The Use of Computer-Assisted Instruction to Improve the Reading Comprehension of Students With Learning Disabilities: An Evaluation of the Evidence Base According to the What Works Clearinghouse Standards

Abstract

The purpose of this study was to investigate the evidence base for using computer-assisted instruction (CAI) to improve the reading comprehension of students with learning disabilities (LD). Twelve peer-reviewed studies (seven comparison group studies, five single-case studies) met selection criteria and were evaluated according to the relevant What Works Clearinghouse (WWC) procedures and standards. Results showed that seven studies (five comparison group and two single-case studies) met WWC standards with or without reservations. Key instructional features employed in CAI studies meeting the WWC standards without reservations included practice opportunities, self-correction and immediate corrective feedback, teacher-directed instruction, and contingencies for enhancing student motivation and engagement. Implications for future research and suggestions for using quality indicators to improve the rigor of future CAI investigations are discussed.

Keywords

WWC standards reading comprehension computer-assisted instruction (CAI)learning disabilities

The ultimate goal of reading instruction is to help students acquire the skills necessary to make meaning and extract useful knowledge from text, and the extent to which this happens is dependent upon a student’s reading comprehension ability (Allington, 2012; Reutzel & Cooter, 2013). Reading comprehension is essential because it is not only the basis of understanding a variety of subjects taught in school but it is also the basis for acquiring important information in their lives (Vacca, Vacca, & Mraz, 2011). The importance of competent reading comprehension skills is highlighted in the Common Core State Standards (CCSS; 2010), which require students to read increasingly complex texts, encounter different text structures, and develop content area knowledge through reading. Unfortunately, many students across the nation continue to struggle with reading comprehension as approximately one third of fourth graders are below basic readers (Kena et al., 2016).

For students with learning disabilities (LD), reading comprehension is a major concern as reading is the most prevalent subtype of specific learning disabilities (SLD; National Center for Learning Disabilities, 2014). Approximately 2.4 million students are identified with LD, 66.2% of whom spend at least 80% of the school day in general education classrooms (Snyder & Dillow, 2015). Students with LD often face a significant challenge meeting grade-level expectations due to limited cognitive and language abilities (Shaywitz & Shaywitz, 2005, 2008; Wise, Sevcik, Morris, Lovett, & Wolf, 2007). If the challenge is not overcome in the elementary grades, students may struggle in secondary level general education content courses (Stetter & Hughes, 2011). Thus, when teaching students with LD, it is imperative that teachers employ evidence-based practices that develop reading comprehension skills to promote school and transition success.

CAI as a Promising Reading Practice for Students With LD

Several investigations have identified effective practices for improving the reading comprehension skills of students with LD, such as questioning strategies and strategy instruction related to summarization and main idea identification (Sencibaugh, 2007; Solis et al., 2012; E. Swanson, 2008). However, students with LD continue to struggle with reading comprehension despite the preponderance of information on effective methods to remediate these difficulties and subsequent efforts to disseminate this information (see McKenna, Shin, & Ciullo, 2015).

CAI may be a promising practice for addressing reading comprehension difficulties. CAI is engaging, permits students to work independently, and provides teachers with flexibility to design a lesson plan, particularly for Tier 2 or Tier 3 interventions (Shores & Chester, 2008). In addition, CAI may be beneficial for students with LD (King-Sears et al., 2015) and students with poor motivation and organizational skills (Stetter & Hughes, 2011). However, CAI should be combined with effective teaching strategies and designed to be sensitive to students’ learning needs (King-Sears & Evmenova, 2007).

For the purposes of this review, we define CAI as, “a method of instruction in which there is a purposeful interaction between a learner and the computer device for helping the individual learner to achieve the desired instructional objectives” (Beichner & Schwartz, 1999, p. 540). In this current review, we focused exclusively on laptop and desktop computers for the following reasons. First, unlike tablets or mobiles that have relatively shorter history of use and therefore a small corpus of studies to draw from, CAI studies that employ laptop and desktop computers can include several instructional features found to be effective. Second, considering that CAI has been researched for more than 30 years (e.g., Harper & Ewing, 1986), a review focusing exclusively on laptop and desktop computers could potentially identify promising practices to address the reading comprehension difficulties of students with LD.

Instructional Features of CAI

CAI includes many instructional features that may be beneficial to text-based reading for students with LD. CAI via electronic text enables students to read text on a computer screen, which may enhance student motivation (Cullen, Alber-Morgan, Schnell, & Wheaton, 2014) and improve reading performance (Cullen et al., 2014; Harper & Ewing, 1986). CAI may also include computer-based speech technology (speech-to-text or text-to-speech software; E. L. Higgins & Raskind, 2004; Raskind & Higgins, 1999), which can be used to provide additional opportunities for intense practice. With speech technology, teachers can ask students to follow along, read aloud stories, and provide decoding support from digitized speech. Furthermore, CAI provides immediate feedback, which may enable students with LD to learn independently (Cullen et al., 2014; E. L. Higgins & Raskind, 2004; Jiménez et al., 2003); built-in feedback enables students to detect errors immediately and recognize correct responses. CAI may also include multimedia such as video that facilitates students’ understanding and maintains their attention (Fitzgerald, Miller, Higgins, Pierce, & Tandy, 2012). Electronic dictionaries may also be incorporated into CAI approaches, which offer choices of words and their definitions. An audio component speaks the word and its meanings, providing an opportunity to students to choose the correct word or definition (Moore, 2009). Finally, when reading strategies are combined with CAI, the strategies can be taught and practiced using the multiple functions (e.g., electronic dictionary, multimedia, immediate feedback) that computer software provides (Stetter & Hughes, 2011).

Syntheses on CAI Approaches for Students With LD

Three comprehensive syntheses have investigated the effects of using computers and technology with students with disabilities (Hall, Hughes, & Filbert, 2000; MacArthur, Ferretti, Okolo, & Cavalier, 2001; Stetter & Hughes, 2010). Hall et al. completed a synthesis of CAI interventions with reading outcomes (e.g., word recognition, vocabulary, comprehension) for students with LD. Stronger effects for improving reading skills were found in studies that provided explicit and systematic instruction with elaborate feedback and included correction procedures based on students’ learning needs and performance level. Among the 17 studies that met study selection criteria, only four included reading comprehension as a dependent variable. Furthermore, the researchers reported no specific findings related to the effects of CAI on reading comprehension.

MacArthur et al. (2001) reviewed technology-assisted instruction (e.g., CAI, synthesized speech feedback, electronic texts) interventions for developing the literacy skills (word identification, text comprehension, and writing) of students with mild disabilities and reading difficulties. Forty-seven studies met study selection criteria, 13 (28 %) of which included a reading comprehension dependent variable. Overall, studies focused on the use of electronic texts with a variety of enhancements (e.g., speech technology, definitions, graphics, and supplementary text) to compensate for poor reading skills such as word identification and vocabulary knowledge. Seventy-seven percent of studies with a reading comprehension dependent variable used researcher-developed or teacher-made assessments (e.g., comprehension questions, daily knowledge quizzes). Twenty-three percent of the studies investigating effects on reading comprehension used standardized tests to measure student outcomes. No statistically significant effects were demonstrated on either word identification or reading comprehension.

