View in EDS HTML Full Text PDF Full Text

Structure Altering Effects of a Multicomponent Reading Intervention: An Application of the Direct and Inferential Mediation (DIME) Model of Reading Comprehension in Upper Elementary Grades

Saved in:
Bibliographic Details
Title: Structure Altering Effects of a Multicomponent Reading Intervention: An Application of the Direct and Inferential Mediation (DIME) Model of Reading Comprehension in Upper Elementary Grades
Language: English
Authors: Ahmed, Yusra, Miciak, Jeremy, Taylor, W. Pat, Francis, David J.
Source: Journal of Learning Disabilities. Jan-Feb 2022 55(1):58-78.
Availability: SAGE Publications and Hammill Institute on Disabilities. 2455 Teller Road, Thousand Oaks, CA 91320. Tel: 800-818-7243; Tel: 805-499-9774; Fax: 800-583-2665; e-mail: journals@sagepub.com; Web site: http://sagepub.com
Peer Reviewed: Y
Page Count: 21
Publication Date: 2022
Sponsoring Agency: Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD) (DHHS/NIH)
Contract Number: R01HD096262
P50HD052117
Document Type: Journal Articles
Reports - Research
Education Level: Elementary Education
Early Childhood Education
Grade 3
Primary Education
Grade 4
Intermediate Grades
Grade 5
Middle Schools
Descriptors: Reading Instruction, Reading Comprehension, Elementary School Students, Reading Difficulties, Grade 3, Grade 4, Grade 5, Intervention, Instructional Effectiveness, After School Programs
Assessment and Survey Identifiers: Gates MacGinitie Reading Tests, Woodcock Johnson Tests of Achievement, Kaufman Brief Intelligence Test
DOI: 10.1177/0022219421995904
ISSN: 0022-2194
Abstract: We evaluate the direct and inferential mediation (DIME) model for reading comprehension with a sample of struggling readers in Grades 3 to 5 (N = 364) in the context of a large-scale randomized controlled trial (RCT) investigating two theoretically distinct reading interventions (text processing + foundational skills [n = 117] or text processing only [n = 120]) and a control condition (n = 127). We investigate whether the intervention affects not just reading comprehension levels, but also how variables within the reading system interrelate. This approach allows the focus to shift from intervention as influencing a change in reading comprehension status to a complex set of processes. We fit structural equation models (SEMs) to evaluate the DIME model at baseline and a change model that included reading comprehension and word reading at posttest. There were no significant mean differences between groups in reading comprehension. However, significant differences emerged on the direct and indirect effects of background knowledge, vocabulary, word reading, strategies, and inferencing on comprehension across grade levels and treatment conditions. Related to treatment groups, background knowledge, vocabulary, and inferencing were significantly related to comprehension at posttest for students who received text processing and/or foundational skills interventions. The results have implications for the direct instruction of higher-order reading skills in the context of multicomponent interventions.
Abstractor: As Provided
Entry Date: 2022
Accession Number: EJ1321503
Database: ERIC
Full text is not displayed to guests.
FullText Links:
  – Type: pdflink
    Url: https://content.ebscohost.com/cds/retrieve?content=AQICAHj0k_4E0hTGH8RJwT4gCJyBsGNe_WN95AvKlDbXJGqwxwHuVt346cMd9IgCs2u833R_AAAA4TCB3gYJKoZIhvcNAQcGoIHQMIHNAgEAMIHHBgkqhkiG9w0BBwEwHgYJYIZIAWUDBAEuMBEEDATa86Qw-bg2QdQAuAIBEICBmevZQbinyUM2GwMdSn7LMli4SjVCFfHgcZ2MuB_Znr3Z-VQjaVIs_AvFunD1EeKvUMUdFCFx6H81N4i0jZVR3Efa4IeLnCKQIXBOXnCUcELHP4gS_w2rFKsH2eQU1Bv-SN4v55ODpQrP-h1Efqf5Fum-eq-URXC4vfk92iAr9ZGoCSV3Dl5D311twAj13v4_RanL9ERFqlBsRw==
Text:
  Availability: 1
  Value: <anid>AN0154068092;led01jan.22;2021Dec14.06:59;v2.2.500</anid> <title id="AN0154068092-1">Structure Altering Effects of a Multicomponent Reading Intervention: An Application of the Direct and Inferential Mediation (DIME) Model of Reading Comprehension in Upper Elementary Grades </title> <p>We evaluate the direct and inferential mediation (DIME) model for reading comprehension with a sample of struggling readers in Grades 3 to 5 (N = 364) in the context of a large-scale randomized controlled trial (RCT) investigating two theoretically distinct reading interventions (text processing + foundational skills [ n = 117] or text processing only [ n = 120]) and a control condition (n = 127). We investigate whether the intervention affects not just reading comprehension levels, but also how variables within the reading system interrelate. This approach allows the focus to shift from intervention as influencing a change in reading comprehension status to a complex set of processes. We fit structural equation models (SEMs) to evaluate the DIME model at baseline and a change model that included reading comprehension and word reading at posttest. There were no significant mean differences between groups in reading comprehension. However, significant differences emerged on the direct and indirect effects of background knowledge, vocabulary, word reading, strategies, and inferencing on comprehension across grade levels and treatment conditions. Related to treatment groups, background knowledge, vocabulary, and inferencing were significantly related to comprehension at posttest for students who received text processing and/or foundational skills interventions. The results have implications for the direct instruction of higher-order reading skills in the context of multicomponent interventions.</p> <p>Keywords: reading comprehension; vocabulary; background knowledge; inferencing; reading strategies; reading intervention</p> <p>In a recent essay on the topic of building usable knowledge in education, [<reflink idref="bib35" id="ref1">35</reflink>] identified three major challenges facing educational research: (a) dealing with the replication crisis that has plagued other scientific fields, (b) establishing better evidence for the generalizability of educational research, and (c) adapting rigorous research designs to match the increasing complexity of educational interventions. This third challenge is particularly salient in reading intervention research, a field celebrated for its successes in establishing effective intervention practices in early elementary grades but which to date has failed to produce a similarly robust evidence base for intervention practices in later grades. This failure to find large treatment effects for interventions implemented in grades beyond early elementary is frequently attributed to the complexity of the reading comprehension construct at these ages and the limited malleability of some component skills (i.e., linguistic comprehension, background knowledge; [<reflink idref="bib12" id="ref2">12</reflink>]; [<reflink idref="bib18" id="ref3">18</reflink>]; [<reflink idref="bib76" id="ref4">76</reflink>]). Studies that implement specific strategies and which may yield positive effects on researcher-created proximal measures of reading have failed to produce detectable effects on more distal standardized measures of reading comprehension ([<reflink idref="bib66" id="ref5">66</reflink>], [<reflink idref="bib65" id="ref6">65</reflink>]). As a result, recent trends in reading intervention research have included a move away from specific strategy instruction and toward more complex, multicomponent interventions (e.g., [<reflink idref="bib76" id="ref7">76</reflink>]).</p> <p>However, complex multicomponent interventions present challenges related to the identification of casual mechanisms. Main effects analyses cannot inform questions about which component or sequence of components are most efficacious ([<reflink idref="bib35" id="ref8">35</reflink>]); thus, main effects analyses leave unanswered questions about <emph>why</emph> or <emph>how</emph> a complex intervention works ([<reflink idref="bib30" id="ref9">30</reflink>]). In addition, main effects analyses featuring a complex, multifaceted outcome such as reading comprehension may not identify effects on constituent skills, changes in the interrelations of constituent skills, or potential interactions with student attributes ([<reflink idref="bib29" id="ref10">29</reflink>]). For example, it is possible that an intervention that failed to produce detectable between-group differences on a complex construct such as reading comprehension may still alter the relationships between constituent skills and the outcome of interest. Such results, when informed by theory and resulting from rigorous experimental designs that generally permit causal inferences, can be particularly informative for future intervention research as they suggest promising intervention practices and unpack complexity ([<reflink idref="bib21" id="ref11">21</reflink>]).</p> <p>In the present study, we evaluate an empirically validated, theoretical model of reading comprehension in the context of a rigorous randomized trial with students in upper elementary grades: the direct and inferential mediation model (DIME) of reading comprehension with a sample of struggling readers in Grades 3 to 5. The study is guided by two goals. First, we seek to replicate the DIME model for the first time with a sample of struggling readers in late elementary. This replication extends research on the generalizability of the DIME model to new students and contexts and permits comparisons with findings from previous studies with older students (e.g., [<reflink idref="bib2" id="ref12">2</reflink>]; [<reflink idref="bib19" id="ref13">19</reflink>]; [<reflink idref="bib20" id="ref14">20</reflink>]). Second, we seek to compare covariance structures among constituent reading skills within randomly assigned intervention and comparison groups. An evaluation of within-group covariances permits inspection of whether participation in one of two distinct reading intervention programs potentially altered the relations between constituent skills. This analysis serves to "unpack" the intervention and parse its potential effects on and among constituent reading skills, an important goal in the context of experiments evaluating complex intervention packages through large-scale, time-, and resource-intensive randomized trials.</p> <hd id="AN0154068092-2">Reading Interventions in Late Elementary</hd> <p>Despite success in developing and validating early interventions for students in early elementary grades, efficacious interventions for older students have been slower to emerge. This fact is often attributed to the changing nature of the reading task in late elementary grades. First recognized as a phenomenon in which children with average reading abilities in early elementary began to experience difficulty in later elementary—called the "fourth grade slump" ([<reflink idref="bib13" id="ref15">13</reflink>])—there is widespread recognition that as students progress through elementary school, the primary academic task shifts from learning to read toward reading to learn. Students are expected to read and understand grade-level tasks featuring more complex ideation, language, and structure. As a result, the strong coupling between foundational reading skills (i.e., decoding and fluency) observed in early elementary grades weakens and more complex skills such as background knowledge, language, and inferencing become increasingly predictive ([<reflink idref="bib12" id="ref16">12</reflink>]).</p> <p>However, recent empirical work suggests that these more complex predictive skills may not be as malleable to intervention as word-level skills that form the target of early elementary interventions ([<reflink idref="bib51" id="ref17">51</reflink>]; [<reflink idref="bib76" id="ref18">76</reflink>]). Evidence for limited malleability is summarized in recent meta-analytic work that found that average effect sizes for interventions in late elementary on standardized measures of reading comprehension are much smaller than those observed in early elementary ([<reflink idref="bib65" id="ref19">65</reflink>]). Identifying key component reading skills that are malleable to intervention in late elementary and secondary settings represents a critical research priority.</p> <hd id="AN0154068092-3">Reading Comprehension: The Simple View of Reading (SVR)</hd> <p>There is considerable practical interest in understanding the component skills underlying reading comprehension; multiple theoretical models have been proposed that demonstrate varying levels of support in empirical literature. Among the most prominent, the SVR ([<reflink idref="bib32" id="ref20">32</reflink>]) posits that reading comprehension is the product of decoding and linguistic comprehension. The SVR has proven to be a robust, influential model for understanding the component skills underlying reading comprehension, and empirical studies have consistently found that it explains a significant degree of variance in reading comprehension across different ages and measures (e.g., [<reflink idref="bib1" id="ref21">1</reflink>]; [<reflink idref="bib27" id="ref22">27</reflink>]). However, there is also growing recognition that the SVR provides a limited framework for designing multicomponent reading interventions targeting component reading skills ([<reflink idref="bib11" id="ref23">11</reflink>]; [<reflink idref="bib18" id="ref24">18</reflink>]). The key aspect of the explanatory elegance of the SVR—its "simplicity"—presents a challenge in developing comprehensive theories of change that are responsive to the complexity of the comprehension task and might guide development for interventions for struggling students in Grades 3+. The critical challenge is to unpack the subcomponents of linguistic comprehension and identify those component skills (e.g., content knowledge) that may be malleable and targeted through intervention ([<reflink idref="bib11" id="ref25">11</reflink>]).