group with poor decoding. The analysis showedsignificant effects of accommodations, F(1, 87) = 4.61, p < .04, butno interaction of WJ Basic Reading and accommodations, F(1, 85) < 1.Thus, Severity of decoding difficulties was not related to the effectsof the accommodations. Figure 1 plots the relation of the WJ BasicReading Skills and TAKS scale Scores in relation to the pass cut pointfor the TAKS. Although the lower end of the reading distribution (scoresbelow 80) was less well represented than scores closer to the averagerange, the lack of a relation is dearly depicted de��pict?tr.v. de��pict��ed, de��pict��ing, de��picts1. To represent in a picture or sculpture.2. To represent in words; describe. See Synonyms at represent. . The 133.6 higher unadjusted mean difference (see Table 2) in theTAKS scale score of children with poor decoding who receivedaccommodations translated into a significant increase in the TAKS passrate, with 18 of the 44 (41%) accommodated students with poor decodingachieving a passing score compared to 4 of the 47 students with poordecoding (9%) who received the standard administration. The logisticregression In statistics, logistic regression is a regression model for binomially distributed response/dependent variables. It is useful for modeling the probability of an event occurring as a function of other factors. analysis predicting the log odds of passing, that is, thenatural logarithm of percentage passing divided by percentage failing asa function of vocabulary, Accommodations, Group, and their interaction,indicated a statistically significant difference in the pass rate, WaldChi-square = 12.82, p < .0005, after controlling for vocabulary. Thisdifference translated into a roughly seven-fold increase in the odds ofpassing the test when students with word decoding problems were givenaccommodations (.41/.59 = .695 with accommodations versus .09/.91 = .099without accommodations). Accommodations had no effect on the pass rateof students in the comparison group, which ranged from 77% for thosereceiving accommodations to 83% for those not receiving accommodations. DISCUSSION We hypothesized that accommodations designed for students withdecoding difficulties would be associated with improved outcomes on ahigh-stakes Grade 3 reading comprehension test, yielding differentialbenefits for students with poor decoding skills when compared withaverage decoders. The findings clearly establish that the package ofaccommodations (increased time, reading of proper nouns, and reading ofcomprehension stems) resulted in Significant improvement in theperformance of poor decoders on the Grade 3 TAKS reading assessment. Thesignificant increase in the TAKS scale score translated to a significantdifference in the odds of passing for those with word decoding problemswho received accommodations. The effect size was in the large range forthe interaction of group and condition (.91) and for the simplecomparison of performance by accommodated and unaccommodated un��ac��com��mo��dat��ed?adj.1. Not adapted or accommodated: new arrivals who were unaccommodated to the heat of the tropics.2. students(.86) with Word decoding problems. The improvement could not beattributed to the student's level of vocabulary development or theseverity of reading decoding difficulties. There was no effect of accommodations on the performance of thecomparison group of average decoders, suggesting that the accommodationswere specific to the students with poor decoding. The absence of benefitto the average decoders indicates that the accommodations help optimizestudent performance without invalidating the test (Fuchs et al., inpress). The results also suggest that the modifications do not introduceconstruct-relevant variance to the test, but rather eliminateconstruct-irrelevant variance from the scores of students withdisabilities, thus assuring that the latent variable In statistics, Latent variables (as opposed to observable variables), are variables that are not directly observed but are rather inferred (through a mathematical model) from other variables that are observed and directly measured. , readingcomprehension, is more validly assessed for students with a disabilitywithout compromising the validity of the test for students with nodisability. The accommodations did not result in students with worddecoding problems