Stetter and Hughes (2010) completed a synthesis investigating the effects of CAI on the reading comprehension of students with disabilities and struggling readers. Of the 26 studies that met study selection criteria, standardized measures were used in only 35% (n = 9) of studies. The majority of studies used nonstandardized tests such as daily comprehension quizzes and oral retells. Overall, CAI demonstrated positive effects on reading comprehension; greater gains were demonstrated from studies that paired synthesized speech with additional supports compared with studies with synthesized speech only. Although the synthesis by Stetter and Hughes represents a significant contribution to the field, this synthesis did not specifically focus on students with LD. Furthermore, none of the aforementioned syntheses evaluated study rigor according to an established set of quality indicators (e.g., What Works Clearinghouse [WWC] standards, 2011).

Study Purpose

Recent syntheses on CAI approaches for students with LD found that CAI is a promising practice for improving reading comprehension. However, given the rapid growth of using instructional practices combined with computers (e.g., Hall et al., 2000; Regan, Berkeley, Hughes, & Kirby, 2014; Stetter & Hughes, 2010), the priority of reading comprehension, and the difficulties students with LD encounter in this era of high-stakes tests and increased academic demands, it is critical to identify evidence-based practices for improving reading comprehension. In 2014, the WWC released procedures for evaluating the rigor of research studies to inform future research and to assist with the identification of evidence-based practices. This current review of the literature employed the WWC standards (Institute of Education Sciences, 2014) to CAI intervention research for students with LD to identify trends in research, promising practices for improving reading comprehension, and areas for future research. This review of the literature was guided by the following research question: Is CAI an evidence-based practice for improving the reading comprehension skills of students with LD according to the WWC standards?

Method

Study Selection Criteria

Studies meeting the following criteria were included in the current study. First, only intervention studies with an experimental, quasi-experimental, or single-case research design were included. For experimental or quasi-experimental studies, only those that (a) included at least one treatment and one control or comparison group and (b) reported effect sizes or sufficient data to calculate them were considered “eligible” for evaluation according to quality indicators. Second, studies were published in a peer-reviewed journal in English between 1980 and April 2015. Unpublished empirical studies (e.g., book chapters, conference papers, dissertations, research reports, theses) were not included (a) to avoid possible overlap and (b) to include only studies that were rigorously evaluated by professionals in the field of special education. Third, studies included a CAI reading intervention for K–12 students with LD. The independent variable was considered a relevant CAI intervention when the study included a reading intervention provided through desktop or laptop computers and the intervention focused on an interaction between student participants and learning content presented by those devices. In some studies, LD was labeled as learning disabled (e.g., Bahr, Kinzer, & Rieth, 1991; Harper & Ewing, 1986; Keene & Davey, 1987); for those studies, authors reviewed the reported definition and identification procedures for LD to reach a final decision for inclusion. Only studies that indicated that students had an LD were included. Studies were excluded if data for students with LD were not disaggregated. Fourth, studies had to include at least one reading comprehension dependent measure. For the current review, reading comprehension was defined as the ability to understand written text, which provides a basis of understanding the meaning in content students are taught in schools, in addition to a process of acquiring important information based on students’ knowledge and experience (Duke & Carlisle, 2010).

Search Procedures

Studies published from 1980, when CAI research for students with disabilities emerged, to 2015 were included in this review. First, an electronic search using ERIC, PsycINFO, and Academic Search Premier (EBSCO) was completed using all possible combinations of the following search terms: computer-assisted instruction, computer-based instruction, technology-assisted instruction, CAI, CBI, computers, special education, learning disabilities, reading, and reading comprehension. Second, the abstracts of studies identified in the electronic search were reviewed to determine if they met inclusion criteria. Third, the first and the second authors conducted a hand search of the reference lists of previously identified CAI intervention studies and comprehensive syntheses (Hall et al., 2000; MacArthur et al., 2001; Stetter & Hughes, 2010).

The following procedures were used to identify studies germane to this review. First, the titles and abstracts for each study were read and examined in relation to the selection criteria. After this initial examination, 54 studies were selected for further scrutiny. Next, each study was read in its entirety to determine if it met selection criteria, resulting in a corpus of 12 studies (seven experimental or quasi-experimental, five single-case). Studies were commonly excluded for the following reasons: (a) the absence of a reading comprehension measure (e.g., Bryant et al., 2015), (b) the use of a technology-based intervention that did not include a computer (e.g., Xin & Rieth, 2001), (c) the absence of students with LD (e.g., Clarfield & Stoner, 2005; Solan, Shelley-Tremblay, Ficarra, Silverman, & Larson, 2003), (d) use of an ineligible design, and (e) the absence of disaggregated data (e.g., Kim, Samson, Fitzgerald, & Hartry, 2010; Montali & Lewandowski, 1996). Based on the inclusion and exclusion criteria, we identified 12 studies to be analyzed for the current review.

Descriptive Coding

Six characteristics were coded for each study: (a) type of design (single-case, experimental, quasi-experimental), (b) participant characteristics (age, grade level, gender, ethnicity), (c) intervention description (type of CAI, length and duration of intervention, instructional setting, interventionist, measurement, data collection reliability, and integrity), (d) student training procedures, (e) results, and (f) social validity. The presence of the following CAI features was also coded: (a) computer-based electronic text (text on computer screen), (b) computer-based speech technology (e.g., text-to-speech, speech recognition), (c) computer-based supported text (electronic text with additional features such as vocabulary support through electronic dictionaries, multimedia, immediate feedback etc.), or (d) reading strategy instruction (CAI with a reading strategy embedded) based on CAI features indicated in previous CAI syntheses (Hall et al., 2000; Stetter & Hughes, 2010).

WWC Coding

Using a coding sheet adapted from a previous study (Kang, McKenna, Arden, & Ciullo, 2015), studies were coded according to the WWC Procedures and Standards Handbook Version 3.0 (Institute of Education Sciences, 2014). According to this procedure, studies may receive one of three ratings: (a) Meets WWC Design Standards Without Reservations, (b) Meets WWC Design Standards With Reservations, and (c) Does Not Meet WWC Design Standards. Studies that employ a randomized control group design are evaluated according to the degree to which sample attrition is low and baseline equivalence (i.e., no statistically significant difference between the groups) is established. Randomized controlled trials with low attrition are rated as “Meeting WWC Group Design Standards Without Reservations” and those with high sample attrition are rated as “Meeting WWC Group Design Standards With Reservations” or “Not Meeting WWC Group Design Standards” depending on baseline equivalence (no statistically significant difference between the groups). To complete this analysis, author queries were conducted for studies in which information on attrition was not reported (K. Higgins, Boone, & Lovitt, 1996; Keene & Davey, 1987; Raskind & Higgins, 1999; H. Swanson & Trahan, 1992; see Note in Table 3).