</p> <p>Using the SVR as a theoretical framework, [<reflink idref="bib61" id="ref26">61</reflink>] took a theory-driven approach to intervention evaluation in a sample of struggling adolescents for whom differences in posttest means were not robust. The researchers posited that a moderated mediation model in which decoding and listening comprehension were regressed on reading comprehension would result in different covariance structures for students who received an intensive, yearlong self-regulated strategy treatment versus those in the business as usual condition. In addition to the components of the SVR, they evaluated the effect of verbal knowledge (measured using a test that probes mainly vocabulary). That is, decoding and listening comprehension mediated the relations among verbal knowledge and reading comprehension, and strategy instruction moderated the mediation. The authors hypothesized that self-regulated strategy treatment would disrupt the mediation of verbal knowledge by way of listening comprehension because the treatment provided more sophisticated alternatives than those afforded by children's verbal knowledge alone (e.g., local cues available in the text). The direct effects of decoding, listening comprehension, and verbal knowledge were similar across treatment and control conditions at pretest, indicating that both groups relied on verbal knowledge in addition to components of the SVR, a finding that generally does not hold in typically developing children. At posttest, the direct effect of decoding was significant in both groups (and larger in the treatment group). The effect of listening comprehension was only significant in the treatment group, whereas the effect of verbal knowledge was significant only in the control group. Corresponding to their hypothesis, the indirect effect of listening comprehension was significant only in the treatment group, whereas the indirect effect of decoding was significant in both groups. Two important limitations of the study are that (a) although verbal knowledge was included in the model, it was not manipulated in the study, and (b) factors such as comprehension strategies, inference-making, and prior knowledge were part of the theory of change but were also not manipulated nor explicitly measured. Nonetheless, the study shows that interrelations among variables of the SVR changed over time in response to intervention. Next, we describe the DIME model, an alternative model of reading comprehension with promising potential to identify specific and malleable intervention targets.</p> <hd id="AN0154068092-4">The DIME Model of Reading Comprehension</hd> <p>The DIME model (Figure 1, left) represents a component-based empirical model of reading comprehension that could serve to inform intervention development in upper elementary grades. It was built upon a recognition of the changing reading task throughout school, in which foundational reading skills such as decoding diminish in importance as students age and more complex skills such as background knowledge and inferencing become more predictive ([<reflink idref="bib12" id="ref27">12</reflink>]; [<reflink idref="bib23" id="ref28">23</reflink>]). In recognition of the need for theoretical models developed and validated in secondary settings, [<reflink idref="bib19" id="ref29">19</reflink>] conducted an extensive review of experiments investigating predictors of reading comprehension in high school reading. Based on that review, they hypothesized a model of reading comprehension that included five components: (a) word reading, (b) vocabulary, (c) background knowledge, (d) reading strategies, and (e) inferencing. The DIME components focus on interindividual differences in text processing, or engagement with reading materials rather than discourse-level oral language. Among these skills, networks of potential direct and indirect relations were put forth, varying in some path specifications. Thus, the DIME model is different from the SVR in two ways: (a) instead of focusing on listening comprehension it includes higher-order linguistic processes that take place during reading, and (b) it explicates important mediated relations among the component skills. Some have characterized the DIME model as an extension of the SVR ([<reflink idref="bib2" id="ref30">2</reflink>]), by unpacking of linguistic comprehension into lower- and higher-level language subcomponents, whereas others have suggested that the DIME model represents a contrasting model ([<reflink idref="bib39" id="ref31">39</reflink>]). The DIME model is suitable for students in upper elementary grades and constitutes a comprehensive model of reading comprehension to the extent that it incorporates multiple processes and products of comprehension (e.g., [<reflink idref="bib60" id="ref32">60</reflink>]; [<reflink idref="bib67" id="ref33">67</reflink>]), extra-textual dimensions such as strategy use, and emergent properties of theoretical frameworks such as construction-integration and landscape models because it specifies relations among lower and higher-order processes from these frameworks (see [<reflink idref="bib42" id="ref34">42</reflink>]; [<reflink idref="bib50" id="ref35">50</reflink>]).</p> <p>Graph: Figure 1. DIME model at baseline (left; Model 1), and change model including word reading and comprehension measured at pre- and posttest (right; Model 2). Note. Paths 1 to 8 were unconstrained between control and treatment groups. All other paths (e.g., RS-to-INF) were constrained equal across conditions in the change model. DIME = direct and inferential mediation; COMP = reading comprehension; WR = word reading; RS = reading strategies, INF = inferencing.</p> <p>Systematic replications of the DIME model have included variations in age range, including students in middle and high school ([<reflink idref="bib2" id="ref36">2</reflink>]) and college students ([<reflink idref="bib20" id="ref37">20</reflink>]), as well as variations in the number and types of measures, including domain-general/domain-specific measures, experimental and standardized tests, as well as state-assessments. Despite considerable progress in validating the DIME model with students in primarily middle, secondary, and postsecondary settings, there remain significant gaps in the DIME model literature. First, the full DIME model has not been tested with students in late elementary, a particularly salient age in light of the changing nature of the reading task at this age. In addition, few studies of the DIME model have focused specifically on struggling readers, a population of considerable interest. Previous studies of the DIME model have utilized a cross-sectional research design, with simultaneous measurement occurring at one time. It is unclear how these relationships may predict changes over time, particularly in response to systematic instructional differences.</p> <hd id="AN0154068092-5">Purpose of the Present Study</hd> <p>This study utilizes data from a large, randomized trial evaluating the efficacy of two after-school reading comprehension interventions implemented with students enrolled in Grades 3 to 5 ([<reflink idref="bib62" id="ref38">62</reflink>]). These grades may represent a key inflection point in students' reading trajectory; studies of component reading skills are necessary and informative, particularly in the context of intensive reading interventions with strong experimental controls. In the present study, we address this important research need. The study is guided by two research goals. First, we replicate the DIME model within this sample of younger students and disaggregate by grade. <bold>This replication permits comparison of previous investigations of the DIME model with older students and allows for inspection of how these component skills relate during this key educational transition. (RQ1). Second, we investigate whether posttest performance in reading comprehension is predicted differentially by pretest performance on component reading skills and whether the interrelations of those skills differ by randomly assigned treatment group. (RQ2).</bold> Thus, both of our research questions consist of moderated mediations with grade and intervention condition as moderators for RQ1 and RQ2, respectively. For the first research question (RQ1), we use data on component skills of the DIME model measured concurrently at baseline. We hypothesized that the DIME model would replicate for this sample of younger students, many with reading comprehension difficulties, but that there would be differences in the relative importance of higher-order component skills for reading comprehension. Specifically, based on previous research with students in late elementary and beyond ([<reflink idref="bib12" id="ref39">12</reflink>]; [<reflink idref="bib23" id="ref40">23</reflink>]) and because most struggling readers demonstrate comprehensive reading deficits ([<reflink idref="bib17" id="ref41">17</reflink>]), we expected that word reading and vocabulary would represent a more robust predictor of reading comprehension than in previous research on the DIME model with older, typical readers. In addition, due to the increased relative importance of word reading, we expected that more complex component skills (inferencing, reading strategies) would demonstrate weaker, but still significant, relations to comprehension in Grade 3 relative to Grades 4 and 5. For the second research question (RQ2), we add reading comprehension and decoding measured longitudinally at posttest. Like [<reflink idref="bib61" id="ref42">61</reflink>], the intervention in the current study did not manipulate the component skills of the theoretical model, but we hypothesized that relations among higher-order malleable skills of reading comprehension (e.g., inferencing and reading strategies) would be affected by treatment. Specifically, based on the findings of [<reflink idref="bib61" id="ref43">61</reflink>], we hypothesized that a multicomponent reading intervention that included foundational skills instruction (described below) would reduce the direct effects of word reading and allow students to apply more complex component skills to the reading task (e.g., vocabulary, background knowledge, inferencing, and strategies). Similarly, we hypothesized that because both intervention groups included extensive text-processing instruction, for both groups, the effects of background knowledge and vocabulary would be mediated by inferencing and reading strategies when compared with a business as usual comparison condition.</p> <hd id="AN0154068092-6">Method</hd> <p></p> <hd id="AN0154068092-7">Sample</hd> <p>The intent-to-treat randomized sample included 419 students who participated in a larger RCT of an after-school reading program ([<reflink idref="bib62" id="ref44">62</reflink>]). Students were sampled from seven participating elementary schools distributed across two school districts in two distinct urban regions in the southwestern United States. Students were identified as struggling readers if they scored at or below the 25th percentile on the Test of Reading Efficiency and Comprehension (TOSREC; [<reflink idref="bib81" id="ref45">81</reflink>]) and were then randomized to Treatment 1 (text processing with foundational skills [TP + FS]), Treatment 2 (text processing without foundational skills [TP]), or "business as usual" (BAU) conditions, with approximately 140 students per condition. The analytic sample for the present study (<emph>n</emph> = 364) included a subsample of students in Grades 3 (<emph>n</emph> = 105), 4 (<emph>n</emph> = 136), and 5 (<emph>n</emph> = 123) for whom data were available on the measures included in the present study. The students ranged in ages from 6.8 to 12.5 years. This sample included 117 students in Treatment 1, 120 in Treatment 2, and 127 in BAU conditions. No differences in reading or demographic variables were found at pretest between students in the treatment versus BAU conditions ([<reflink idref="bib62" id="ref46">62</reflink>]). The sample was representative of school districts' populations and consisted of 56% White, 45% Black, 39% Hispanic, 3% American Indian and Alaskan Native, 3% Asian, and <1% Native Hawaiian and Other Pacific Islander (some students identified one or more races). Most students were economically disadvantaged (73% free/reduced-cost lunch), and only a small portion of the sample was Limited English Proficient (LEP; 19%) or enrolled in special education (21%). Eighteen students (<5%) were dropped from the study at posttest because they moved (<emph>n</emph> = 9), withdrew participation from the study (<emph>n</emph> = 7), or were absent (<emph>n</emph> = 1), but these students were not significantly different from the retained sample at pretest on gender, χ<sups>2</sups>(<reflink idref="bib1" id="ref47">1</reflink>) = 0.04, <emph>p</emph> =.84; race, χ<sups>2</sups>(<reflink idref="bib4" id="ref48">4</reflink>) = 1.84, <emph>p</emph> =.76; LEP status, χ<sups>2</sups>(<reflink idref="bib1" id="ref49">1</reflink>) = 0.04, <emph>p</emph> =.84; special education status, χ<sups>2</sups>(<reflink idref="bib1" id="ref50">1</reflink>) = 0.12, <emph>p</emph> =.73; or measures of decoding, χ<sups>2</sups>(<reflink idref="bib333" id="ref51">333</reflink>) = 342.73, <emph>p</emph> =.34; reading comprehension, χ<sups>2</sups>(<reflink idref="bib317" id="ref52">317</reflink>) = 307.27, <emph>p</emph> =.64; vocabulary, χ<sups>2</sups>(<reflink idref="bib340" id="ref53">340</reflink>) = 334.96, <emph>p</emph> =.57; inferencing, χ<sups>2</sups>(<reflink idref="bib333" id="ref54">333</reflink>) = 332.09, <emph>p</emph> =.50; background knowledge, χ<sups>2</sups>(<reflink idref="bib3" id="ref55">3</reflink>) = 0.39, <emph>p</emph> =.94; or strategies, χ<sups>2</sups>(<reflink idref="bib205" id="ref56">205</reflink>) = 181.40, <emph>p</emph> =.88. Although there were significant differences in socioeconomic status (SES) (as measured by free or reduced lunch status) between students lost to attrition and those who remained in the sample, χ<sups>2</sups>(<reflink idref="bib3" id="ref57">3</reflink>) = 23.13, <emph>p</emph> <.0001, there was no differential attrition among study groups, χ<sups>2</sups>(<reflink idref="bib4" id="ref58">4</reflink>) = 7.52, <emph>p</emph> =.11.