obtaining scores that were comparable to the averagereaders, representing evidence for the weaker version of the interactionhypothesis as modified by Sireci et al. (2003). Under this modification,interactions of group and accommodation may be observed, but notnecessarily result in the accommodated group obtaining scores that arecomparable to typical achievers. More important is the evidence that theaccommodations benefited only students with word decoding problems. The literature offers different positions concerning how to studyaccommodations. Schulte et al. (2001) proposed that accommodationsalways be bundled because that is how accommodations are usuallyprovided in schools, whereas Thompson et al. (2002) suggested thataccommodations be bundled because single accommodations are likely to beweak in their impact on performance. In contrast, Fuchs et al. (inpress) argued that the effects of different accommodations should bestudied individually so that the contribution of each accommodation canbe assessed. The study was not designed to establish the differentialimpact of the three accommodations on student performance. Although thisquestion is interesting, it was actually unrelated to the rationale forthe design of the accommodations. When these accommodations wereformulated, they were selected to respond specifically to thedifficulties experienced by most individuals with word reading problemsWhen taking high-stakes assessment tests. The research currentlyavailable did not indicate that any single accommodation would haveenough of an impact to generate an effect that would be of a reasonablemagnitude and meet all of the criteria for an appropriate accommodation. Follow-up research could certainly unpackage the accommodations andevaluate each one in isolation. Given the heterogeneity in thepopulation of students with word reading difficulties, the possibilityexists that different components of the package are effective fordifferent students. Isolating the effects of individual components couldaid in the identification or design of more effective accommodations fordifferent test settings and different groups of students withdisabilities, as some components may be more generalizable gen��er��al��ize?v. gen��er��al��ized, gen��er��al��iz��ing, gen��er��al��iz��esv.tr.1. a. To reduce to a general form, class, or law.b. To render indefinite or unspecific.2. than others.However, the notion of identifying as many accommodations as possiblethat relate to the type of disability is a rationale for bundlingaccommodations. Although there is some utility in knowing if one or morecomponents of the package are responsible for the effect, unless thereis substantial financial or behavioral cost to the schools in providingdifferent components of the package, the answer to the question mayappear to be primarily of academic interest. In evaluating the accommodations, it is not likely that providingthe test in two sessions accounted for the effect that was observed. TheTAKS test is untimed and accommodations that extend the time typicallyare maximized on timed tests (Sireci et al., 2003). The only study thatisolated the effects of multiday assessments found no evidence of thebenefits of this accommodation (Walz, Albus, Thompson, & Thurlow,2000). To be certain of the effects of double versus single sessions onperformance for students with poor decoding, this study would need to berepeated with and without two sessions while retaining the otheraccommodations. There is no research that assesses the effects of reading propernouns and comprehension stems to students with reading difficulties. Ourrationale for reading proper nouns as an accommodation was that thesetypes of words can "slow down" the performance of studentswith poor decoding because they occur infrequently in��fre��quent?adj.1. Not occurring regularly; occasional or rare: an infrequent guest.2. and are often notdecodable. For students who have adequate decoding ability but poorcomprehension skills, it is not likely that these accommodations inisolation would make much difference in their level of performance onreading comprehension tests. The majority of the proper nouns were namesof individuals, towns, states, and countries. In evaluating the contentof the passages and comprehension questions on the TAKS, nocomprehension questions asked for information