Comparison group studies that did not include random assignment are rated as “Meeting WWC Group Design Standards Without Reservations” or “Not Meeting WWC Group Design Standards” depending on baseline equivalence. For example, quasi-experimental design studies with no statistically significant differences between groups at baseline receive a rating of “Meeting WWC Group Design Standards With Reservations.” Quasi-experimental studies with statistically significant differences between groups at baseline are rated as “Not Meeting WWC Group Design Standards.”

Single-case studies were evaluated according to the following four WWC standards: (a) the independent variable was systematically manipulated by the researcher, (b) interassessor agreement was collected for at least 20% of the data points across phases and met reliability thresholds, (c) the design provided at least three opportunities to demonstrate an effect at three different points in time, and (d) the degree to which study phases met requirements for the minimum number of data points. For example, multiple-baseline design studies with at least six phases with five points in each phase could potentially meet design standards “without reservations,” depending on the degree to which the study met other criteria. Multiple-baseline designs with three to four data points for at least six phases could potentially meet design standards “with reservations” and studies with less than six phases and less than three data points in at least one phase would “not meet design standards.”

Interrater Agreement

A multistep procedure was followed to establish interrater agreement for descriptive and WWC coding. First, the researchers independently coded one randomly selected comparison group study for training purposes and to establish initial reliability. All areas of disagreement were discussed until 100% agreement was obtained. Second, a similar procedure was followed with a randomly selected single-case design study. Third, upon the establishment of initial reliability with a comparison group and a single case, all remaining studies were independently double coded for descriptive information. All discrepancies in coding were discussed and resolved for each study. The overall interrater agreement was 100% for descriptive information across the 12 studies.

For WWC standards, all studies were double coded by the first and second authors. Overall initial agreement was 99% with 100% agreement for comparison group studies and 98% agreement for single-case studies. All discrepancies were discussed until 100% agreement was obtained.

Results

Overview of Selected Studies

A total of 299 students participated in the 12 studies included in this review. The mean number of students per study was 24.9 (SD = 21.5) and ranged from 3 to 60. All participants were school-aged students ranging from Grades 1 to 12. Students were reported as having an LD in all studies and the primary focus was improving reading comprehension. Nine studies (75%) included secondary school students, five (42%) included students in elementary grades, and two studies (E. L. Higgins & Raskind, 2004; Raskind & Higgins, 1999) included both elementary and secondary grade students.

Of the 12 studies that met inclusion criteria, five studies (42%) examined electronic text, and three studies (25%) included speech technology. Supported text and computer-based strategy instruction was examined in two studies each (17%). Regarding the type of CAI employed by grade level, electronic texts were employed with elementary and secondary school students. All speech technology studies and studies with strategy instruction were conducted with secondary grade students. Elementary and secondary grade students were included in studies investigating supported text (e.g., vocabulary support through electronic dictionaries, multimedia, immediate feedback) interventions. Tables 1 and 2 provide a summary of study characteristics.

Table 1.

Summary of Experimental or Quasi-Experimental Studies.

Study	Participants	Group comparison	n	Types of CAI	Effect size d
Bahr, Kinzer, and Rieth (1991)	9th–12th	[1] CAI with trained teacher (n = 19) vs. CAI with untrained teacher (n = 29); [2] CAI in heterogeneous group (n = 19) vs. CAI in homogeneous group (n = 29)	48	Electronic text	[1] Literal comprehension (ES = −.33), inferential comprehension (ES = −.41), and reading level (Bader) (ES = .17) [2] literal comprehension (ES = −.31), inferential comprehension (ES = −.22), and reading level (ES = −.36)
K. Higgins, Boone, and Lovitt (1996)	9th grade; mean 14.6 years old	[1] CAI (n = 2) vs. non-CAI (traditional lecture, n = 6); [2] CAI + Lecture (n = 5) vs. non-CAI (traditional lecture only)	13	Supported text	[1] ES = 72.69 [2] ES = 235.26
E. L. Higgins and Raskind (2004)	Aged 8 to 18 (mean 12.2)	[1] CAI with speech synthesis (n = 28) vs. non-CAI group (n = 16); [2] CAI without speech synthesis (n = 28) vs. non-CAI group (n = 16)	44	Speech synthesis	[1] 0.90 [2] 1.09
Jiménez et al. (2003)	103.6 months (8.6 years old)	CAI for students with dyslexia (n = 14) vs. non-CAI for students with reading difficulties (n = 28)	14	Supported text	ES = 0.70
Keene and Davey (1987)	9th–12th	CAI (electronic text) (n = 25) vs. non-CAI (printed text) (n = 26)	51	Electronic text	In the immediate testing (ES = 0.06) In the delayed testing (ES = 0.30)
Raskind and Higgins (1999)	12.9, and ranged from 9.3 to 17.7	CAI with speech synthesis (n = 19) vs. CAI without speech synthesis (n = 20)	39	Speech synthesis	ES = 0.76
H. Swanson and Trahan (1992)	Grades 4–6	CAI (no reread [n = 15] and reread [n = 15]) vs. paper (n = 15)	60	Electronic text	Nelson reading (ES = 0.08), traditional questions (ES = −0.18), and cloze questions (ES = 0.02)

Note. CAI = computer-assisted instruction; ES = effect size.

Table 2.

Summary of Single-Case Design Studies.

Study	n	Participants	Types of CAI	PND (%)
Cullen, Alber-Morgan, Schnell, and Wheaton (2014)	4	5th	CAI with electronic text	75 for AIMSweb Maze assessments; 82 for OAA
Fitzgerald, Miller, Higgins, Pierce, and Tandy (2012)	5	6th–7th	CAI with a reading strategy embedded	NA
Harper and Ewing (1986)	9	11.5–13.5 years old (junior high)	CAI with electronic text	[1] CAI vs. baseline: 84; [2] CAI vs. non-CAI: 44
Meyer and Bouck (2014)	3	High school: 13, 14, 15 years old	CAI with speech synthesis	20
Stetter and Hughes (2011)	9^a	9th, 14–15 years old	CAI with a reading strategy embedded	0.7

Note. CAI = computer-assisted instruction; PND = percentage of nonoverlapping data; OAA = Ohio Achievement Assessment; NA = not applicable (data were not collected [at least] at three different points).

Six in intervention groups and three in control group.

Application of WWC Standards and Evaluation Results

Comparison group studies

Of the seven studies that employed an experimental or quasi-experimental design, four met WWC standards without reservations (K. Higgins et al., 1996; Keene & Davey, 1987; Raskind & Higgins, 1999; H. Swanson & Trahan, 1992) and one study met WWC standards with reservations (Bahr et al., 1991). The four studies meeting the WWC standards without reservations used a random process for generating groups and did not have evidence of high attrition. Of the three studies using a quasi-experimental design (Bahr et al., 1991; E. L. Higgins & Raskind, 2004; Jiménez et al., 2003), only one study (Bahr et al., 1991) met WWC standards with reservations. This study demonstrated baseline equivalence for pretest scores in treatment and comparison groups. In the following section, five intervention studies that met the WWC standards without or with reservations are presented, followed by two studies that did not meet standards.