</p> <hd id="AN0154068092-8">Intervention Conditions and Procedures</hd> <p>The intervention consisted of an after-school reading program designed to improve struggling readers' text processing plus foundational skills (TP + FS; Treatment 1) or text processing (TP; Treatment 2). The BAU condition consisted of a "business as usual" comparison group that did not participate in the researcher-provided after-school reading program or any other after-school tutoring program. Thus, students in the BAU and treatment conditions received all instruction and interventions typically available in the school (described in [<reflink idref="bib62" id="ref59">62</reflink>], but students in the BAU condition did not receive the additional reading instruction provided after school. Both intervention conditions included (a) 30 min of computer-based reading instruction plus (b) 30 min of small-group (~three to six students per group) instruction, for a total of 60 min of instruction per session, and up to four sessions per week from approximately November to May of a single school year (see [<reflink idref="bib62" id="ref60">62</reflink>], for additional details). The small-group instruction occurred in two phases as described below, where the second phase started after the 20th session and consisted of advanced lessons of either self-regulation or writing (see [<reflink idref="bib62" id="ref61">62</reflink>]). Thus, in Phase 2, intervention condition (TX1 or TX2) was crossed with instruction modality (writing or self-regulation). The two intervention conditions featured instruction focused on text processing, which we conceptually contrast with <emph>text-based strategy use</emph> or <emph>reading strategy interventions</emph> that seek to directly train specific reading skills or strategies before, after, or during reading (i.e., previewing text, inferencing, etc.; [<reflink idref="bib4" id="ref62">4</reflink>]; [<reflink idref="bib77" id="ref63">77</reflink>]; [<reflink idref="bib82" id="ref64">82</reflink>]). The goal of text-processing instruction was for students to develop reading comprehension by reading, thinking about, and discussing text. Thus, the intervention provided implicit instruction in higher-level skills such as inferencing and reading strategies.</p> <hd id="AN0154068092-9">TP + FS (Treatment 1)</hd> <p>Prior to beginning the computer component for the TP + FS intervention, each student completed a program-specific pretest to determine their placement in the adaptive curriculum. Progress through the pretest curriculum determined each student's individual sequence through the curriculum. Each day, students participated in computer instruction targeting domains of weakness, while progressing to more complex skills and domains as they achieved mastery of foundational skills (i.e., decoding, fluency). The TP + FS computer-based program featured explicit instruction in phonological awareness, decoding, reading fluency, vocabulary, grammar, and text-reading. This instruction took the form of "games" that required students to apply a specific component reading skill after a brief explicit lesson (e.g., identifying the number of sounds in a word for phonological awareness; selecting the correct definition for a vocabulary word). To motivate students, their progress was tracked through mastery badges and points.</p> <p>The focus of the teacher-directed small-group instruction was on explicit instruction in decoding and fluency, with embedded comprehension instruction each day. Activities included explicit phonics instruction, structured reading fluency activities, explicit vocabulary instruction, text-reading with questions and summarization, and syntax and semantics. Instruction was organized in two phases, with Phase 1 (Lessons 1–20) emphasizing word reading and reading fluency with embedded comprehension instruction. Phase 2 (Lessons 20+) featured more advanced foundational skills instruction (i.e., multisyllabic word reading, extended fluency with text). In addition, Phase 2 instruction included more complex vocabulary and text, including two instructional days per week with a 30-min "stretch text," which was an extended grade-level text to give students practice reading and understanding text with more complex structure, language, and ideation. Example small group lessons from the TP + FS condition are accessible at https://<ulink href="http://www.texasldcenter.org/files/lesson-plans/TCLD%5f3-5%5fT1-TBI-SelfReg.pdf">www.texasldcenter.org/files/lesson-plans/TCLD%5f3-5%5fT1-TBI-SelfReg.pdf</ulink> and https://<ulink href="http://www.texasldcenter.org/files/lesson-plans/TCLD%5f3-5%5fT1-TBI-Writing.pdf">www.texasldcenter.org/files/lesson-plans/TCLD%5f3-5%5fT1-TBI-Writing.pdf</ulink>.</p> <hd id="AN0154068092-10">Text Processing (TP; Treatment 2)</hd> <p>The computer component for the TP intervention was not adaptive and required students to read paper or digital texts cataloged within the computer program and complete multiple-choice questions about the text following reading. Students who passed a quiz accrued reading points, which were utilized as part of a program at their school to earn prizes and rewards. The program included a large variety of available texts and quizzes, and students could choose specific texts under the guidance of the supervising tutor. After reading the text, students completed a short multiple-choice quiz on the computer to ensure that they had read and understood the text. The small-group instruction in the TP condition consisted of a "book-club" format, with students reading both instructional and grade-level content and narrative texts. Texts were cooperatively chosen by the students and tutor. Prior to reading, students were asked to make predictions about the text. During reading, tutors asked a variety of comprehension questions. Text reading included a variety of reading formats, including tutor modeling, choral reading, partner reading, and individual reading. In contrast with the TP + FS condition, students in the TP condition did not receive systematic and explicit instruction in word reading, reading fluency, or vocabulary. Instead, instructional time focused on maximizing reading time to improve text processing skills. Example lessons from the TP condition are available at https://<ulink href="http://www.texasldcenter.org/files/lesson-plans/TCLD%5f3-5%5fT2-BCI-SelfRegLite.pdf">www.texasldcenter.org/files/lesson-plans/TCLD%5f3-5%5fT2-BCI-SelfRegLite.pdf</ulink> and https://<ulink href="http://www.texasldcenter.org/files/lesson-plans/TCLD%5f3-5%5fT2-BCIWriting.pdf">www.texasldcenter.org/files/lesson-plans/TCLD%5f3-5%5fT2-BCIWriting.pdf</ulink>.</p> <hd id="AN0154068092-11">BAU</hd> <p>Students in the BAU condition did not participate in the after-school reading intervention and did not have access to any of the researcher-provided intervention materials. They received all instruction typically available at the participating schools.</p> <hd id="AN0154068092-12">Measures</hd> <p>The pretest battery included all the measures described below, whereas the posttest battery included only decoding and reading comprehension measures because the larger study was designed to test the main effect of intervention on reading. Reading comprehension, background knowledge, and strategies measures were group-administered, whereas word reading, vocabulary, and inference measures were individually administered in the schools by rigorously trained test administrators.</p> <hd id="AN0154068092-13">Reading Comprehension</hd> <p>Reading comprehension was measured using two standardized tests, the Gates–MacGinitie Reading Comprehension Test (GMRT; [<reflink idref="bib47" id="ref65">47</reflink>]) and two forms of the Test of Silent Reading Comprehension and Efficiency (TOSREC; [<reflink idref="bib81" id="ref66">81</reflink>]). The GMRT requires students to read a series of narrative, expository (texts whose intent is primarily to instruct), or setting passages (portion of texts that do not move a story forward in time), and each passage is associated with a series of multiple-choice questions that test literal or inferential understanding of the passage. Kuder-Richardson 20 (K-R 20) reliability is high for Grades 3 (.94), 4 (.93), and 5 (.93) for Form T used in the current study. The TOSREC requires students to read sentences and students are asked to verify the truthfulness of as many sentences as possible. Alternate-form reliability is high for Grades 3 (.93), 4 (.86), and 5 (.89) for Forms A and O used in the current study.</p> <hd id="AN0154068092-14">Word Reading</hd> <p>Word reading (decoding) was measured using two standardized tests, the Woodcock Johnson III Letter Word Identification (WJ-LWID; [<reflink idref="bib83" id="ref67">83</reflink>]) and the Sight Word Efficiency (SWE) subtest of the Test of Word Reading Efficiency (TOWRE; [<reflink idref="bib73" id="ref68">73</reflink>]). Both measures require reading words of increasing difficulty, with less frequent words appearing toward the end of the list. The TOWRE-SWE measures both speed and accuracy of reading words. Both subtests demonstrate excellent psychometric properties, with a median reliability of.91 for ages 5 to 19 for the WJ-LWID. Alternate-form reliability ranged from.90 to.93 for ages 6 to 12/Grades 3 to 5 for the TOWRE-SWE Form A used in this study.</p> <hd id="AN0154068092-15">Vocabulary</hd> <p>Vocabulary was measured using the verbal knowledge subtest of the Kaufman Brief Intelligence Test (KBIT-2; [<reflink idref="bib37" id="ref69">37</reflink>]). The verbal knowledge subtest measures receptive language and general information and does not require reading or spelling. The student points to a picture that shows the meaning of a word or provides the answer to a question. Internal consistency coefficients (split-half) for the verbal scores for ages 6 through 12 (covering Grades 3 through 5) range from.86 to.93 for verbal knowledge.</p> <hd id="AN0154068092-16">Inference</hd> <p>We administered a paper-and-pencil version of the Bridge-IT ([<reflink idref="bib58" id="ref70">58</reflink>]) test of inference-making. The Bridge-IT is designed to measure the effect of textual distance (near/far) on bridging inferences ability. Students were presented with short narrative passages consisting of four sentences followed by three continuation sentences. Students were asked to choose the continuation sentence that best follows from the passage. In the far condition, the passages consisted of a statement sentence (e.g., "Tim was very full because his mother had made him a very big lunch") followed by three intervening sentences (e.g., "They played some video games and then opened presents"), and in the near condition the passage consisted of three intervening sentences followed by the statement sentence. To integrate information presented in the statement sentence and the correct continuation sentence (e.g., "Tim asked for a small piece of cake"), students must rule out inconsistent continuation sentences (e.g., "Tim asked for a big piece of cake"). Parallel forms reliability is reported as.73 in Grades 3 to 8 ([<reflink idref="bib58" id="ref71">58</reflink>]).</p> <hd id="AN0154068092-17">Background Knowledge</hd> <p>A measure of background knowledge was derived using items from an experimental measure of reading and writing known as the Assessment of Writing, Self-Monitoring and Reading (AWSM Reading) developed for the larger study. Students were asked to read three expository passages on the following topics: preserving sand dunes, Yellowstone National Park, and barefoot running. The background knowledge items required students to answer questions such as "What is found inside Yellowstone National Park?" prior to reading the passage. The open-ended questions were scored for accuracy. Average Kuder-Richardson 20 reliability coefficients for the AWSM Reading ranged from.62 to.69 in this sample.</p> <hd id="AN0154068092-18">Reading Strategies</hd> <p>Strategies were assessed via the Student Contextual Learning Scale (SCL; [<reflink idref="bib15" id="ref72">15</reflink>]). The SCL is a student report measure that evaluates attitudes, beliefs, and habits related to reading and learning. Items capture information related to the constructs of self-efficacy, effort, and enjoyment of the reading task. In the present study, we utilized items that load onto the strategies scale, which included items such as "I ask myself questions to make sure I know the material I've been studying." In previous evaluations of the SCL, we have validated the factor structure and differential item functioning ([<reflink idref="bib16" id="ref73">16</reflink>]). Kuder-Richardson 20 reliability ranged from.71 to.82 in this sample. Summary writing items from the AWSM reading test were also used to form a latent variable for strategies. These open-ended items required students to read the three AWSM Reader passages and subsequently write a summary for each passage. The items were scored for the number of important ideas from the passage expressed accurately. Inter-rater reliability (Cohen's kappa) for this sample was.92 to.97 for summary writing items across grade levels.</p> <hd id="AN0154068092-19">Analytic Approach</hd> <p>All models were fit using Mplus 8.1 ([<reflink idref="bib52" id="ref74">52</reflink>]). We first established measurement invariance across grade levels for RQ1 (replication) and across treatment conditions for RQ2 (intervention). Then we estimated a multigroup structural model at baseline for Grades 3, 4, and 5 using all measures at pretest for RQ1 (Model 1: concurrent model) in which reading comprehension was regressed on (a) background knowledge, (b) vocabulary, (c) word reading (decoding), (d) reading strategies, and (e) inferences. The direct effects of background knowledge and vocabulary were mediated by reading strategies and inferences, and the effect of reading strategies was further mediated by inferences. The structural model for RQ2 (Model 2: change model) included decoding and reading comprehension measured longitudinally at posttest moderated by experimental condition. All the covariances from the baseline model were estimated (i.e., all direct and indirect effects from Models 1 and 3 correlations), and equality constraints were imposed such that treatment conditions did not differ at pretest. All paths regressed on reading comprehension at pretest in the baseline models were also regressed on reading comprehension at posttest. This model included the auto-regressor effects of decoding and reading comprehension at posttest regressed on decoding and reading comprehension at pretest, respectively. These eight paths (Figure 1, right) were unconstrained across conditions because we expected treatment to alter relations with posttest comprehension. Details about the statistical analyses are presented next.</p> <hd id="AN0154068092-20">Model Fit Indices</hd> <p>Traditional indices and cut-off values (e.g., root mean square error of approximation [RMSEA] ≤ 0.08, standardized root mean square residual [SRMR] ≤ 0.06, and comparative fit index [CFI] and Tucker Lewis index [TLI] ≥.95) were used to evaluate the fit of individual models for <emph>n</emph> < 500 ([<reflink idref="bib36" id="ref75">36</reflink>]). We used nested models for comparing among competing alternatives and testing equivalence of model parameters. Specifically, we used chi-square difference test for models estimated with maximum likelihood and Satorra–Bentler scaled chi-square difference test for models estimated with robust estimators ([<reflink idref="bib64" id="ref76">64</reflink>]).