that involved knowledge ofthe proper nouns. As proper nouns tend to be content neutral,particularly names of persons, cities, and states, it seems doubtfulthat knowledge of proper nouns is likely to influence readingcomprehension directly other than reduce reliance on decoding skills,which can be very frustrating for students with poor decoding skills. Although not empirically discernable from this study, wehypothesize hy��poth��e��size?v. hy��poth��e��sized, hy��poth��e��siz��ing, hy��poth��e��siz��esv.tr.To assert as a hypothesis.v.intr.To form a hypothesis. that the largest impact from the accommodations was fromreading the comprehension stems. This accommodation reduced demands fordecoding and also introduced pacing. It is possible that comprehensionstems could be written in a way that provides dues to the meaning of thepassage. However, on this particular comprehension measure, thequestions are mostly multiple choices with a few fill in the blank. Avariety of skills involving knowledge of vocabulary, critical reasoning,and factual detail are queried. The stems themselves appear to beneutral, including questions such as "Paragraph 2 is mainlyabout," "In paragraph 5, the word considerate con��sid��er��ate?adj.1. Having or marked by regard for the needs or feelings of others. See Synonyms at thoughtful.2. Characterized by careful thought; deliberate. means," and"Josh wants to learn some words in Japanese to," all requiringmultiple choice responses. To establish that reading proper nouns andcomprehension stems eliminates construct-irrelevant variance from thescores of disabled readers, additional research could compare theperformance of students who receive these accommodations in isolation. These results are limited to students who have characteristicssimilar to those used to identify the participants for this study.Finding effective accommodations for students with word decodingdifficulties is important because this does not represent a small orunique group of students. It has been estimated that up to 92% ofstudents identified with learning disabilities have problems in the areaof reading (Kavale & Reese, 1992), and that 80% to 90% of thesestudents have word recognition problems (Lerner, 1989). Althoughstudents identified with learning disabilities in the area of readingmay have problems in addition to poor word recognition skills, it iscommonly the area of difficulty that results in identification. Althoughmost students in this study were not served in special education, it islikely that students in special education in the same age range withword recognition difficulties would benefit from these accommodations.It is less likely that such accommodations would be effective with olderstudents. If the gap between the comprehension assessment andstudents' sight word vocabulary increases, decoding the passagewill become increasingly difficult. Middle or high school students inspecial education with a level of decoding ability comparable to thestudents in this study may not benefit as much from this package ofaccommodations. Although we did not observe relations of the severity ofword decoding ability and response to accommodations, the number ofstudents who had severe word recognition problems below the 10thpercentile was small. Accommodations are reasonable in so far as they permit validassessments of student performance. Although more students would havepassed the state accountability test in Grade 3 under the accommodatedconditions in this study, the results should not mask the facts thateven students who "pass" still have potentially remedial REMEDIAL. That which affords a remedy; as, a remedial statute, or one which is made to supply some defects or abridge some superfluities of the common law. 1 131. Com. 86. The term remedial statute is also applied to those acts which give a new remedy. Esp. Pen. Act. 1. reading problems and that most accommodated students with word decodingproblems (59%) did not achieve a passing score. It is possible thatstudents who did not benefit had other reading difficulties or problemsoutside of reading that kept them from benefiting from theaccommodation. It would be helpful to assess more extensively studentson other reading skills and capabilities that might help explainindividual responses to the accommodations, although this issue does notlikely affect