Study that met WWC standards with reservations

Bahr et al. (1991) used a quasi-experimental two-by-two factorial design to examine two factors for 48 students with LD: teacher training on how to employ the computer software (training vs. no training) and student grouping based on pretest scores (homogeneous vs. heterogeneous) on vocabulary and reading comprehension (literal and inferential) measures. The Computer-Based Informal Reading Inventory System (CIRIS, 1986) software was used, with students silently reading a passage and answering multiple-choice questions during 15-min sessions. The software provided positive feedback for correct responses, corrective feedback for incorrect responses, and generated graphs indicating the percentage of questions answered correctly. Two reading comprehension subtests of the Stanford Diagnostic Reading Test (SDRT; Literal Comprehension and Inferential Comprehension) and Bader Informal Reading and Language Inventory (vocabulary and comprehension) were used as dependent measures. According to study findings, there were no significant differences for teacher training and for student grouping.

Studies that met WWC standards without reservations

K. Higgins et al. (1996) examined the effect of hypermedia study guides on the content area reading comprehension of 13 students with LD and remedial students. Students were randomly assigned to one of three conditions: (a) teacher lecture, (b) teacher lecture and hypermedia text, and (c) hypermedia text only. In this study, hypermedia text included vocabulary support (e.g., short explanation of target word, synonyms), multiple-choice questions, and graphics. A 50-item multiple-choice test based on questions from daily reading passage quizzes was used to measure reading comprehension. Students with LD in the teacher lecture plus hypermedia text condition outperformed students with LD in the two other conditions: lecture and hypermedia text vs. lecture (effect size [ES] = 235.26) and hypermedia text vs. lecture (ES = 72.69).

Keene and Davey (1987) examined the effects of reading computer-based text with 51 students with LD. Students were randomly assigned to computer-displayed text condition (n = 25) or a printed page condition (n = 26). Students in each group were instructed to use strategy cues, answer questions, and read two passages and reading comprehension was assessed with researcher-developed multiple-choice questions (factual and inferential). In this study, statistically significant differences were not found on immediate and delayed assessments of comprehension.

Raskind and Higgins (1999) examined the effects of a CAI intervention on the writing performance (e.g., self-selected compositions, class assignments) of 39 students with LD. Students in the experimental condition (n = 19) used speech recognition technology to complete writing activities and control students (n = 20) completed simple keyboarding activities such as typing. Reading comprehension was assessed using the silent reading portion of the Formal Reading Inventory (Wiederholt, 1986) to examine possible transfer effects from CAI writing. In this study, a statistically significant (ES = .76) effect was demonstrated in favor of the intervention condition.

H. Swanson and Trahan (1992) examined the effects of computer-mediated text by randomly assigning 60 students with LD to one of four treatment conditions: (a) a control condition in which students received typical instruction and no direct instruction in reading comprehension strategies, (b) a condition in which students read printed text passages, (c) a CAI condition in which students read passages on a computer screen and were not permitted to reread previous frames before completing a cloze assessment, and (d) a CAI condition in which students used a “back arrow” feature to reread all or portions of passages before answering comprehension questions. Level B of the Nelson Reading Skills Test (Hanna, Schell, & Schreiner, 1977) and cloze comprehension questions were used as dependent measures. According to study findings, the two computer-mediated text conditions were not superior to the printed text reading conditions: Nelson Reading (ES = .08), traditional questions (ES = −.18), and cloze questions (ES = .02).

Studies that did not meet WWC standards

E. L. Higgins and Raskind (2004) examined the effectiveness of two CAI interventions with 44 students with LD: a computer speech recognition-based program (SRBP) and a computer and text-based automaticity program (AP). In SRBP, students read self-selected stories on screen at their own pace. Also, they clicked single words to listen to prerecorded human voices and were asked to choose the correct spoken word from five similar words provided on the next screen. In the AP condition, students worked with a teacher or a partner on decoding and fluency-building activities (e.g., repeated reading). Students in the AP condition were also encouraged to use text-to-speech software such as Quicktionary Reading Pen or Kurzweil 3000. Students in a comparison condition received teacher provided reading instruction with no CAI. In this study, reading comprehension was assessed with the silent reading portion of the Formal Reading Inventory (Wiederholt, 1986). According to study findings, both the SRBP (ES = .90) and AP groups (ES = 1.09) demonstrated statistically significant differences to the comparison group.

Jiménez et al. (2003) examined the effects of a CAI intervention in which 14 students with LD were presented segmented words through digitized speech and a visual model. During 40-min sessions, students first attempted to pronounce a segment of a target word and then were presented an audio model by the computer. Students could then either repeat the task with the same segment or move on to the next segment in the target word. Upon successfully pronouncing the target word, students were presented a new word to work with. Using the Spanish Standardized Reading Skills Test (Cuetos, Rodríguez, & Ruano, 1996), the performance of students with LD in this intervention group was compared with a group of poor readers who received the same CAI intervention and to a group of students with reading disabilities who did not receive CAI. While the effects of CAI were not evident for the control-treatment group comparisons (ES = .18), within-group differences (pretest to posttest) were significant (ES = .70), indicating an improvement in reading comprehension. An overall summary of quality indicator ratings for comparison group studies is presented in Table 3.

Table 3.

Quality Indicators Ratings for Experimental or Quasi-Experimental Studies.

Quality indicators/criteria	Bahr, Kinzer, and Rieth (1991)	K. Higgins, Boone, and Lovitt (1996)	E. L. Higgins and Raskind (2004)	Jiménez et al. (2003)	Keene and Davey (1987)	Raskind and Higgins (1999)	H. Swanson and Trahan (1992)
A primary analysis of the effect of an intervention	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Study design	Quasi	Randomized	Quasi	Quasi	Randomized	Randomized	Randomized

Random process?	No	Yes	No	No	Yes	Yes	Yes
If yes go to sample attrition If no, go to baseline equivalence
Sample attrition		No			No	No	No
Is the combination of overall and differential attrition high? If no, study meets WWC group design standards without reservations
Baseline equivalence: If yes, study meets group design standards with reservations	Yes		No	No
Determination	Meets standards with reservations	Meets standards without reservations	Does not meet standards	Does not meet standards	Meets standards without reservations	Meets standards without reservations	Meets standards without reservations

Note. For sample attrition, author query was conducted to increase the evaluation accuracy, and no supplementary information (against the evaluation results) was found. WWC = What Works Clearinghouse.