</p> <hd id="AN0154068092-21">Moderated Mediation</hd> <p>To test the DIME model at different levels of the moderating variables, multigroup modeling was used with grade as a moderator in Model 1 and treatment condition as a moderator in Model 2 because both grade and condition are categorical variables. Multigroup, moderated mediation with categorical mediators concerns the difference in the indirect effects between groups and the moderator does not appear in the model as a variable ([<reflink idref="bib63" id="ref77">63</reflink>]). Direct and indirect effects are interpreted as <emph>conditional</emph> if the effects are significant in one group and not another ([<reflink idref="bib59" id="ref78">59</reflink>]). Bias-corrected bootstrap standard errors and confidence intervals from 5,000 replications were obtained for testing mediation effects ([<reflink idref="bib7" id="ref79">7</reflink>]; [<reflink idref="bib46" id="ref80">46</reflink>]).</p> <hd id="AN0154068092-22">Clustering</hd> <p>Clustering adjustments were made because reading skills for students within classrooms may be correlated, as well as within condition within classrooms. In the present study, there were 93 classrooms with small cluster sizes (<emph>M</emph> = 3.91; <emph>SD</emph> = 2.27) and treatment assignment was not at the classroom level. Most classes (<emph>n</emph> = 63) included children in all experimental conditions, 20 classes included children in treatment conditions only, and 10 classes included children in the control condition only. Thus, we used condition within classrooms as the clustering units (<emph>n</emph> = 157), with small cluster sizes (<emph>M</emph> = 2.33; <emph>SD</emph> = 1.46). Because an explicit multi-level modeling approach requires many clusters and relatively large cluster sizes, we were not able to explicitly model at both the student-within-classroom and classroom levels simultaneously. Rather, we modeled the total-groups covariance structure (within- and between-classrooms), but we addressed the effects of clustering on the standard errors. Specifically, we addressed the clustering of students within condition and classrooms by modeling data at the student-level and obtaining standard errors that are robust to non-normality and nonindependence. Maximum likelihood estimation with robust standard errors (MLR) was used with sampling specified as complex using the TYPE = COMPLEX command. Furthermore, the MLR approach, which relies on distributional assumptions, was compared with maximum likelihood estimation with bootstrapping using 5,000 draws to increase the confidence of the results from the MLR approach. Results of the sampling from the empirical distribution (MLR) were largely comparable with the bootstrap resampling approach, but we present the results from ML approach with bootstrapping for Models 1 and 2 because of the extremely small cluster sizes in the MLR approach (namely, approximately 50% of the condition within classroom clusters consisted of <emph>n</emph> = 1). While our approach conflates the within- and between-classroom (i.e., cluster) covariance, it is not feasible to disentangle the two sources of covariation given the limited number of clusters and the small number of students per cluster. However, it is possible to address the impact of clustering on the standard errors using the approaches taken here.</p> <hd id="AN0154068092-23">Measurement Invariance</hd> <p>To establish measurement invariance across groups, a series of increasingly restrictive confirmatory factor analytic models (CFA) were fit. The baseline (configural) model estimated the DIME components without equality constraints across groups. Equality constraints were imposed on the following parameters: factor loadings (metric model), factor loadings and intercepts (scalar model), factor loadings and select intercepts (partial scalar model), and factor loadings, intercepts, and residual variances (strict model). The models were estimated separately for grade invariance and condition invariance. The condition invariance models included longitudinal invariance (pre- and posttest) for word reading and comprehension as these factors were measured at two time-points.</p> <hd id="AN0154068092-24">Structural Invariance</hd> <p>For RQ1, we present the results of a noninvariant structural model because we were interested in exploring the developmental differences in the DIME model, but we note that an omnibus model of structural noninvariance was not significantly different from a grade-invariant model, χ<sups>2</sups>Δ(<reflink idref="bib26" id="ref81">26</reflink>) = 37.89, <emph>p</emph> >.05, indicating that a parsimonious model aggregating Grades 3 through 5 fit as well as a model in which the covariances in the model varied as a function of grade level. Because we did not expect differences across experimental conditions at pretest, in the structural model for RQ2, all single-headed paths to reading comprehension at pretest were constrained to be equal across conditions, but all paths to reading comprehension at posttest were freely estimated (see Figure 1, right). This model was significantly different from an omnibus test of structural noninvariance, χ<sups>2</sups>Δ(<reflink idref="bib14" id="ref82">14</reflink>) = 23.23, <emph>p</emph> =.05, indicating that constructs did not relate in the same way at pre- to posttest and posttest as a function of treatment. Although we do not assume that structural equation models (SEMs) establish causal relations from associations alone, in what follows we use causal language to the extent that SEM is appropriate for mediation analysis and has empirical implications ([<reflink idref="bib6" id="ref83">6</reflink>]).</p> <hd id="AN0154068092-25">Results</hd> <p>The means of vocabulary (<emph>M</emph> = 89.79, <emph>SD</emph> = 15.51), background knowledge (<emph>M</emph> = 2.26, <emph>SD</emph> = 0.83), strategies (Summary 1 [<emph>M</emph> = 0.95, <emph>SD</emph> = 1.18], Summary 2 [<emph>M</emph> = 0.99, <emph>SD</emph> = 1.20], Summary 3 [<emph>M</emph> = 0.41, <emph>SD</emph> = 0.69], SCL [<emph>M</emph> = 17.45, <emph>SD</emph> = 4.96]), and inferencing (near bridging inferences [<emph>M</emph> = 5.57, <emph>SD</emph> = 2.16], far bridging inferences [<emph>M</emph> = 4.53, <emph>SD</emph> = 1.88]) at pretest were within the expected range for struggling readers. Table 1 presents descriptive statistics for decoding and comprehension at pre- and posttest, reported for the full sample at pretest because the treatment and BAU conditions did not differ at pretest. Table 1 shows there were also no substantial differences between the groups at posttest on reading comprehension or decoding. The negligible effects of treatment on mean performance are presented in more detail in [<reflink idref="bib62" id="ref84">62</reflink>], but the intent-to-treat standardized effect of treatment was <emph>g</emph> =.01, <emph>SE</emph> =.07, <emph>p</emph> =.92, on the GMRT. Correlations among all measures can be found in the online supplemental materials table. Covariance coverage among measures exceeded 90%, and all missing data were handled in Mplus using full information maximum likelihood (FIML) because data were missing completely at random (MCAR) as determined by Little's MCAR test, χ<sups>2</sups>(<reflink idref="bib252" id="ref85">252</reflink>) = 271.24, <emph>p</emph> =.19 ([<reflink idref="bib44" id="ref86">44</reflink>]).</p> <p>Graph</p> <p>Table 1. Descriptive Statistics for Word Reading and Reading Comprehension.</p> <p> <ephtml> <table><colgroup><col align="left" /><col align="char" char="." /><col align="char" char="." /><col align="char" char="." /><col align="char" char="." /><col align="char" char="." /><col align="char" char="." /><col align="char" char="." /><col align="char" char="." /></colgroup><thead><tr><th /><th align="center" colspan="2" rowspan="2">Pretest (<italic>n</italic> = 364)</th><th align="center" colspan="6">Posttest</th></tr><tr><th /><th align="center" colspan="2">Control (<italic>n</italic> = 127)</th><th align="center" colspan="2">Treatment 1 (<italic>n</italic> = 117)</th><th align="center" colspan="2">Treatment 2 (<italic>n</italic> = 110)</th></tr><tr><th align="center">Variable</th><th align="center"><italic>M</italic></th><th align="center"><italic>SD</italic></th><th align="center"><italic>M</italic></th><th align="center"><italic>SD</italic></th><th align="center"><italic>M</italic></th><th align="center"><italic>SD</italic></th><th align="center"><italic>M</italic></th><th align="center"><italic>SD</italic></th></tr></thead><tbody><tr><td>Word reading</td><td><bold>0.00</bold><xref ref-type="table-fn" rid="tfn2">a</xref></td><td><bold>0.77</bold><xref ref-type="table-fn" rid="tfn2">a</xref></td><td><bold>0.00</bold></td><td><bold>0.68</bold></td><td><bold>0.06</bold></td><td><bold>0.54</bold></td><td>−<bold>0.04</bold></td><td><bold>0.57</bold></td></tr><tr><td> TOWRE SWE<xref ref-type="table-fn" rid="tfn2">b</xref></td><td>85.02</td><td>12.96</td><td>86.37</td><td>13.29</td><td>87.14</td><td>12.40</td><td>87.05</td><td>12.34</td></tr><tr><td> WJ3 LWID<xref ref-type="table-fn" rid="tfn2">b</xref></td><td>94.26</td><td>11.79</td><td>94.10</td><td>13.04</td><td>95.44</td><td>12.35</td><td>96.08</td><td>12.77</td></tr><tr><td>Reading comprehension</td><td>−<bold>0.00</bold><xref ref-type="table-fn" rid="tfn2">a</xref></td><td><bold>0.46</bold><xref ref-type="table-fn" rid="tfn2">a</xref></td><td>−<bold>0.01</bold></td><td><bold>0.53</bold></td><td><bold>0.06</bold></td><td><bold>0.52</bold></td><td><bold>0.08</bold></td><td><bold>0.61</bold></td></tr><tr><td> TOSREC 1<xref ref-type="table-fn" rid="tfn2">c</xref></td><td>13.41</td><td>5.37</td><td>17.78</td><td>7.04</td><td>18.41</td><td>6.07</td><td>19.04</td><td>6.77</td></tr><tr><td> TOSREC 2<xref ref-type="table-fn" rid="tfn2">c</xref></td><td>13.62</td><td>5.61</td><td>18.12</td><td>7.54</td><td>19.05</td><td>7.29</td><td>19.62</td><td>7.99</td></tr><tr><td> GMRT<xref ref-type="table-fn" rid="tfn2">b</xref></td><td>451.63</td><td>31.98</td><td>465.30</td><td>33.14</td><td>460.58</td><td>32.92</td><td>457.95</td><td>35.80</td></tr></tbody></table> </ephtml> </p> <p>1 <emph>Note</emph>. Values in bold are latent variable means and standard deviations from Model 2. TOWRE SWE = Test of Word Reading Efficiency: Sight Word Efficiency; WJ3 LWID = Woodcock Johnson III Letter Word Identification; TOSREC = Test of Silent Reading Efficiency and Comprehension; GMRT = Gates–MacGinitie Reading Test.</p> <p>2 Averaged across experimental conditions. <sups>b</sups>Standard scores. <sups>c</sups>Raw scores.</p> <p>Model fit indices for the measurement invariance models are presented in Table 2. The best fitting grade-invariant model was a partial scalar model in which factor loadings were constrained to be equal across groups, and intercepts were freed for inferences and reading strategies. As expected, component skills were condition-invariant at pretest, as the strict model was not significantly different from the more constrained scalar model at pretest. This finding was expected because invariance is tenable at the pretest due to randomization (i.e., any differences at pretest are due to chance). The best fitting condition-invariant model for RQ2 was a scalar model in which the factor loadings and intercepts were constrained to be equal across experimental conditions, holding partial scalar longitudinal invariance for word reading and comprehension at pre- and posttest, based on the results of the reduced longitudinal model (see Table 2). This model indicated that differences in factor means of word reading and comprehension were interpretable as mean differences and not a failure of the scales to hold over time and conditions.</p> <p>Graph</p> <p>Table 2. Model Fit Indices for the Measurement Invariance Models.</p> <p> <ephtml> <table><colgroup><col align="left" /><col align="char" char="." /><col align="char" char="." /><col align="char" char="." /><col align="char" char="." /><col align="char" char="." /><col align="char" char="." /><col align="char" char="." /><col align="char" char="." /><col align="char" char="." /></colgroup><thead><tr><th align="center">Model</th><th align="center">AIC</th><th align="center">Adjusted BIC</th><th align="center">χ<sup>2</sup></th><th align="center"><italic>df</italic></th><th align="center">RMSEA [90% CI]</th><th align="center">CFI</th><th align="center">TLI</th><th align="center">SRMR</th><th align="center">Δχ<sup>2</sup>(<italic>df</italic>) (vs. preceding model)</th></tr></thead><tbody><tr><td colspan="10">Grade (pretest)</td></tr><tr><td> Configural</td><td>10,915.53</td><td>11,031.66</td><td>163.54</td><td>144</td><td>0.03 [0.00, 0.06]</td><td>0.98</td><td>0.97</td><td>0.06</td><td /></tr><tr><td> Metric</td><td>10,909.31</td><td>11,015.77</td><td>185.32</td><td>158</td><td>0.04 [0.00, 0.06]</td><td>0.97</td><td>0.96</td><td>0.07</td><td>21.78 (14)</td></tr><tr><td> Partial scalar<xref ref-type="table-fn" rid="tfn4">a</xref></td><td>10,857.73</td><td>10,940.68</td><td>201.74</td><td>192</td><td>0.02 [0.00, 0.05]</td><td>0.99</td><td>0.99</td><td>0.07</td><td>16.42 (34)</td></tr><tr><td> Scalar</td><td>10,930.57</td><td>11,010.75</td><td>282.58<xref ref-type="table-fn" rid="tfn6">**</xref></td><td>196</td><td>0.06 [0.05, 0.08]</td><td>0.92</td><td>0.90</td><td>0.10</td><td>80.84 (4)<xref ref-type="table-fn" rid="tfn6">**</xref></td></tr><tr><td> Strict</td><td>10,999.25</td><td>11,058.70</td><td>411.26<xref ref-type="table-fn" rid="tfn6">**</xref></td><td>226</td><td>0.08 [0.07, 0.10]</td><td>0.82</td><td>0.82</td><td>0.14</td><td>128.68 (30)<xref ref-type="table-fn" rid="tfn6">**</xref></td></tr><tr><td colspan="10">Experimental condition (pretest)</td></tr><tr><td> Configural</td><td>11,072.21</td><td>11,188.34</td><td>163.55</td><td>144</td><td>0.03 [0.00, 0.06]</td><td>0.98</td><td>0.97</td><td>0.05</td><td /></tr><tr><td> Metric</td><td>11,054.59</td><td>11,161.04</td><td>173.93</td><td>158</td><td>0.03 [0.00, 0.05]</td><td>0.99</td><td>0.98</td><td>0.06</td><td>10.38 (14)</td></tr><tr><td> Scalar</td><td>11,005.73</td><td>11,085.91</td><td>201.07</td><td>196</td><td>0.02 [0.00, 0.04]</td><td>1.00</td><td>0.99</td><td>0.06</td><td>27.14 (38)</td></tr><tr><td> Strict</td><td>11,007.30</td><td>11,066.75</td><td>262.65</td><td>226</td><td>0.04 [0.00, 0.06]</td><td>0.97</td><td>0.96</td><td>0.08</td><td>61.58 (30)</td></tr><tr><td colspan="10">Experimental condition (pre- and posttest): reduced model<xref ref-type="table-fn" rid="tfn4">c</xref></td></tr><tr><td> Configural</td><td>6,974.39</td><td>7,061.49</td><td>119.43<xref