the results of this study given random assignment. Thecritical issue with any disability is the extent to which it hampers theoverall adaptive functioning adaptive functioning,n the relative ability of a person to effectively interact with society on all levels and care for one's self; affected by one's willingness to practice skills and pursue opportunities for improvement on all levels. of a person. If an accommodation allows aperson with a disability to pass a state accountability test, does thismean that the person is not disabled? In fact, the functionalimplications of passing or failing any high-stakes test has not beendemonstrated, so that this conclusion should not be made solely on thebasis of passing an accommodated state accountability test. Futureresearch should focus more specifically on the types of reading skillsrequired by different tests in students with a variety of readingdifficulties and across a broad age range. REFERENCES Breslow, N. R, & Clayton, D. G. (1993). Approximate inference (logic) inference - The logical process by which new facts are derived from known facts by the application of inference rules.See also symbolic inference, type inference. in generalized linear mixed models. Journal of the American StatisticalAssociation Established in 1888 and published quarterly in March, June, September, and December, the Journal of the American Statistical Association (JASA) has long been considered the premier journal of statistical science. , 88, 9-25. Chiu, C. W., & Pearson, P. D. (1999, June). Synthesizing theeffects of test accommodations for special education and limited Englishproficient students. Paper presented at the National Conference on LargeScale Assessment, Snowbird snowbird:see junco. , UT. Crawford, L., & Tindal, G. (2004). Effects of a read-aloudmodification on a standardized reading test. Exceptionality, 12, 89-106. Elbaum, B., Arguelles, M. E., Campbell, Y., & Saleh, M. B.(2004). Effects of a student-reads-aloud accommodation on theperformance of students with and without learning disabilities on a testof reading comprehension. Exceptionality, 12, 71-87. Elliott, S. N., & Marquart, A. M. (2004). Extended time as atesting accommodation: Its effects and perceived consequences.Exceptional Children, 70, 349-367. Fletcher, J. M., Francis, D. J., Stuebing, K. K., Shaywitz, B. A.,Shaywitz, S. E., Shankweiler, D. E, Katz, L., & Morris, R. (1996).Conceptual and methodological issues in construct definition. In G. R.Lyon & N. A. Krasnegor (Eds.), Attention, memory, and executivefunctions Executive functions is a term synonymous with cognitive control, and used by psychologists and neuroscientists to describe a loosely defined collection of brain processes whose role is to guide thought and behaviour in accordance with internally generated goals or plans. (pp. 17-42). Baltimore: Paul H. Brookes. Fuchs, L. S., Fuchs, D., & Capizzi, A. M. (in press).Identifying appropriate test accommodations for students with learningdisabilities. Focus on Exceptional Children. Fuchs, L. S., Fuchs, D., Eaton, S. B., Hamlett, C. L., Binkley, E.,& Crouch, R. (2000). Using objective data sources to enhance teacherjudgments about test accommodations. Exceptional Children, 67, 67-81. Harker, J. K., & Feldt, L. S. (1993). A comparison ofachievement test performance of nondisabled students under silentreading and reading plus listening modes of administration. AppliedMeasurement in Education, 6, 307-320. Johnson, E. (2000). The effects of accommodations on performanceassessments. Remedial and Special Education, 21, 261-267. Kavale, K. A., & Reese, L. (1992). The character of learningdisabilities: An Iowa profile. Learning Disability Quarterly, 15, 74-94. Lerner, J. (1989). Educational intervention in learningdisabilities. Journal of the American Academy of Child and AdolescentPsychiatry A branch of psychiatry that specialises in work with children, teenagers, and their families. HistoryAn important antecedent to the specialty of child psychiatry was the social recognition of childhood as a special phase of life with its own developmental stages, starting with , 28, 326-331. McCullagh, P., & Nelder, J. A. (1989). Generalized linearmodels (2nd cd.). New York New York, state, United StatesNew York,Middle Atlantic state of the United States. It is bordered by Vermont, Massachusetts, Connecticut, and