Single-case design studies

A total of five single-case studies met selection criteria, two of which met WWC standards without reservations (Cullen et al., 2014; Harper & Ewing, 1986). However, the analysis of one of the two reading comprehension outcomes in Cullen et al. met standards with reservations due to the presence of only four data points during an intervention phase for one participant. A total of three single-case studies did not meet design standards. Two studies did not meet standards due to insufficient attempts to demonstrate effects and one study did not meet standards due to the absence of reliability data. Ratings for each study according to the WWC standards are presented in Table 4. A summary of single-case studies included in this review is reported in the following section.

Table 4.

Quality Indicators Ratings for Single-Case Design Studies.

Quality indicators/criteria	Cullen, Alber-Morgan, Schnell, and Wheaton (2014)	Fitzgerald, Miller, Higgins, Pierce, and Tandy (2012)	Harper and Ewing (1986)	Meyer and Bouck (2014)	Stetter and Hughes (2011)
The independent variable systematically manipulated with the researcher determining when and how the independent variable conditions change	Yes	Yes	Yes	Yes	Yes
Outcome variables must be measured systematically (i.e., repeatedly) over time by more than one assessor	Yes	Yes	Yes	Yes	No
The study needs to collect interassessor agreement on at least 20% of the data points in each condition (e.g., baseline, intervention)	Yes	Yes	Yes	Yes	No
Interassessor agreement must meet minimum thresholds. Minimum thresholds include .80 for percentage agreement indices and .60 for kappa measures	Yes	Yes	Yes	Yes	No
The study must include at least three attempts to demonstrate an intervention effect at three different points in time or with three different phase repetitions	Yes	No	Yes	No	Yes
For a phase to qualify as an attempt to demonstrate an effect, the phase must have a minimum of three data points	Yes	NA	Yes	Yes	Yes
Final rating	With and without reservations	Does not meet standards	Without reservations	Does not meet standards	Does not meet standards

Studies that met WWC standards with and without reservations

Using a multiple baseline across participants design, Cullen et al. (2014) examined the effects of the Headsprout Comprehension program with four students with LD. In the intervention condition, students read expository and narrative text and answered questions related to main idea identification, making inferences, and vocabulary. Dependent variables included the AIMSweb Maze assessment and Ohio Achievement Assessment (OAA) passage comprehension questions. Across intervention phases, the average percentage of nonoverlapping data (PND) was 75% for AIMSweb Maze assessments and 82% for OAA. This study received a final rating of meets design standards without reservations as WWC conventions stipulate that single-case studies are assigned a final rating based on the highest design standard rating received from relevant dependent variables (in this case, reading comprehension) within the study.

Harper and Ewing (1986) used an alternating treatment design to compare the effects of microcomputer instruction (CAI) and workbook instruction for nine students with LD. Students worked with either a microcomputer (text on screen) or workbook (text on paper) for their silent reading comprehension performance; comprehension was measured by the Comprehensive Test of Basic Skills (CTBS; 1973) and an informal reading inventory. For eight of nine students, CAI was superior to the workbook condition. Specifically, when CAI was compared to baseline, the average PND was 84%; when CAI was compared to workbook instruction, the average PND was 44%. This study met design standards without reservations.

Studies that did not meet WWC standards

Using a multiple-probe design, Fitzgerald et al. (2012) investigated the effect of online modules for five students with LD. During the CAI intervention condition, participants (a) divided words into prefixes, stems, and suffixes; (b) orally read a passage and had their reading recorded as an MP3 file; (c) answered comprehension questions; and (d) emailed the completed word study activity, MP3 file, and comprehension question answers to a special education teacher. The Woodcock Johnson Passage Comprehension subtest (Woodcock, McGrew, & Mather, 2001) and 10 multiple-choice comprehension questions from the Timed Readings series (Spargo, 1989) were used as dependent measures. Although the authors indicated that the intervention was effective at improving reading comprehension, this study did not include at least three attempts to demonstrate an effect at three different points in time. Therefore, this study did not meet WWC design standards.

Meyer and Bouck (2014) examined the effects of a computer-based text-to-speech intervention with three adolescents with LD. Students in the text-to-speech condition listened to reading passages and answered comprehension questions that accompanied each passage. Findings suggest that computer-based text-to-speech did not have an effect on reading comprehension as PND was 20%. However, this study used nonconcurrent multiple-baseline design, which lacks sufficient rigor according to the WWC protocol. Therefore, this study did not meet design standards.

Using a multiple-baseline design, Stetter and Hughes (2011) examined the effects of CAI on the reading comprehension of nine high school students with LD. CAI focused on teaching students a story mapping strategy (e.g., character, plot, setting, theme) and vocabulary development and daily quizzes and the Gates–MacGinitie (MacGinitie, MacGinitie, Maria, & Dreyer, 2001) were used to assess reading comprehension. Although this study reported that students showed improvements on the Gates–MacGinitie, this analysis can only be considered exploratory since single-case designs rely on repeated measurement over time. Average PND for quizzes was 2%. This study did not meet WWC design standards due to the absence of reliability data.

Discussion

This current review summarized and evaluated the rigor of CAI intervention studies for improving the reading comprehension of students with LD. Twelve studies met study selection criteria and were reviewed according to the WWC design standards (Institute of Education Sciences, 2014) to determine whether computer-based CAI was an evidence-based practice. Our analysis indicated that seven of 12 studies met design standards with or without reservations. Of the five comparison group studies, two (K. Higgins et al., 1996; Raskind & Higgins, 1999) demonstrated positive effects on reading comprehension. Both single-case studies (Cullen et al., 2014; Harper & Ewing, 1986) also demonstrated positive effects. Considering the evidence provided by comparison group and single-case studies as a whole, it appears that CAI may have potentially positive effects on the reading comprehension of students with LD as at least one study had a statistically significant effect, the total number of studies with noneffects was fewer than the number of studies with positive effects, and no studies demonstrated negative effects.

Key Instructional Features of CAI Studies Meeting the WWC Standards

Although a small number of studies (five out of 12 selected studies) met standards without reservations, there are common instructional features across these studies that warrant consideration. First, three studies meeting WWC standards without reservations and reporting high effect sizes included interventions that provided frequent opportunities to respond and/or practice. For example, students with LD had an opportunity to read texts on a computer screen and demonstrate their comprehension by answering multiple-choice questions on a computer screen (Cullen et al., 2014; K. Higgins et al., 1996) or participated in writing activities using speech synthesis (E. L. Higgins & Raskind, 2004). Effective CAI interventions also automatically determined lesson sequence based on student response and provided prompting (Cullen et al., 2014; K. Higgins et al., 1996). Students also received teacher support on how to work at computers; as they became comfortable with their independent work, teacher support was gradually faded (E. L. Higgins & Raskind, 2004).