ref-type="table-fn" rid="tfn6">**</xref></td><td>69</td><td>0.08 [0.05, 0.10]</td><td>0.97</td><td>0.94</td><td>0.04</td><td /></tr><tr><td> Metric</td><td>6,958.53</td><td>7,035.26</td><td>133.56<xref ref-type="table-fn" rid="tfn6">**</xref></td><td>84</td><td>0.07 [0.05, 0.09]</td><td>0.97</td><td>0.96</td><td>0.05</td><td>14.13 (15)</td></tr><tr><td> Partial scalar<xref ref-type="table-fn" rid="tfn4">b</xref></td><td>6,919.56</td><td>6,974.17</td><td>158.60*</td><td>116</td><td>0.06 [0.03, 0.08]</td><td>0.98</td><td>0.97</td><td>0.06</td><td>25.04 (32)</td></tr><tr><td> Scalar</td><td>6,966.09</td><td>7,017.24</td><td>215.12<xref ref-type="table-fn" rid="tfn6">**</xref></td><td>121</td><td>0.08 [0.06, 0.10]</td><td>0.95</td><td>0.94</td><td>0.08</td><td>56.52 (5)<xref ref-type="table-fn" rid="tfn6">**</xref></td></tr><tr><td> Strict</td><td>6,961.82</td><td>6,993.61</td><td>266.85<xref ref-type="table-fn" rid="tfn6">**</xref></td><td>149</td><td>0.08 [0.07, 0.10]</td><td>0.93</td><td>0.94</td><td>0.10</td><td>51.73 (28)*</td></tr><tr><td colspan="10">Experimental condition (pre- and posttest): full model<xref ref-type="table-fn" rid="tfn4">d</xref>,<xref ref-type="table-fn" rid="tfn4">e</xref></td></tr><tr><td> Configural</td><td>14,154.99</td><td>14,313.28</td><td>417.33<xref ref-type="table-fn" rid="tfn6">**</xref></td><td>338</td><td>0.05 [0.03, 0.06]</td><td>0.97</td><td>0.96</td><td>0.06</td><td /></tr><tr><td> Metric</td><td>14,146.40</td><td>14,299.16</td><td>424.74<xref ref-type="table-fn" rid="tfn6">**</xref></td><td>346</td><td>0.04 [0.03, 0.06]</td><td>0.97</td><td>0.96</td><td>0.06</td><td>7.41(8)</td></tr><tr><td> Scalar</td><td>14,125.40</td><td>14,267.11</td><td>435.74<xref ref-type="table-fn" rid="tfn6">**</xref></td><td>362</td><td>0.04 [0.02, 0.06]</td><td>0.97</td><td>0.96</td><td>0.06</td><td>11.00 (16)</td></tr><tr><td> Strict</td><td>14,141.11</td><td>14,266.23</td><td>499.45<xref ref-type="table-fn" rid="tfn6">**</xref></td><td>386</td><td>0.05 [0.04, 0.06]</td><td>0.95</td><td>0.94</td><td>0.07</td><td>63.71 (24)<xref ref-type="table-fn" rid="tfn6">**</xref></td></tr></tbody></table> </ephtml> </p> <ulist> <item>3 <emph>Note</emph>. AIC = Akaike information criterion; BIC = Bayesian information criterion; RMSEA = root mean square error of approximation; CFI = comparative fit index, TLI = Tucker Lewis index, and SRMR = standardized root mean square residual.</item> <item>4 Intercepts for the inferencing and reading strategies manifest variables were unconstrained. <sups>b</sups>Intercepts were constrained equal across conditions but not across time. <sups>c</sups>Reduced model including only word reading and reading comprehension. <sups>d</sups>Full model including all variables. <sups>e</sups>Holding partial scalar longitudinal invariance for word reading and reading comprehension measured at pre- and posttest.</item> <item>5 <emph>p</emph> <.05.</item> <item>6 <emph>p</emph> <.001.</item> </ulist> <p>Model fit indices for the DIME model are presented in Table 3 and show that all models provided an adequate fit to the data, and comparable with those of prior systematic replications of the DIME model (e.g., RMSEA = 0.01–0.05; SRMR = 0.02–0.10; CFI = 0.90–1.00). The models explained 10% to 13% of the variance in reading strategies, 25% to 47% in inference-making, and 53% to 56% in reading comprehension at pretest. Model 2 explained 86% to 100% of the variance in reading comprehension at posttest, and 76% to 95% of the variance in word reading at posttest. The unstandardized solution for the measurement models is presented in Table 4. All observed variables loaded adequately on their respective latent variables, with the exception of the SCL strategy measure (e.g., in Model 1 the standardized loading was λ =.12, <emph>SE</emph> =.05, <emph>p</emph> =.01 in Grade 3) as this self-report measure was different from the performance measure of summary writing.</p> <p>Graph</p> <p>Table 3. Model Fit Indices for the DIME Model.</p> <p> <ephtml> <table><colgroup><col align="left" /><col align="char" char="." /><col align="char" char="." /><col align="char" char="." /><col align="char" char="." /><col align="char" char="." /><col align="char" char="." /><col align="char" char="." /><col align="char" char="." /><col align="char" char="." /><col align="char" char="." /><col align="char" char="." /><col align="char" char="." /><col align="char" char="." /></colgroup><thead><tr><th align="center">Model</th><th align="center">AIC</th><th align="center">Adj. BIC</th><th align="center">χ<sup>2</sup></th><th align="center"><italic>df</italic></th><th align="center">RMSEA [90% CI]</th><th align="center">CFI</th><th align="center">TLI</th><th align="center">SRMR</th><th align="center"><italic>R</italic><sup>2</sup> (pretest COMP)</th><th align="center"><italic>R</italic><sup>2</sup> (posttest COMP)</th><th align="center"><italic>R</italic><sup>2</sup> (INF)</th><th align="center"><italic>R</italic><sup>2</sup> (RS)</th><th align="center"><italic>R</italic><sup>2</sup> (posttest WR)</th></tr></thead><tbody><tr><td>Concurrent model (by grade level)</td><td>10,882.85</td><td>10,961.65</td><td>238.86<xref ref-type="table-fn" rid="tfn8">**</xref></td><td>198</td><td>.04 [.02,.06]</td><td>.96</td><td>.96</td><td>.09</td><td>G3:.57G4:.61G5: 1.00</td><td>NA</td><td>G3:.31G4:.31G5:.40</td><td>G3:.11G4:.17G5:.07</td><td>NA</td></tr><tr><td>Change model (by treatment condition)</td><td>14,503.37</td><td>14,578.00</td><td>577.65<xref ref-type="table-fn" rid="tfn8">**</xref></td><td>464</td><td>.05 [.03,.06]</td><td>.95</td><td>.95</td><td>.11</td><td>TX1:.56TX2:.54Con:.53</td><td>TX1:1.00TX2:.86Con:.90</td><td>TX1:.37TX2:.25Con:.47</td><td>TX1:.12TX2:.13Con:.10</td><td>TX1:.95TX2:.85Con:.76</td></tr></tbody></table> </ephtml> </p> <ulist> <item>7 <emph>Note</emph>: The change model includes word reading and reading comprehension measured at pre- and posttest. All other covariates were measured at pretest. DIME = direct and inferential mediation; AIC = Akaike information criterion; BIC = Bayesian information criterion; RMSEA = root mean square error of approximation; CFI = comparative fit index; TLI = Tucker Lewis index; SRMR = standardized root mean square residual; COMP = reading comprehension; INF = inferencing; RS = reading strategies; WR = word reading; TX = treatment (1 = Foundational Skills + Text Processing; 2 = Text Processing); Con = control.</item> <item>8 <emph>p</emph> <.001.</item> </ulist> <p>Graph</p> <p>Table 4. Unstandardized Solution (and Standard Errors) for the Measurement Models.</p> <p> <ephtml> <table><colgroup><col align="left" /><col align="char" char="." /><col align="char" char="." /></colgroup><thead><tr><th align="center">Variable</th><th align="center">Concurrent (Model 1)</th><th align="center">Change (Model 2)</th></tr></thead><tbody><tr><td colspan="3">Inference</td></tr><tr><td>Bridge-IT Far</td><td>0.75 (0.14)</td><td>0.79 (0.13)</td></tr><tr><td colspan="3">Reading strategies</td></tr><tr><td> AWSM Summary 1</td><td>5.01 (2.25)</td><td>5.19 (2.96)</td></tr><tr><td> AWSM Summary 2</td><td>5.29 (2.42)</td><td>5.33 (3.10)</td></tr><tr><td> AWSM Summary 3</td><td>4.71 (2.14)</td><td>5.11 (2.40)</td></tr><tr><td colspan="3">Word reading (pre)</td></tr><tr><td> WJ3 LWID</td><td>1.12 (0.15)</td><td>1.03 (0.12)</td></tr><tr><td colspan="3">Word reading (post)</td></tr><tr><td> WJ3 LWID</td><td>NA</td><td>1.03 (0.12)</td></tr><tr><td colspan="3">Reading comp (pre)</td></tr><tr><td> TOSREC 1</td><td>0.93 (0.11)</td><td>1.02 (0.09)</td></tr><tr><td> TOSREC 2</td><td>1.05 (0.10)</td><td>1.12 (0.09)</td></tr><tr><td colspan="3">Reading comp (post)</td></tr><tr><td> TOSREC 1</td><td>NA</td><td>1.02 (0.09)</td></tr><tr><td> TOSREC 2</td><td>NA</td><td>1.12 (0.09)</td></tr></tbody></table> </ephtml> </p> <p>9 <emph>Note</emph>. The following loadings were fixed to 1.00 for identification of latent variables: Bridge-IT Near (inference), SCL strategies (reading strategies), TOWRE SWE (word reading), Gates–MacGinitie (reading comprehension). AWSM = Assessment of Writing, Self-Monitoring; WJ3 LWID = Woodcock Johnson III Letter Word Identification; TOSREC = Test of Reading Efficiency and Comprehension.</p> <hd id="AN0154068092-26">Dime Model at Baseline</hd> <p></p> <hd id="AN0154068092-27">Direct Effects</hd> <p>Figure 2A shows the standardized results for the structural model by grade, distinguishing between direct effects that were not significant (short-dash lines), significant (solid lines), and significant in one or more grades (long-short dash lines). For ease of interpretation, the spider chart in Figure 2B also presents the pattern of results of DIME components on comprehension, reading strategies, and inferencing, with each dependent-independent variable pair having its own axis, which starts at the center. The values are plotted along the axis and joined, with values further away from the center representing larger direct effects. As shown in the figure, decoding consistently predicted comprehension in Grades 3 to 5. As with prior studies on the DIME model, reading strategies did not significantly predict reading comprehension. However, contrary to our expectations, vocabulary was also not significantly related to reading comprehension after controlling for the direct effects of decoding, inferencing, background knowledge, and strategies. Grade-level differences were found for the path of inferencing to reading comprehension, which was significant in Grades 4 and 5, and larger in Grade 5 compared with Grade 4, suggesting that the role of inferencing becomes more important in later grades. The path from vocabulary to inferencing was significant in Grade 3 only, and the path from vocabulary to strategies was significant in Grade 4 only, suggesting that vocabulary may play a different role at different developmental stages. Strategies predicted inferencing in Grades 3 and 4, but not Grade 5. Similar to vocabulary, background knowledge predicted strategies in Grade 4, but this path was also significant for Grade 5.</p> <p>Graph: Figure 2. Standardized solution for Model 1 (concurrent model at pretest). Note. In Figure 2A, estimates are displayed for Grades 3, 4, and 5. Solid lines = statistically significant in all grades; long-short dash lines = significant in some grades (ns = not significant); short dashed lines = not significant in any grade; WR = word reading; RS = reading strategies, INF = inferencing; COMP = reading comprehension; VOC = vocabulary; BK = background knowledge.</p> <hd id="AN0154068092-28">Indirect Effects</hd> <p>The indirect effect of both background knowledge (β =.11, <emph>p</emph> <.05) and vocabulary on inferencing (β =.12, <emph>p</emph> <.05) was significant in Grade 4, suggesting that vocabulary and background knowledge impacted inferencing via strategies rather than impacting inferencing or reading comprehension directly (see Table 5). The indirect effect of background knowledge on inferencing was also significant in Grade 5, suggesting a complete mediation. No other indirect effects were significant. The correlations among background knowledge, vocabulary, and decoding were also significant only in some grade levels: All of the variables were significantly correlated in Grade 4, and decoding and background knowledge were also correlated in Grade 3. Unlike prior research of the DIME model with older students, the correlation between background knowledge and vocabulary was lower in the current study, as were the correlations between vocabulary and decoding, and background knowledge and decoding because in the present study these skills were not measured in a domain-specific context (i.e., the content of the tests did not overlap with the content of the reading comprehension tests) and because of the restriction inherent in our sampling plan.</p> <p>Graph</p> <p>Table 5. Total and Indirect Effects for Models 1 and 2.</p> <p> <ephtml> <table><colgroup><col align="left" /><col align="char" char="." /><col align="char" char="." /><col align="char" char="." /><col align="char" char="." /><col align="char" char="." /><col align="char" char="." /></colgroup><thead><tr><th /><th align="center" colspan="3">Total effect [95% CI]</th><th align="center" colspan="3">Total indirect effect [95% CI]</th></tr><tr><th align="center">Model 1</th><th align="center">Grade 3</th><th align="center">Grade 4</th><th align="center">Grade 5</th><th align="center">Grade 3</th><th align="center">Grade 4</th><th align="center">Grade 5</th></tr></thead><tbody><tr><td>BK-to-INF</td><td>0.15 [−0.60, 0.36]</td><td><bold><italic>0.20</italic></bold> [0.02, 0.38]</td><td><bold><italic>0.24</italic></bold> [0.05, 0.43]</td><td>0.11 [0.02, 0.26]</td><td><bold><italic>0.11</italic></bold> [0.03, 0.21]</td><td><bold><italic>0.15</italic></bold> [0.04, 0.27]</td></tr><tr><td>VOC-to-INF</td><td><bold>0.38</bold> [0.22, 0.56]</td><td><bold><italic>0.27</italic></bold> [0.10, 0.48]</td><td>0.21 [−0.01, 0.42]</td><td>0.03 [−0.05, 0.11]</td><td><bold><italic>0.12</italic></bold> [0.04, 0.20]</td><td>−0.01 [−0.12, 0.10]</td></tr><tr><td>BK-to-COMP</td><td>0.06 [−0.01, 0.16]</td><td><bold><italic>0.23</italic></bold> [0.07, 0.39]</td><td><bold><italic>0.37</italic></bold> [0.16, 0.54]</td><td>0.09 [−0.11, 0.27]</td><td>0.07 [−0.01, 0.15]</td><td>0.15 [0.00, 0.29]</td></tr><tr><td>VOC-to-COMP</td><td>0.10 [−0.06, 0.27]</td><td>0.15 [−0.09, 0.36]</td><td><bold><italic>0.32</italic></bold> [0.09, 0.51]</td><td>0.10 [−0.10, 0.23]</td><td>0.09 [0.02, 0.18]</td><td>0.13 [−0.02, 0.32]</td></tr><tr><td>RS-to-COMP</td><td>0.16 [0.04, 0.31]</td><td>017 [−0.01, 0.34]</td><td>0.32 [0.08, 0.53]</td><td>0.09 [−0.10, 0.23]</td><td>0.14 [0.03, 0.29]</td><td>0.33 [0.11, 0.71]</td></tr><tr><td>Model 2</td><td>BAU</td><td>TX1</td><td>TX2</td><td>BAU</td><td>TX1</td><td>TX2</td></tr><tr><td>BK-to-INF</td><td><bold>0.27</bold> [0.15, 0.39]</td><td><bold>0.25</bold> [0.14, 0.38]</td><td><bold>0.20</bold> [0.11, 0.30]</td><td><bold>0.15</bold> [0.09, 0.21]</td><td><bold>0.13</bold> [0.08, 0.20]</td><td><bold>0.11</bold> [0.06, 0.16]</td></tr><tr><td>VOC-to-INF</td><td><bold>0.32</bold> [0.19, 0.43]</td><td><bold>0.29</bold> [0.18, 0.40]</td><td><bold>0.26</bold> [0.16, 0.36]</td><td><bold><italic>0.07</italic></bold> [0.01, 0.12]</td><td><bold><italic>0.06</italic></bold> [0.01, 0.11]</td><td><bold><italic>0.06</italic></bold> [0.01, 0.11]</td></tr><tr><td>Reading Comprehension at Pretest</td><td /><td /><td /><td /><td /><td /></tr><tr><td> BK-to-COMP</td><td><bold>0.22</bold> [0.13, 