the Atlantic Ocean (E), New Jersey and Pennsylvania (S), Lakes Erie and Ontario and the Canadian province of : Chapman & Hall. McKevitt, B. C., & Elliott, S. N. (2003). Effects and perceivedconsequences of using read-aloud and teacher-recommended testingaccommodations on a reading achievement test. School Psychology Review,32, 583-600. Meloy, L., Devine, C., & Frisbie, D. (2000, April). The effectsof a reading accommodation on standardized test scores of learningdisabled and non-learning disabled students. Paper presented for theNational Council on Measurement in Education Annual Meeting, NewOrleans New Orleans(ôr`lēənz –lənz, ôrlēnz`), city (2006 pop. 187,525), coextensive with Orleans parish, SE La., between the Mississippi River and Lake Pontchartrain, 107 mi (172 km) by water from the river mouth; founded , LA. Morris, R. D., Stuebing, K. K., Fletcher, J. M., Shaywitz, S. E.,Lyon, G. R., Shankweiler, D. P. et al. (1998). Subtypes of readingdisability: Variability around a phonological core. Journal ofEducational Psychology, 90, 347-373. Schulte, A. A., Elliott, S. N., & Kratochwill, T. R. (2001).Effects of testing accommodations on standardized mathematics testscores: An experimental analysis of the performances of students withand without disabilities. School Psychology Review, 30, 527-547. Shaywitz, S. E. (1996). Dyslexia. Scientific American Scientific AmericanU.S. monthly magazine interpreting scientific developments to lay readers. It was founded in 1845 as a newspaper describing new inventions. By 1853 its circulation had reached 30,000 and it was reporting on various sciences, such as astronomy and , 275, 98-104. Sireci, S. G., Li, S., & Scarpati, S. (2003). The effects oftest accommodation on test performance: A review of the literature.Center for Educational Assessment Research Report No. 485. RetrievedSeptember 9, 2004, from http://education.umn.edu/NCEO/Onlinepubs/TestaccommLitReview.PDF (Portable Document Format) The de facto standard for document publishing from Adobe. On the Web, there are countless brochures, data sheets, white papers and technical manuals in the PDF format. Swanson, H. L., & Saez, L. (2003). Memory difficulties inchildren and adults with learning disabilities. In H. L. Swanson, K. R.Harris, & S. Graham (Eds.), Handbook of learning disabilities (pp.182-198). New York: Guilford. Texas Education Agency. (2004). Technical digest 2002-2003. Austin,TX: Author. Retrieved September 9, 2004, fromwww.tea.state.tx.us/studentassessment/ resources/techdig/index.html Thompson, S., Blount, A., & Thurlow, M. (2002). A summary ofresearch on the effects of test accommodations: 1999 through 2001.(Technical Report 34.) Minneapolis, MN: University of Minnesota (body, education) University of Minnesota - The home of Gopher.http://umn.edu/.Address: Minneapolis, Minnesota, USA. ,National Center on Educational Outcomes. Retrieved September 9, 2004,from http://education.umn.edu/NCEO/OnlinePubs/ Technical 34.htm Tindal, G., & Fuchs, L. (2000). A summary of research on testchanges: An empirical basis for defining accommodations., Lexington:University of Kentucky Coordinates: The University of Kentucky, also referred to as UK, is a public, co-educational university located in Lexington, Kentucky. Mid-South Regional Resource Center,Interdisciplinary in��ter��dis��ci��pli��nar��y?adj.Of, relating to, or involving two or more academic disciplines that are usually considered distinct.interdisciplinaryAdjective Human Development Institute. Tindal, G., Heath, B., Hollenbeck, K., Almond, P.., & Harniss,M. (1998). Accommodating students with disabilities on large-scaletests: An experimental study. Exceptional Children, 64, 439-450. Walz, L., Albus, D., Thompson, S., & Thurlow, M. (2000). Effectof a multiple day test accommodation on the performance of specialeducation students. Minneapolis: University of Minnesota, NationalCenter on Educational Outcomes. Retrieved September 9, 2004, fromhttp://education.umn.edu/NCEO/Onlinepubs/ MnReport34.html Weston, T. J. (2003). The validity of oral accommodation intesting. (Working Paper No. 2003-06). National Assessment of EducationalProgress Validity Studies Panel. Retrieved September 9, 2004, fromhttp://nces.ed.gov/pubs2003/200306.pdf Woodcock, R. W. (1991). Woodcock Language ProficiencyBattery-Revised. Itasca, IL: Riverside. Woodcock, R. W., McGrew, K. S.,& Mather, N. (2001) Woodcock-Johnson III Tests of Achievement.Itasca, IL: Riverside. JACK M. FLETCHER (CEC (Central Electronic Complex) The set of hardware that defines a mainframe, which includes the CPU(s), memory, channels, controllers and power supplies included in the box. Some CECs, such as IBM's Multiprise 2000 and 3000, include data storage devices as well. TX Federation), Professor and AssociateDirector, Department of Pediatrics/Center for Academic and ReadingSkills, University of Texas Health Science Center at Houston. DAVID David, in the BibleDavid,d. c.970 B.C., king of ancient Israel (c.1010–970 B.C.), successor of Saul. The Book of First Samuel introduces him as the youngest of eight sons who is anointed king by Samuel to replace Saul, who had been deemed a failure. J.FRANCIS, Professor, Chair, and Director, Department of Psychology/ TexasInstitute for Measurement, Evaluation, and Statistics, University ofHouston. AMY A`my´n. 1. A friend. BOUDOUSQUIE, Research Associate, Department ofPediatrics/Center for Academic and Reading Skills, University of TexasHealth Science Center at Houston. KIM COPELAND, Assistant Professor,Department of Psychology/Texas Institute for Measurement, Evaluation,and Statistics, University of Houston. VICTORIA YOUNG, Director of TAKSAssessment, Student Assessment Division, Texas Education Agency, Austin.SHARON KALINOWSKI, Research Associate, Department of Pediatrics/ Centerfor Academic and Reading Skills, University of Texas Health ScienceCenter at Houston. SHARON VAUGHN (TX--Council for Exceptional Children)Professor and Director, Department of Special Education, Vaughn GrossCenter for Reading and Language Arts language artspl.n.The subjects, including reading, spelling, and composition, aimed at developing reading and writing skills, usually taught in elementary and secondary school. , College of Education, Universityof Texas-Austin. Address correspondence to Jack M. Fletcher, Department ofPediatrics/Center for Academic Reading Skills, University of TexasHealth Science Center at Houston, 7000 Fannin-UCT 2478, Houston, TX77030. (e-mail: Jack.Fletcher@uth. tmc.edu) This research was supported in part by funds from the TexasEducation Agency, Region X Education Service Center, and NICHD NICHDNational Institute of Child Health and Human Development. grant P50HD25802, Center for Learning and Attention Disorders. The authors wish to thank the superintendents, principals, andGrade 3 teachers of the school districts for their participation in thisstudy, especially the dyslexia coordinators in each district: RoseanneJoseph, Ed Mills, Pauline Vetor, Karen Coffey, Ricki Fischer, and TeresaRinglein; Pearson Educational Measurement, especially KimberlyO'Malley, for preparing and scoring the test; Sandy Maddox of theRegion X Education Service Center; Ed Miller of the Texas EducationAgency for helpful comments on the design; and Rita Taylor formanuscript preparation. Manuscript received September 2004; accepted December 2004.TABLE 1Age, Gender, Ethnicity, and Free Lunch Status for Dyslexic andComparison Group by Condition Dyslexic Comparison Accomodated Standard Accomodated StandardN 44 47 44 47Age (months) M 115.3 115.9 110.6 111.2 SD 6.4 7.4 4.5 4.9Gender N (%) Female 21 (48) 22 (47) 20 (46) 22 (47)Ethnicity White 17 (39) 22 (47) 15 (34) 19 (40) African-American 10 (23) 6 (13) 10 (23) 7 (15) Hispanic 16 (37) 18 (38) 17 (39) 21 (45) Asian/Other 1 (2) 1 (2) 2 (5) 0 (0)Free Lunch Status (a) N 43 47 35 38 N (%) 23 (54) 23 (49) 17 (49) 20 (53)(a) Data on free lunch status was missing on 19 students.TABLE 2Means and Standard Deviations for the WJIII Basic Reading StandardScore, WLPB-R Picture Vocabulary Standard Score, and Grade 3 TAKSReading Scale Score for Dyslexic and Comparison Group by Condition Dyslexic ComparisonVariable Accommodated Standard Accommodated StandardN 44 47 44 47Basic ReadingSkills Cluster (a) M 83.6 81.8 102.2 102.7 SD 6.0 6.7 3.0 4.1 Range 62-89 55-90 98-109 96-110Picture Vocabulary (a) M 92.0 94.6 95.4 95.0 SD 12.1 14.2 12.2 10.3 Range 66-132 61-138 71-134 75-122TAKS Reading M 2055.3 1921.7 2184.8 2166.7 SD 162.5 132.3 122.0 116.0 Range 1838-2588 1745-2309 1944-2460 1911-2588Note. WJ = Woodcock-Johnson; WLPB-R = Woodcock Language ProficiencyBattery-Revised; TAKS = Texas Assessment of Knowledge and Skills.(a) Mean (SD) = 100 (15).