Opportunities for self-correction and immediate corrective feedback were also included in studies meeting WWC criteria with high effect sizes. CAI programs supported self-correction by reinforcing correct answers through text or visual supports (e.g., graphics) or by prompting students provide a different response (Cullen et al., 2014). When students continued making incorrect responses, CAI provided additional instruction and practice to assist with self-correction (E. L. Higgins & Raskind, 2004). When a student chose an incorrect answer, digitized speech produced the sound “try again,” and the student had multiple opportunities to select a correct answer until she/he reached the correct answer. In another study, incorrect answers resulted in students being rerouted back to the text and the student engaging in self-correction (K. Higgins et al., 1996). Corrective feedback was also evident in Cullen et al., where the built-in corrective feedback was presented by virtually erasing incorrect responses, highlighting the correct answer, and providing an explanation of why the student’s answer was incorrect.

Activities designed to engage and motivate students to read were also commonly included in studies with evidence of effectiveness. In one study, students read text on self-selected topics (K. Higgins et al., 1996). In another study, students earned virtual coins for correct responses that were redeemed for time to play short videos or games (Cullen et al., 2014).

Teacher-directed instruction was often embedded in CAI or provided along with CAI (e.g., Cullen et al., 2014; K. Higgins et al., 1996; Stetter & Hughes, 2011). For example, in one study students received a teacher-led lecture prior to participating in CAI sessions (K. Higgins et al., 1996). Finally, different types of questioning (factual, inferential, and note questions) were also applied to CAI programs. The questions ranged from factual and inferential questions to note questions (questions that can be answered by only reading the assigned reading text) where the students were asked to check pop-up text and then respond (K. Higgins et al., 1996). Students also received instruction on how to answer several types of questions for reading comprehension, such as finding a main idea, understanding unknown words, and making inferences (Cullen et al., 2014).

Findings Related to Methodological Rigor

Although this review contributes to the existing knowledge regarding the effectiveness of CAI interventions for reading comprehension and common instructional features within studies demonstrating effectiveness, the following methodological concerns also warrant consideration: reporting of treatment fidelity, the manner in which CAI interventions were described, and the internal validity of the corpus of studies included in this review.

Treatment fidelity

Swanson and her colleagues (E. Swanson, Wanzek, Haring, Ciullo, & McCulley, 2013) noted that less than 70% of research studies published in high-impact journals reported information on fidelity. Similarly, studies included in this review tended to omit information on fidelity. For example, no group design studies described procedures for measuring treatment fidelity or reported formal data. Although the WWC standards do not include fidelity as a quality indicator, the absence of treatment fidelity calls into question the degree to which interventions were provided as intended. As recommended by Swanson et al., researchers should report fidelity measurement procedures and the degree to which implementation aligns with intervention procedures. Fidelity ensures that an instructional practice and its content are consistently conveyed across different agents (e.g., teachers) and in different settings (Glasgow, Lichtenstein, & Marcus, 2003). Without sufficient documentation of implementation, the ability to make causal claims regarding CAI interventions is compromised (Cook, 2014; Smith, Daunic, & Taylor, 2007).

Description of CAI intervention

Adequate description of intervention and baseline conditions is necessary to ensure that empirical findings are reliable and to provide a foundation for replication (Gersten et al., 2005; Shadish, Campbell, & Cook, 2002). For experimental or quasi-experimental design studies, information on interventionist characteristics and description of comparison conditions was limited. In regard to single-case design studies, description of independent variables including instructional materials, procedures, and duration of sessions and intervention conditions were not clearly reported. Although technology is the main component of CAI interventions, it is essential to describe the role of the interventionist during implementation. Although most studies included in this review lacked detailed descriptions of the intervention and CAI instructional features (e.g., corrective feedback, modeling), intervention effects can be enhanced when combined with explicit and well-planned instruction (Regan et al., 2014). For example, CAI with additional computer-based supports such as speech technology and vocabulary supports (e.g., K. Higgins et al., 1996; E. L. Higgins & Raskind, 2004; Jiménez et al., 2003) appeared to be more effective than interventions that required students to only reading text on computer screen (e.g., Bahr et al., 1991; Keene & Davey, 1987).

Threats to internal validity

For comparison group studies in which random assignment was not employed, baseline equivalence is necessary to make certain mean differences between groups are due to the intervention (Shadish et al., 2002). In this review, two quasi-experimental studies did not meet the WWC standards due to a failure to document the comparability of participants between conditions (e.g., demographic variables, reading measures). Because the dependent variable in each study selected for this review was reading comprehension, the comparability of participants’ reading ability and equivalence of groups across conditions are vital to assure the treatment effect of CAI instruction for target students. In regard to single-case studies, three of five studies did not meet design standards. A failure to employ a design that permitted the establishment of an effect at three points in time was a common concern. Although single-case designs offer many advantages to researchers such as the need to recruit fewer participants, causal inferences can only be made when designs with sufficient rigor are employed.

Limitations

Five limitations are associated with this study. First, other technology-assisted approaches (e.g., tablet PC, iPads, iPhones, and other kinds of technology) were not included in this review. In addition, a limited number of studies met inclusion criteria due to specific content and student group targeted (reading comprehension, students with LD). Second, although the WWC standards have been frequently used to evaluate study rigor, the process of developing and refining quality indicators is ongoing as additional factors warrant consideration (e.g., inconsistencies in how to measure and interpret implementation fidelity, evaluation of mixed-method designs; for example, Chatterji, 2004; Schoenfeld, 2006). Use of another set of quality indicators may have yielded different findings. Third, the current study did not include unpublished manuscripts or dissertations. This decision may have resulted in fewer studies included in this analysis or fewer studies that did not have evidence of an intervention effect. Fourth, PND was used to evaluate the effectiveness of single-case studies. Although commonly used, PND has several limitations and therefore could produce misleading findings (see Parker, Hagan-Burke, & Vannest, 2007). Lastly, the studies that did not disaggregate data for students with LD were excluded. It is possible that at least some excluded studies could meet the WWC standards and thus contribute to the research base regarding the degree to which CAI is an evidence-based practice for developing the reading comprehension skills of students with LD.

Implications for Practice

Findings from this study suggest a few implications for school practice. Considering the quality of the literature, the use of CAI to improve the reading comprehension of students with LD should be viewed with caution. There are a number of technologies available to schools that lack a research base (Edyburn, 2013) and findings from this study suggest that CAI may be a promising practice for improving the reading comprehension of students with LD. However, it may be advantageous for schools to place greater emphasis on teacher-directed interventions that have an evidence base (see Berkeley, Scruggs, & Mastropieri, 2010; Ciullo, Lo, Wanzek, & Reed, 2014). CAI reading interventions may best be viewed as supplementary to these efforts and as a means of supporting students with disabilities to have “new opportunities for accessing and engaging in the general curriculum” (Edyburn, 2013, p. 9). Schools should carefully consider the evidence base when selecting technology-based interventions to integrate into typical school practice so that this process is driven by research findings rather than other factors (Edyburn, 2013). Schools may find it beneficial to select CAI programs that include instructional elements present in studies with strong effects and that met WWC criteria (e.g., frequent opportunities to respond, procedures for matching task difficulty to student performance levels, opportunities for self-correction, prompt fading). Instead of simply providing texts on screen (electronic text only), providing additional computer-based supports (e.g., speech synthesis, feedback, vocabulary support) and strategy instruction may be more effective. Schools should also provide ongoing professional development and coaching to assist teacher efforts to integrate new instructional practices (e.g., use of CAI) into their typical teaching practice (McKenna & Ciullo, 2016).