0.34]</td><td><bold>0.22</bold> [0.13, 0.33]</td><td><bold>0.19</bold> [0.11, 0.29]</td><td><bold><italic>0.08</italic></bold><xref ref-type="table-fn" rid="tfn11">a</xref> [0.04, 0.15]</td><td><bold><italic>0.08</italic></bold><xref ref-type="table-fn" rid="tfn11">a</xref> [0.04, 0.14]</td><td><bold><italic>0.07</italic></bold><xref ref-type="table-fn" rid="tfn11">a</xref> [0.03, 0.13]</td></tr><tr><td> VOC-to-COMP</td><td><bold>0.20</bold> [0.10, 0.30]</td><td><bold>0.20</bold> [0.10, 0.30]</td><td><bold>0.19</bold> [0.09, 0.30]</td><td><bold><italic>0.10</italic></bold><xref ref-type="table-fn" rid="tfn11">b</xref> [0.05, 0.16]</td><td><bold><italic>0.10</italic></bold><xref ref-type="table-fn" rid="tfn11">b</xref> [0.05, 0.16]</td><td><bold><italic>0.09</italic></bold><xref ref-type="table-fn" rid="tfn11">b</xref> [0.04, 0.16]</td></tr><tr><td> RS-to-COMP</td><td>0.17 [0.05, 0.32]</td><td>0.16 [0.05, 0.28]</td><td>0.14 [0.04, 0.25]</td><td>0.16 [0.07, 0.29]</td><td>0.15 [0.07, 0.28]</td><td>0.13 [0.06, 0.23]</td></tr><tr><td>Reading Comprehension at Posttest</td><td /><td /><td /><td /><td /><td /></tr><tr><td> WORD-to-COMP</td><td><bold>0.70</bold> [0.51, 0.85]</td><td><bold><italic>0.70</italic></bold> [0.58, 0.81]</td><td><bold><italic>0.39</italic></bold> [0.18, 0.62]</td><td>0.22<xref ref-type="table-fn" rid="tfn11">c</xref> [−0.11, 0.59]</td><td>1.05<xref ref-type="table-fn" rid="tfn11">c</xref> [−0.02, 3.76]</td><td>0.94<xref ref-type="table-fn" rid="tfn11">c</xref> [0.19, 2.85]</td></tr><tr><td> BK-to-COMP</td><td><bold><italic>0.23</italic></bold> [0.08, 0.34]</td><td><bold><italic>0.20</italic></bold> [0.07, 0.33]</td><td><bold>0.32</bold> [0.19, 0.45]</td><td><bold><italic>0.16</italic></bold> [0.09, 0.24]</td><td><bold><italic>0.14</italic></bold> [0.06, 0.22]</td><td><bold>0.16</bold> [0.09, 0.25]</td></tr><tr><td> VOC-to-COMP</td><td><bold><italic>0.20</italic></bold> [0.08, 0.32]</td><td><bold>0.30</bold> [0.15, 0.41]</td><td><bold><italic>0.23</italic></bold> [0.06, 0.37]</td><td><bold><italic>0.14</italic></bold> [0.06, 0.23]</td><td><bold><italic>0.11</italic></bold> [0.04, 0.17]</td><td><bold>0.18</bold> [0.10, 0.26]</td></tr><tr><td> RS-to-COMP</td><td>0.26 [0.11, 0.43]</td><td>0.23 [0.07, 0.39]</td><td>0.24 [0.09, 0.40]</td><td>0.14 [0.01, 0.32]</td><td>0.08 [−0.03, 0.17]</td><td>0.19 [0.10, 0.30]</td></tr><tr><td> INF-to-COMP</td><td>0.26 [0.03, 0.53]</td><td>0.17 [−0.04, 0.33]</td><td><bold><italic>0.51</italic></bold> [0.29, 0.72]</td><td>0.12 [0.05, 0.20]</td><td>0.12 [0.05, 0.22]</td><td>0.13 [0.05, 0.22]</td></tr></tbody></table> </ephtml> </p> <ulist> <item>10 <emph>Note</emph>. Estimates in bold were significant at <emph>p</emph> ≤.001. Estimates in italics were significant at <emph>p</emph> <.05. BK = Background knowledge; INF = inferencing; VOC = vocabulary; COMP = reading comprehension; RS = reading strategies; BAU = Business as Usual; TX1 = Treatment 1 (Foundational and Text Processing Skills); TX2 = Treatment 2 (Text Processing Skills).</item> <item>11 Significant specific indirect effect: Knowledge → Strategies → Inferencing → Comprehension (BAU, TX1 and TX2: β = 0.04, <emph>p</emph> =.04). <sups>b</sups>Significant specific indirect effect: Vocabulary → Inferencing → Comprehension (BAU: β = 0.08, <emph>p</emph> =.03; TX1 and TX2: β = 0.07, <emph>p</emph> =.03). <sups>c</sups>Significant specific indirect effect: Word Reading → Pretest Comprehension → Posttest Comprehension (BAU: β = 0.19, <emph>p</emph> =.001; TX1 and TX2: β = 0.18, <emph>p</emph> =.002).</item> </ulist> <hd id="AN0154068092-29">Change Model</hd> <p></p> <hd id="AN0154068092-30">Pretest</hd> <p>Figure 3 shows the standardized direct effects for Model 2, with the pattern of results for comprehension at posttest plotted in the spider chart (Figure 3B). As noted above, interrelations among the component skills and between these skills and reading comprehension at pretest were similar across conditions by design because struggling readers did not differ in any meaningful way at pretest: the paths from decoding, background knowledge, and inferences to reading comprehension were statistically significant, but the paths from vocabulary and strategies to comprehension were not. In Model 2, background knowledge predicted strategies, and vocabulary and strategies predicted inferences in all groups at pretest. This pattern of results is different from that of Model 1, suggesting that in general these effects are significant for children in upper elementary grades, but that important differences emerge when the sample is disaggregated by grade level. Table 5 presents the total and indirect effects, with significant specific indirect effects noted for indirect effects composed of more than one mediator. Indirect effects of background knowledge (β =.07–.08, <emph>p</emph> <.05) and vocabulary (β =.09–.10, <emph>p</emph> <.05) on comprehension via strategies were significant, as were indirect effects of background knowledge (β =.11–.15, <emph>p</emph> <.001) and vocabulary on inference (β =.06–.07, <emph>p</emph> <.05) via strategies. These results suggest that strategies may play a supporting role of other components rather than a direct role in comprehension.</p> <p>Graph: Figure 3. Standardized solution for Model 2 (change model: pre- and posttest). Note. In Figure 3A, estimates are displayed for control, followed by Treatment 1 and Treatment 2. Gray lines: relations among variables measured at pretest; black lines: relations between pre- and posttest variables. Solid lines: statistically significant in all conditions; long-short dash lines = significant in some conditions (ns = not significant); short dashed lines = not significant in any condition; WR = word reading; RS = reading strategies, INF = inferencing; COMP = reading comprehension; VOC = vocabulary; BK = background knowledge.</p> <hd id="AN0154068092-31">Posttest</hd> <p>As expected, the auto-regressor effects (i.e., pretest to posttest) of comprehension and decoding were significant in all conditions. As shown in Figure 3, the main differences between BAU and treatment conditions were for the path from pretest word reading to posttest comprehension which was significant in the BAU group only. Although the magnitude of the path from posttest decoding to posttest comprehension was larger in both treatment groups than in BAU, this path was not significant in any group. The path from pretest vocabulary to posttest comprehension was significant in TX1, which directly targeted foundational skills, including vocabulary, plus some text processing. Pretest inferencing and background knowledge were related to posttest comprehension in TX2, which focused exclusively on text processing and taught these skills implicitly.</p> <p>Similar to pretest, the indirect effects of background knowledge (β =.14–.16, <emph>p</emph> <.05) and vocabulary (β =.11–.18, <emph>p</emph> <.05) to comprehension via strategies were significant. The total effect of pretest word reading on posttest comprehension was significant, as was the specific indirect effect via pretest comprehension. However, the total indirect effect of pretest word reading on posttest comprehension was not significant in any experimental condition after controlling for all DIME components. The total effect of inferencing on posttest comprehension via pretest comprehension was significant in Grade 5, but the indirect effect was not significant.</p> <hd id="AN0154068092-32">Discussion</hd> <p>The important role of each of the five DIME components has been established for young children with reading difficulties ([<reflink idref="bib8" id="ref87">8</reflink>]; [<reflink idref="bib25" id="ref88">25</reflink>]; [<reflink idref="bib54" id="ref89">54</reflink>]), although there is relatively less research on the higher-order skills (background knowledge, inference, and reading strategies) than the foundational skills (decoding and vocabulary) in this age group. While research has also found a direct link among DIME components in upper elementary grades (e.g., vocabulary and inferencing, [<reflink idref="bib9" id="ref90">9</reflink>]; vocabulary and strategies, [<reflink idref="bib84" id="ref91">84</reflink>]; background knowledge and reading strategies, [<reflink idref="bib78" id="ref92">78</reflink>]; background knowledge and inferencing, [<reflink idref="bib3" id="ref93">3</reflink>]; [<reflink idref="bib10" id="ref94">10</reflink>]; [<reflink idref="bib49" id="ref95">49</reflink>]), more effortful aspects of the comprehension process and interrelations among them are noticeably absent from components skills models of reading comprehension, making it difficult to understand the relative importance of each skill ([<reflink idref="bib50" id="ref96">50</reflink>]). Recent component skills models of reading comprehension have evaluated some, but not all DIME components, and typically included four out of five components with more studies including inferencing and fewer studies including strategies or knowledge as the higher-order components ([<reflink idref="bib22" id="ref97">22</reflink>]; [<reflink idref="bib39" id="ref98">39</reflink>], [<reflink idref="bib40" id="ref99">40</reflink>], [<reflink idref="bib41" id="ref100">41</reflink>]; [<reflink idref="bib42" id="ref101">42</reflink>]; [<reflink idref="bib48" id="ref102">48</reflink>]; [<reflink idref="bib53" id="ref103">53</reflink>]; [<reflink idref="bib56" id="ref104">56</reflink>], [<reflink idref="bib55" id="ref105">55</reflink>]; [<reflink idref="bib68" id="ref106">68</reflink>]; [<reflink idref="bib70" id="ref107">70</reflink>]; [<reflink idref="bib72" id="ref108">72</reflink>]; [<reflink idref="bib79" id="ref109">79</reflink>]). The DIME model takes into consideration the complex and multidimensional nature of reading comprehension, more so than preceding models such as the SVR, but less so than personalized models that also include discourse characteristics (e.g., [<reflink idref="bib29" id="ref110">29</reflink>]). As such, we note that the results of the current study speak to the processing demands on the reading comprehension of struggling readers on average rather than processing demands on individuals of varying profiles (e.g., low knowledge and high strategy skills). We also note that the results pertain to direct and indirect effects among variables rather than changes in the levels of the component skills, but the results have important implications for interventions in one or more areas. In the sections that follow, we discuss the findings of the DIME model replication at pretest and change model that incorporates the effects of treatment.</p> <hd id="AN0154068092-33">DIME Replication</hd> <p>We hypothesized that the DIME model would substantively replicate with this sample of younger struggling readers, but that some relations would differ due to differences in age and reading skill with previous investigations. Specifically, we hypothesized that word reading and vocabulary would prove to be robust predictors of comprehension and that more complex skills would demonstrate relatively weaker, but still significant, relations with reading comprehension. At pretest, decoding was strongly related to reading comprehension across all grades, with the strongest relations in third grade. As expected, our latent factor of reading comprehension (measured by two sentence-level and one text-level standardized measures of comprehension) always exerted processing demands of word reading, even after controlling for higher-order skills. This direct effect was significant in prior DIME studies, apart from the sample of college students ([<reflink idref="bib20" id="ref111">20</reflink>]) and in a weaker model with confounding method bias in a sample of struggling adolescent readers ([<reflink idref="bib2" id="ref112">2</reflink>]). It is possible that the scope of the reading situation is such that texts in upper elementary have higher ease of processing because they are less difficult and therefore require more word reading fluency and less skilled comprehension processes such as reading strategies, particularly in Grade 3 ([<reflink idref="bib49" id="ref113">49</reflink>]; [<reflink idref="bib50" id="ref114">50</reflink>]). However, as we discuss below, the word reading–comprehension relationship was altered in the context of both interventions, suggesting that reading comprehension was driven by other components after receiving implicit instruction.</p> <p>Related to the more complex reading component skills, inferencing was significant predictor of reading comprehension for older students in Grade 5. Research has shown that struggling readers in upper elementary grades are poor at making inferences because they are not able to integrate information locally or produce a global situation model, and they do not generate as many inferences as skilled readers ([<reflink idref="bib10" id="ref115">10</reflink>]). Our findings confirmed the inferencing demands of comprehension in younger students, and that relations among inference (measured by a latent variable comprising both near and far bridging inferences) and comprehension are indeed malleable, as the treatment focused exclusively on text processing bolstered this connection. It is noteworthy that the central role of inferencing of the DIME model was replicated for younger readers in Grade 5 because the relative importance of inferencing compared with other linguistic comprehension processes has not been well established.</p> <p>In previous studies of the DIME model, one consistent finding has been that component skills were directly related to comprehension, but the relations were also mediated through inferencing. In the current study, we expected both a direct path from vocabulary to comprehension and an indirect effect mediated through inferencing, but we found complete mediation as the effect of vocabulary was not significant after controlling for inferencing. This finding is surprising, given the strong link between vocabulary and reading comprehension after controlling for background knowledge and inferencing ([<reflink idref="bib2" id="ref116">2</reflink>]; [<reflink idref="bib19" id="ref117">19</reflink>]; [<reflink idref="bib20" id="ref118">20</reflink>]; [<reflink idref="bib55" id="ref119">55</reflink>], [<reflink idref="bib56" id="ref120">56</reflink>]; [<reflink idref="bib70" id="ref121">70</reflink>]), and the fact that in this study the observed correlations of vocabulary with other measures were low but comparable with studies with younger samples (e.g., [<reflink idref="bib38" id="ref122">38</reflink>]). One notable exception in the current study is that comprehension was driven by vocabulary for the intervention condition that directly targeted vocabulary (FS + TP; see below). Thus, it is possible that we didn't find a significant vocabulary–comprehension relation because children with reading difficulties had depressed vocabulary scores at pretest. Overall, the findings are consistent with those of previous studies that found partial mediation (e.g., [<reflink idref="bib22" id="ref123">22</reflink>]; [<reflink idref="bib42" id="ref124">42</reflink>]; [<reflink idref="bib48" id="ref125">48</reflink>]; [<reflink idref="bib68" id="ref126">68</reflink>]) and support the hypothesis that the vocabulary–reading connection in younger children may be due to a third variable, inference-making, because inferencing taps semantic relations among words and because readers derive meaning from text on the basis of context ([<reflink idref="bib22" id="ref127">22</reflink>]; [<reflink idref="bib80" id="ref128">80</reflink>]).</p> <p>Prior systematic investigations of the DIME model suggest that comprehension is not driven by strategies after controlling for inferencing ([<reflink idref="bib2" id="ref129">2</reflink>]; [<reflink idref="bib19" id="ref130">19</reflink>]; [<reflink idref="bib20" id="ref131">20</reflink>]). This relation was also not supported in the present study likely because cognitive and learning strategies per se are not important, but rather strategies are important because they facilitate integration of information from background knowledge and across different sections of texts. It is possible that when both background knowledge and inferencing are controlled for the role strategies diminishes.</p> <p>Another robust effect was that of strategies on inferencing in Grades 3 and 4, which was consistent with research by Cromley and colleagues with ninth grade and college students, using performance measures of strategy, and with prior research on strategies and inferencing in younger children ([<reflink idref="bib54" id="ref132">54</reflink>]). By contrast, [<reflink idref="bib2" id="ref133">2</reflink>] did not find a relation between strategies and inferences in middle and high school students, likely because that study utilized self-report measures. The present study utilized both self-report strategy use and a performance measure, summary writing, but our latent variable was heavily related to summary writing and not the self-report measure (as evidenced by the magnitude of the factor loadings). Reading strategies is a broad construct encompassing self-regulatory and comprehension-monitoring strategies such as detecting inconsistencies in text, recalling information, paraphrasing, and skimming. Future work is needed to identify strategies that are most effective for inferencing in the context of the DIME model, and better measurement of this construct including mixed methods and multiple strategies.</p> <p>Finally, background knowledge and vocabulary were related to reading strategies in Grade 4, and background knowledge was also related to reading strategies in Grade 5. That reading strategies were driven by background knowledge is not surprising because the summary writing measure of strategies was directly linked to the content of the knowledge test, whereas the inference task was not linked to the content of knowledge or strategy measures. Furthermore, both knowledge and strategies measures (except for the self-report measure) consisted of open-ended question requiring expressive rather than receptive language skills. Thus, in the present study, readers with greater knowledge were likelier to produce better summaries than to make better inferences. Overall, the results of the replication of the DIME model showed important developmental differences in the skills that drive the comprehension of struggling readers.</p> <hd id="AN0154068092-34">Intervention Effects</hd> <p>The second research question investigated whether the two reading interventions disrupted relations within the model. While the treatments did not significantly improve reading comprehension compared with the BAU comparison group ([<reflink idref="bib62" id="ref134">62</reflink>]), we hypothesized that the multicomponent intervention would disrupt how component skills in the reading comprehension system interact with one another, although instruction did not correspond explicitly to the components of the DIME model ([<reflink idref="bib61" id="ref135">61</reflink>]). Prior research on decoding with younger children has found that students learn to utilize component skills such as phonological awareness when these skills are taught directly but also when they are taught implicitly ([<reflink idref="bib26" id="ref136">26</reflink>]). However, the role of component skills on comprehension instruction has not been explicitly demonstrated beyond variables of the SVR ([<reflink idref="bib61" id="ref137">61</reflink>]), and none have used SEM to test direct/indirect effects among the five DIME components within this age group. An evaluation of the effects of intervention on covariances in addition to means is important given the preponderance of empirically sound studies that show null mean effects of intervention ([<reflink idref="bib66" id="ref138">66</reflink>]). In addition, studies often aggregate findings when effects are not discernible across theoretically different treatment conditions (e.g., [<reflink idref="bib31" id="ref139">31</reflink>]). While the current study does not suggest that changes in covariance structure constitutes <emph>response</emph> to intervention, it does suggest that instruction may alter how children utilize both proximal (inferencing and strategies) and distal (decoding, vocabulary, and background knowledge) skills in text comprehension.</p> <p>An important finding centers on the differential predictors of comprehension at posttest depending on the nature of instruction. For students in the BAU condition who did not receive either of the researcher-provided interventions, only pretest decoding had a direct effect on posttest comprehension. For the treatment aimed at building both foundational skills and text processing, pretest vocabulary predicted posttest comprehension, but this path was not significant in the BAU group or in the group receiving exclusively text-processing instruction. This finding is consistent with the findings of [<reflink idref="bib61" id="ref140">61</reflink>], in which treatment appeared to allow participants to utilize more sophisticated component skills (i.e., listening comprehension) in the reading task. In the present study, we tentatively surmise that the TP + FS treatment, by directly intervening on word reading, fluency, and vocabulary, may have (a) supported vocabulary skills in predicting reading comprehension, but also (b) disrupted the relation of word reading and comprehension, freeing students to call upon more complex comprehension processes.</p> <p>For the text-processing intervention group, posttest comprehension was driven by inferencing skill. For the BAU and text processing plus foundational skills groups, posttest comprehension was not driven by inferencing. When considering this difference, it is important to note that the text-processing component of Treatments 1 and 2 was not identical because the interventions incorporated different materials and utilized different instructional routines (see [<reflink idref="bib62" id="ref141">62</reflink>]). In addition, because Treatment 1 included foundational skills and text processing, the amount of text-processing instruction was lower within that treatment. This difference between treatments may partially explain why the intervention with the larger dose of text-processing instruction disrupted the effect of higher-level skills (e.g., inferencing) at the pretest to a greater extent.</p> <p>Related to the auto-regressive portion of the model, posttest decoding and comprehension were driven by pretest decoding and comprehension, respectively, for all groups. However, posttest comprehension was not driven by posttest decoding in any experimental condition, although the magnitude of the estimates was larger than any other estimates for both treatment conditions (see Figure 2B), and close to zero for BAU. That group differences would emerge in the magnitude of posttest decoding for the intervention groups is not surprising, as recent studies with this age group suggest that decoding and fluency skills are more easily remediated than comprehension skills ([<reflink idref="bib51" id="ref142">51</reflink>]; [<reflink idref="bib76" id="ref143">76</reflink>]). However, it is surprising that there were not significant differences between the two researcher-implemented interventions on the measures of word reading (Hedge's <emph>g</emph> = 0.00–0.15, <emph>p</emph> >.05), and the regression of posttest comprehension on decoding, as the text processing plus foundational skills (TX 1) intervention explicitly taught decoding and fluency skills, whereas the text-processing intervention taught these skills only incidentally when students encountered difficulties when reading text. This finding might suggest the fundamental importance of higher-order processes in disrupting the decoding and comprehension link, as comprehension appeared to be driven by decoding in BAU, vocabulary in TX1, and background knowledge and inference in TX2.</p> <hd id="AN0154068092-35">Implications for Interventions</hd> <p>Although by late elementary grades many students have solidified foundational reading skills, a substantial number of struggling readers will continue to require code-based reading instruction ([<reflink idref="bib27" id="ref144">27</reflink>]; [<reflink idref="bib75" id="ref145">75</reflink>]), whereas others will require interventions in higher-order skills (Language and Reading Research Consortium & [<reflink idref="bib43" id="ref146">43</reflink>]). Recent instructional models emphasize explicit skills-based reading instruction. For example, the cognitive-element instructional model emphasizes that outcomes should consist of near-transfer tasks (e.g., inferencing) instead of, or in addition to, far-transfer effects on general outcome measures of reading ([<reflink idref="bib57" id="ref147">57</reflink>]; [<reflink idref="bib74" id="ref148">74</reflink>]). In addition, instruction should be task-specific and assimilated within existing skill-based reading activities. Older instructional models such as continuous progress models allow students to proceed through a well-specified hierarchy of skills ([<reflink idref="bib69" id="ref149">69</reflink>]; [<reflink idref="bib71" id="ref150">71</reflink>]). Students are tested at each level of the hierarchy using subskill mastery assessments before moving on to the next skill. Students who do not master a subskill are then provided small-group or individualized instruction. In specifying the DIME model, Cromley and Azevedo also followed a direct instruction paradigm because each path in the model was supported by at least one true experimental study conducted with students in Grades 4+.</p> <p>Considerable research has been devoted to developing and evaluating the efficacy of "multidimensional, systematic and intense, and linguistically motivated" reading interventions for children in elementary grades that include multiple foundational and some higher-order reading skills ([<reflink idref="bib45" id="ref151">45</reflink>], p. 904). For example, the Triple Focus program provides explicit instruction in meta-cognitive strategies for word identification and reading comprehension ([<reflink idref="bib45" id="ref152">45</reflink>]). However, to our knowledge, there are currently no interventions that systematically treat all lower and higher-level skills prescribed by the DIME model. Findings from the present study have important implications for future systematic multicomponent interventions. Both continuous progress programs and individualized models could (a) record students' progress through a structured, hierarchical set of learning objectives tied to the five DIME components, and (b) provide differentiated instruction on proximal tasks (e.g., inferencing; [<reflink idref="bib33" id="ref153">33</reflink>]; [<reflink idref="bib60" id="ref154">60</reflink>]) once a minimum level of mastery on a component skills is attained (e.g., vocabulary or background knowledge) and before proceeding to a more complex skill (e.g., reading comprehension), thus leveraging intermediate comprehension processes. We note that in the context of multidimensional interventions, assessments of component skills should be psychometrically sound and should include a mixture of measurement approaches that are not limited to testing on the content of the instructional passages. This approach reduces common method bias that can distort results (see [<reflink idref="bib2" id="ref155">2</reflink>]). Future interventions should also provide a better alignment between the theoretical model of change, intervention components, and empirical models tested ([<reflink idref="bib5" id="ref156">5</reflink>]; [<reflink idref="bib24" id="ref157">24</reflink>]). Finally, additional research is needed to understand <emph>inter</emph>- and <emph>intra</emph>-individual differences in covariances among component skills ([<reflink idref="bib29" id="ref158">29</reflink>]). This distinction is important to understanding the expected change in behavior at the person-level due to placement in intervention.