Implications for Future Research

Additional studies and replications that encompass a broad range of technology uses and across grade levels are needed. As most CAI studies included in this review were conducted in resource rooms or computer labs, investigations in inclusive settings are warranted. Future studies should employ designs with sufficient rigor to permit causal inferences regarding the effects of CAI interventions. In fact, researchers have previously reported a need for more rigorous research on the use of computers in special education (Edyburn, 2013). Future studies should be reported in a manner that permits replication and documents the degree to which the independent variable (intervention) was provided as intended. In addition, explicit instruction in reading comprehension skills has rarely been paired with CAI interventions. Therefore, future research should examine the effects of explicit reading comprehension instruction paired with CAI. Lastly, future research should investigate how findings vary when different sets of quality indicators are used to evaluate a corpus of studies.

Footnotes

Associate Editor: Margaret King-Sears

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

References

Allington

R. L.

(2012). What really matters for struggling readers: Designing research-based programs. Boston, MA: Pearson.

*Bahr

C. M.

Kinzer

C. K.

Rieth

H. J.

(1991). An analysis of the effects of teacher training and student grouping on reading comprehension skills among mildly handicapped high school students using computer-assisted instruction. Journal of Special Education Technology, 11, 136–154.

Beichner

Schwartz

(1999). Essentials of educational technology. Needham Heights, MA: Allyn & Bacon.

Berkeley

Scruggs

Mastropieri

(2010). Reading comprehension instruction for students with learning disabilities, 1995—2006: A meta-analysis. Remedial and Special Education, 31, 423–436.

Bryant

B. R.

Kim

Kang

Bryant

D. P.

Lang

Son

(2015). A comparison of the effects of reading interventions on engagement and performance for fourth-grade students with learning disabilities. Behavior Modification, 39, 167–190.

Chatterji

(2004). Evidence on “what works”: An argument for extended-term mixed-method (ETMM) evaluation designs. Educational Researcher, 33, 3–13.

Ciullo

Wanzek

Reed

(2014). A synthesis of research on informational text reading interventions for elementary students with learning disabilities. Journal of Learning Disabilities, 49, 257–271.

Clarfield

Stoner

(2005). The effects of computerized reading instruction on the academic performance of students identified with ADHD. School Psychology Review, 34, 246–254.

Common Core State Standards . (2010). National Governors Association and Council of Chief State School Officers. Retrieved http://www.corestandards.org/

10.

Comprehensive test of basic skills . (1973). New York, NY: McGraw Hill.

11.

Computer-Based Informal Reading Inventory System [Computer program]. (1986). Nashville, TN: Peabody College of Vanderbilt University.

12.

Cook

B. G.

(2014). A call for examining replication and bias in special education research. Remedial and Special Education, 35, 233–246.

13.

Cuetos

Rodríguez

Ruano

(1996). Assessment test of reading skills for children in primary education (PROLEC). Madrid, Spain: Technical Specialists Association.

14.

*Cullen

J. M.

Alber-Morgan

S. R.

Schnell

S. T.

Wheaton

J. E.

(2014). Improving reading skills of students with disabilities using Headsprout comprehension. Remedial and Special Education, 35, 356–365.

15.

Duke

Carlisle

(2010). The development of reading comprehension. In Kamil

M. L.

Pearson

D. P.

Moje

E. B.

Afflerbach

(Eds.), The handbook of reading research (Vol. IV, pp. 199–228). London, England: Routlege.

16.

Edyburn

(2013). Critical issues in advancing the special education technology evidence base. Exceptional Children, 80, 7–24.

17.

*Fitzgerald

N. S.

Miller

S. P.

Higgins

Pierce

Tandy

(2012). Exploring the efficacy of online strategy instruction for improving the reading abilities of students with learning disabilities. Journal of Special Education Technology, 27, 33–47.

18.

Gersten

Fuchs

Compton

Coyne

Greenwood

Innocenti

(2005). Quality indicators for group experimental and quasi-experimental research in special education. Exceptional Children, 71, 149–164.

19.

Glasgow

R. E.

Lichtenstein

Marcus

A. C.

(2003). Why don’t we see more translation of health promotion research to practice? Rethinking the efficacy-to-effectiveness transition. American Journal of Public Health, 93, 1261–1267.

20.

Hall

T. E.

Hughes

C. A.

Filbert

(2000). Computer assisted instruction in reading for students with learning disabilities: A research synthesis. Education & Treatment of Children, 23, 173–193.

21.

Hanna

Schell

L. M.

Schreiner

(1977). The Nelson reading skills test. Itasca, IL: Riverside.

22.

*Harper

J. A.

Ewing

N. J.

(1986). A comparison of the effectiveness of microcomputer and workbook instruction on reading comprehension performance of high incidence handicapped children. Educational Technology, 26, 40–45.

23.

*Higgins

E. L.

Raskind

M. H.

(2004). Speech recognition-based and automaticity programs to help students with severe reading and spelling problems. Annals of Dyslexia, 54, 365–392.

24.

*Higgins

Boone

R. A.

Lovitt

T. C.

(1996). Hypertext support for remedial students and students with learning disabilities. Journal of Learning Disabilities, 29, 402–412.

25.

Institute of Education Sciences. (2014). What Works Clearinghouse study review standards. Retrieved from https://ies.ed.gov/ncee/wwc/FWW

26.

*Jiménez

J. E.

del Rosario

O. M.

Rodrigo

Hernández-Valle

Ramírez

Estévez

O’Shanahan

de la Luz Trabaue

(2003). Do the effects of computer-assisted practice differ for children with reading disabilities with and without IQ-achievement discrepancy? Journal of Learning Disabilities, 36, 34–47.

27.

Kang

McKenna

Arden

Ciullo

(2015). Integrated reading and writing interventions for students with learning disabilities: A review of the literature. Learning Disabilities Research & Practice, 31, 22–33.

28.

*Keene

Davey

(1987). Effects of computer-presented text on LD adolescents’ reading behaviors. Learning Disability Quarterly, 10, 283–290.

29.

Kena

Hussar

McFarland

de Brey

Musu-Gillette

Wang

. . . Dunlop Velez

(2016). The condition of education 2016 (NCES 2016144). Retrieved from http://nces.ed.gov/pubsearch/pubsinfo.asp?pubid=2016144

30.

Kim

J. S.

Samson

J. F.

Fitzgerald

Hartry

(2010). A randomized experiment of a mixed methods literacy intervention for struggling readers in grades 4–6: Effects on word reading efficiency, reading comprehension and vocabulary, and oral reading fluency. Reading and Writing, 23, 1109–1129.

31.

King-Sears

Evmenova

(2007). Premises, principles, and processes for integrating technology into instruction. Teaching Exceptional Children, 40, 6–14.

32.

King-Sears

Johnson

Berkeley

Weiss

Peters-Burton

Evmenova

. . . Hursh

(2015). An exploratory study of universal design for teaching chemistry to students with and without disabilities. Learning Disability Quarterly, 38, 84–96.

33.

MacArthur

C. A.

Ferretti

R. P.

Okolo

C. M.

Cavalier

A. R.

(2001). Technology applications for students with literacy problems: A critical review. The Elementary School Journal, 101, 273–301.

34.

MacGinitie

W. H.

MacGinitie

R. K.

Maria

Dreyer

L. G.

(2001). Gates-MacGinitie Reading Tests (4th ed.). Itasca, IL: Riverside.

35.

McKenna

Ciullo

(2016). Typical reading instructional practices provided to students with emotional and behavioral disorders in a residential and day treatment setting: A mixed methods study. Residential Treatment for Children & Youth, 33, 225–246.

36.

McKenna

Shin

Ciullo

(2015). Evaluating reading and mathematics instruction for students with learning disabilities: A synthesis of observation research. Learning Disability Quarterly, 38, 195–207.

37.

*Meyer

N. K.

Bouck

E. C.

(2014). The impact of text-to-speech on expository reading for adolescents with LD. Journal of Special Education Technology, 29, 21–33.

38.

Montali

Lewandowski

(1996). Bimodal reading: Benefits of a talking computer for average and less skilled readers. Journal of Learning Disabilities, 29, 271–279.

39.

Moore

(2009). The everything parent’s guide to children with special needs: A reassuring, informative guide to your child’s well-being and happiness. Avon, MA: Adams Media.

40.

National Center for Learning Disabilities. (2014). The state of learning disabilities (3rd ed.). New York, NY: Author.

41.

Parker

R. I.

Hagan-Burke

Vannest

(2007). Percentage of all non-overlapping data (PAND): An alternative to PND. The Journal of Special Education, 40, 194–204.

42.

*Raskind

M. H.

Higgins

E. L.

(1999). Speaking to read: The effects of speech recognition technology on the reading and spelling performance of children with learning disabilities. Annals of Dyslexia, 49, 251–281.

43.

Regan

Berkeley

Hughes

Kirby

(2014). Effects of computer-assisted instruction for struggling elementary readers with disabilities. Journal of Special Education, 48, 106–119.

44.

Reutzel

D. R.

Cooter

R. B.

(2013). Essentials of teaching children to read: The teacher makes the difference (3rd ed.). Boston, MA: Pearson.

45.

Schoenfeld

A. H.

(2006). What doesn’t work: The challenge and failure of the What Works Clearinghouse to conduct meaningful reviews of studies of mathematics curricula. Educational Researcher, 35, 13–21.

46.

Sencibaugh

J. M.

(2007). Meta-analysis of reading comprehension interventions for students with learning disabilities: Strategies and implications. Reading Improvement, 44, 6–22.

47.

Shadish

Cook

Campbell

(2002). Experimental and quasi-experimental designs for generalized causal inference. Boston, MA: Houghton-Mifflin.

48.

Shaywitz

S. E.

Shaywitz

B. A.

(2005). Dyslexia (specific reading disability). Biological Psychiatry, 57, 1301–1309.

49.

Shaywitz

S. E.

Shaywitz

B. A.

(2008). Paying attention to reading: The neurobiology of reading and dyslexia. Development and Psychopathology, 20, 1329–1349.

50.

Shores

Chester

(2008). Using RTI for school improvement. Thousand Oaks, CA: Corwin Press.

51.

Smith

S. W.

Daunic

A. P.

Taylor

G. G.

(2007). Treatment fidelity in applied educational research: Expanding the adoption and application of measures to ensure evidence-based practice. Education & Treatment of Children, 30, 121–134.

52.

Snyder

T. D.

Dillow

S. A.

(2015). Digest of education statistics 2013 (NCES 2015-011). Washington, DC: National Center for Education Statistics, Institute of Education Sciences, U.S. Department of Education.

53.

Solan

H. A.

Shelley-Tremblay

Ficarra

Silverman

Larson

(2003). Effect of attention therapy on reading comprehension. Journal of Learning Disabilities, 36, 556–563.

54.

Solis

Ciullo

Vaughn

Pyle

Hassaram

Leroux

(2012). Reading comprehension interventions for middle school students with learning disabilities: A synthesis of 30 years of research. Journal of Learning Disabilities, 45, 327–340.

55.

Spargo

(1989). Timed readings (3rd ed.). New York, NY: Glencoe McGraw-Hill.

56.

Stetter

M. E.

Hughes

M. T.

(2010). Computer-assisted instruction to enhance the reading comprehension of struggling readers: A review of the literature. Journal of Special Education Technology, 25, 1–16.

57.

*Stetter

M. E.

Hughes

M. T.

(2011). Computer assisted instruction to promote comprehension in students with learning disabilities. International Journal of Special Education, 26, 88–100.

58.

Swanson

(2008). Observing reading instruction for students with learning disabilities: A synthesis. Learning Disability Quarterly, 31, 115–133.

59.

Swanson

Wanzek

Haring

Ciullo

McCulley

(2013). Intervention fidelity in special and general education research journals. Journal of Special Education, 47, 3–13.

60.

*Swanson

Trahan

(1992). Learning disabled readers’ comprehension of computer mediated text: The influence of working memory, metacognition and attribution. Learning Disabilities Research & Practice, 7, 74–86.

61.

Vacca

R. T.

Vacca

J. L.

Mraz

(2011). Content-area reading (10th ed.). Boston, MA: Pearson.

62.

What Works Clearinghouse. (2011). Intervention: Check & Connect. Retrieved from https://ies.ed.gov/ncee/wwc/EvidenceSnapshot/79

63.

What Works Clearinghouse. (2014). Procedures and standards handbook (version 3.0). Retrieved from https://ies.ed.gov/ncee/wwc/Docs/referenceresources/wwc_procedures_v3_0_standards_handbook.pdf

64.

Wiederholt

L. J.

(1986). Formal Reading Inventory. Austin, TX: Pro-Ed.

65.

Wise

J. C.

Sevcik

R. A.

Morris

R. D.

Lovett

M. W.

Wolf

(2007). The relationship among receptive and expressive vocabulary, listening comprehension, pre-reading skills, word identification skills, and reading comprehension by children with reading disabilities. Journal of Speech, Language, and Hearing Research, 50, 1093–1109.

66.

Woodcock

R. W.

McGrew

K. S.

Mather

(2001). Woodcock-Johnson III Tests of Achievement. Itasca, IL: Riverside.

67.

Xin

J. F.

Rieth

(2001). Video-assisted vocabulary instruction for elementary school students with learning disabilities. Information Technology in Childhood Education Annual, 12, 87–104.