</p> <hd id="AN0154068092-36">Limitations</hd> <p>This study should be interpreted considering its limitations, which may influence both the internal and external validity of its findings. First, the interventions implemented were not designed to directly map onto the constructs of the DIME model. A more tailored intervention may have produced more robust disruptions to the means and covariances. In addition, because this intervention occurred with struggling readers, mean scores for this sample were lower than would be observed in peers without reading difficulties, resulting in restriction of range and potentially influencing our results. There are also several limitations related to measurement. First, due to resources and time constraints, the parent study administered a parsimonious battery postintervention with a single indicator per construct. Therefore, we were unable to investigate structural differences in the model postintervention. The lower reliability associated with a single indicator may partially explain some of the low observed correlations in this study, in so far as the reliability of a construct in a SEM is a function of the number of indicators and the size of their loadings ([<reflink idref="bib28" id="ref159">28</reflink>]). In addition, two components of the DIME model were derived from a single experimental measure (AWSM Reader). Thus, items for both components share some overlap in topics and format, and this overlap may partly explain the significant effects of background knowledge and strategies. The large variability in the reading strategies measures might reflect the writing requirements of the summarizing task, which may have affected struggling readers with poor writing skills. Indeed, in the current sample, summary writing was moderately correlated with the Test of Written Language (TOWL; [<reflink idref="bib34" id="ref160">34</reflink>]) at pretest (<emph>r</emph> =.35–.45, <emph>p</emph> <.001). Finally, the current study lacked the power to simultaneously estimate models by grade and condition because the ratio of subjects to number of parameters estimated in a fully unconstrained SEM would be close to 1. Although we adapted rigorous research designs to match the complexity of the educational intervention by using SEM to investigate potential differences introduced by intervention in the covariance structures of the reading comprehension system, our sample size was not large enough to statistically model all complexities of the intervention. For example, personalized models for Phase 1 of the foundational skills plus text-processing intervention would have been informative because students followed a unique sequence through the curriculum. Similarly, cross-classification of writing or self-regulation instruction modality with intervention condition for Phase 2 would have been informative. Future research is needed which directly aligns theory of change, interventions, assessments, and statistical models.</p> <hd id="AN0154068092-37">Conclusion</hd> <p>This study evaluated the DIME model with a younger sample of interest due to established reading risk and the need for effective interventions. As it relates to the three main challenges facing educational research ([<reflink idref="bib35" id="ref161">35</reflink>]), this study deals with the replication crisis in social sciences by systematically replicating a theoretically motivated model which has proven robust in older samples but has not been evaluated for its generalizability to younger, struggling readers in late elementary. Our findings suggest that higher-order skills, particularly inferencing and background knowledge, are more relevant in higher grades, and relations among DIME components appear to be malleable. We showed that multicomponent interventions can support foundational skills such as vocabulary and disrupt the relations among word reading and reading comprehension even when the overall effect size on mean performance of comprehension is negligible. Implicit interventions of higher-order processes can support the use of background knowledge and inferencing and disrupt the relations among word reading and comprehension. We expect that interventions will be more effective if they (a) include direct instruction in higher-order component skills and (b) use component skills actively to improve other skills (e.g., improve vocabulary so students can make better inferences), leading to a more nuanced measurement of response to intervention by focusing not just on reading comprehension outcomes but also prerequisite malleable skills.</p> <hd id="AN0154068092-38">Supplemental Material</hd> <p>sj-docx-1-ldx-10.1177_0022219421995904.docx</p> <p>sj-docx-1-ldx-10.1177_0022219421995904 – Supplemental material for Structure Altering Effects of a Multicomponent Reading Intervention: An Application of the Direct and Inferential Mediation (DIME) Model of Reading Comprehension in Upper Elementary Grades</p> <p></p> <p>Supplemental material, sj-docx-1-ldx-10.1177_0022219421995904 for Structure Altering Effects of a Multicomponent Reading Intervention: An Application of the Direct and Inferential Mediation (DIME) Model of Reading Comprehension in Upper Elementary Grades by Yusra Ahmed, Jeremy Miciak, W. Pat Taylor and David J. Francis in Journal of Learning Disabilities</p> <p></p> <ref id="AN0154068092-39"> <title> References </title> <blist> <bibl id="bib1" idref="ref21" type="bt">1</bibl> <bibtext> Adlof S. M., Catts H. W., Little T. D. (2006). Should the simple view of reading include a fluency component? Reading and Writing, 19, 933–958. https://doi.org/10.1007/s11145-006-9024-z</bibtext> </blist> <blist> <bibl id="bib2" idref="ref12" type="bt">2</bibl> <bibtext> Ahmed Y., Francis D. J., York M., Fletcher J. M., Barnes M., Kulesz P. (2016). Validation of the direct and inferential mediation (DIME) model of reading comprehension in grades 7 through 12. Contemporary Educational Psychology, 44, 68–82. https://doi.org/10.1016/j.cedpsych.2016.02.002</bibtext> </blist> <blist> <bibl id="bib3" idref="ref55" type="bt">3</bibl> <bibtext> Barnes M. A., Dennis M., Haefele-Kalvaitis J. (1996). The effects of knowledge availability and knowledge accessibility on coherence and elaborative inferencing in children from six to fifteen years of age. Journal of Experimental Child Psychology, 61(3), 216–241. https://doi.org/10.1006/jecp.1996.0015</bibtext> </blist> <blist> <bibl id="bib4" idref="ref48" type="bt">4</bibl> <bibtext> Barth A. E., Vaughn S., Capin P., Cho E., Stillman-Spisak S., Martinez L., Kincaid H. (2016). Effects of a text-processing comprehension intervention on struggling middle school readers. Topics in Language Disorders, 36(4), 368–389. https://doi.org/10.1097/TLD.0000000000000101</bibtext> </blist> <blist> <bibl id="bib5" idref="ref156" type="bt">5</bibl> <bibtext> Blase K., Fixsen D. (2013). Core intervention components: Identifying and operationalizing what makes programs work [ASPE Research Brief]. U.S, Department of Health and Human Services.</bibtext> </blist> <blist> <bibl id="bib6" idref="ref83" type="bt">6</bibl> <bibtext> Bollen K. A., Pearl J. (2013). Eight myths about causality and structural equation models. In S. L. Morgan (Ed.), Handbook of causal analysis for social research (pp. 301–328). Springer.</bibtext> </blist> <blist> <bibl id="bib7" idref="ref79" type="bt">7</bibl> <bibtext> Bollen K. A., Stine R. (1990). Direct and indirect effects: Classical and bootstrap estimates of variability. Sociological Methodology, 20, 115–140. https://doi.org/10.2307/271084</bibtext> </blist> <blist> <bibl id="bib8" idref="ref87" type="bt">8</bibl> <bibtext> Cain K., Oakhill J. (2006). Profiles of children with specific reading comprehension difficulties. British Journal of Educational Psychology, 76(4), 683–696. https://doi.org/10.1348/000709905X67610</bibtext> </blist> <blist> <bibl id="bib9" idref="ref90" type="bt">9</bibl> <bibtext> Cain K., Oakhill J. (2014). Reading comprehension and vocabulary: Is vocabulary more important for some aspects of comprehension? L'Année Psychologique, 114(4), 647–662. https://doi.org/10.4074/S0003503314004035</bibtext> </blist> <blist> <bibtext> Cain K., Oakhill J. V., Barnes M. A., Bryant P. E. (2001). Comprehension skill, inference-making ability, and their relation to knowledge. Memory & Cognition, 29(6), 850–859. https://doi.org/10.1037/0022-0663.96.1.31</bibtext> </blist> <blist> <bibtext> Catts H. W. (2018). The simple view of reading: Advancements and false impressions. Remedial and Special Education, 39(5), 317–323. https://doi.org/10.1177/0741932518767563</bibtext> </blist> <blist> <bibtext> Catts H. W., Adlof S. M., Weismer S. E. (2006). Language deficits in poor comprehenders: A case for the simple view of reading. Journal of Speech, Language, and Hearing Research, 49(2), 278–293. https://doi.org/10.1044/1092-4388(2006/023)</bibtext> </blist> <blist> <bibtext> Chall J. S. (1983). Stages of reading development. McGraw-Hill.</bibtext> </blist> <blist> <bibtext> Chall J. S., Jacobs V. A. (1983). Writing and reading in the elementary grades: Developmental trends among low SES children. Language Arts, 60(5), 617–626. <ulink href="http://www.jstor.org/stable/41961511">http://www.jstor.org/stable/41961511</ulink></bibtext> </blist> <blist> <bibtext> Cirino P. T. (2012). The contextual learning scale. Author.</bibtext> </blist> <blist> <bibtext> Cirino P. T., Ahmed Y., Miciak J., Taylor W. P., Gerst E. H., Barnes M. A. (2018). A framework for executive function in the late elementary years. Neuropsychology, 32(2), 176–189. https://doi.org/10.1007/s11145-018-9923-9</bibtext> </blist> <blist> <bibtext> Cirino P. T., Romain M. A., Barth A. E., Tolar T. D., Fletcher J. M., Vaughn S. (2013). Reading skill components and impairments in middle school struggling readers. Reading and Writing, 26(7), 1059–1086. <ulink href="http://doi.org/10.1007/s11145-012-9406-3">http://doi.org/10.1007/s11145-012-9406-3</ulink></bibtext> </blist> <blist> <bibtext> Compton D. L., Miller A. C., Elleman A. M., Steacy L. M. (2014). Have we forsaken reading theory in the name of "quick fix" interventions for children with reading disability? Scientific Studies of Reading, 18(1), 55–73. https://doi.org/10.1080/10888438.2013.836200</bibtext> </blist> <blist> <bibtext> Cromley J. G., Azevedo R. (2007). Testing and refining the direct and inferential mediation model of reading comprehension. Journal of Educational Psychology, 99(2), 311–325. https://doi.org/10.1037/0022-0663.99.2.311</bibtext> </blist> <blist> <bibtext> Cromley J. G., Snyder-Hogan L. E., Luciw-Dubas U. A. (2010). Reading comprehension of scientific text: A domain-specific test of the direct and inferential mediation model of reading comprehension. Journal of Educational Psychology, 102(3), 687–700. https://doi.org/10.1037/a0019452</bibtext> </blist> <blist> <bibtext> Curran P. J., Muthén B. O. (1999). The application of latent curve analysis to testing developmental theories in intervention research. American Journal of Community Psychology, 27(4), 567–595. https://doi.org/10.1023/A:1022137429115</bibtext> </blist> <blist> <bibtext> Daugaard H. T., Cain K., Elbro C. (2017). From words to text: Inference making mediates the role of vocabulary in children's reading comprehension. Reading and Writing, 30(8), 1773–1788. https://doi.org/10.1007/s11145-017-9752-2</bibtext> </blist> <blist> <bibtext> Evans J. J., Floyd R. G., McGrew K. S., Leforgee M. H. (2002). The relations between measures of Cattell-Horn-Carroll (CHC) cognitive abilities and reading achievement during childhood and adolescence. School Psychology Review, 31(2), 246–262. https://doi.org/10.1002/pits.10083</bibtext> </blist> <blist> <bibtext> Fixen D., Blasé K., Naoon S., Duda M. (2013). Implementation drivers: Assessing best practices. National Implementation Science Network.</bibtext> </blist> <blist> <bibtext> Fletcher J. M., Lyon G. R., Fuchs L. S., Barnes M. A. (2018). Learning disabilities: From identification to intervention. Guilford.</bibtext> </blist> <blist> <bibtext> Foorman B. R., Francis D. J., Fletcher J. M., Schatschneider C., Mehta P. (1998). The role of instruction in learning to read: Preventing reading failure in at-risk children. Journal of Educational Psychology, 90(1), 37–55. https://doi.org/10.1037/0022-0663.90.1.37</bibtext> </blist> <blist> <bibtext> Foorman B. R., Koon S., Petscher Y., Mitchell A., Truckenmiller A. (2015). Examining general and specific factors in the dimensionality of oral language and reading in 4th–10th grades. Journal of Educational Psychology, 107(3), 884–899. https://doi.org/10.1037/edu0000026</bibtext> </blist> <blist> <bibtext> Fornell C., Larcker D. F. (1981). Evaluating structural equation models with unobservable variables and measurement error. Journal of Marketing Research, 18(1), 39–50.</bibtext> </blist> <blist> <bibtext> Francis D. J., Kulesz P. A., Benoit J. S. (2018). Extending the simple view of reading to account for variation within readers and across texts: The complete view of reading (CVR i). Remedial and Special Education, 39(5), 274–288. https://doi.org/10.1177/0741932518772904</bibtext> </blist> <blist> <bibtext> Fuchs D., Fuchs L. S. (2019). On the importance of moderator analysis in intervention research: An introduction to the special issue. Exceptional Children, 85(2), 126–128. https://doi.org/10.1177/0014402918811924</bibtext> </blist> <blist> <bibtext> Fuchs D., Hendricks E., Walsh M. E., Fuchs L. S., Gilbert J. K., Zhang Tracy W.,... Peng P. (2018). Evaluating a multidimensional reading comprehension program and reconsidering the lowly reputation of tests of near-transfer. Learning Disabilities Research & Practice, 33(1), 11–23. https://doi.org/10.1111/ldrp.12162</bibtext> </blist> <blist> <bibtext> Gough P. B., Tunmer W. E. (1986). Decoding, reading, and reading disability. Remedial and Special Education, 7(1), 6–10. https://doi.org/10.1177/074193258600700104</bibtext> </blist> <blist> <bibtext> Hall C., Barnes M. A. (2017). Inference instruction to support reading comprehension for elementary students with learning disabilities. Intervention in School and Clinic, 52(5), 279–286. https://doi.org/10.1177/1053451216676799</bibtext> </blist> <blist> <bibtext> Hammill D. D., Larsen S. C. (2009). Test of written language 4 (TOWL-4). Pro-Ed Assessments.</bibtext> </blist> <blist> <bibtext> Hedges L. V. (2018). Challenges in building usable knowledge in education. Journal of Research on Educational Effectiveness, 11(1), 1–21. https://doi.org/10.1080/19345747.2017.1375583</bibtext> </blist> <blist> <bibtext> Hu L. T., Bentler P. M. (1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling: A Multidisciplinary Journal, 6(1), 1–55. <ulink href="http://doi.org/10.1080/10705519909540118">http://doi.org/10.1080/10705519909540118</ulink></bibtext> </blist> <blist> <bibtext> Kaufman A. S., Kaufman N. L. (2004). Kaufman brief intelligence test (2nd ed.). Pearson.</bibtext> </blist> <blist> <bibtext> Kendeou P., Papadopoulos T. C., Spanoudis G. (2012). Processing demands of reading comprehension tests in young readers. Learning and Instruction, 22(5), 354–367. https://doi.org/10.1016/j.learninstruc.2012.02.001</bibtext> </blist> <blist> <bibtext> Kim Y. S. G. (2017). Why the simple view of reading is not simplistic: Unpacking component skills of reading using a direct and indirect effect model of reading (DIER). Scientific Studies of Reading, 21(4), 310–333. https://doi.org/10.1080/10888438.2017.1291643</bibtext> </blist> <blist> <bibtext> Kim Y. S. G. (2020a). Hierarchical and dynamic relations of language and cognitive skills to reading comprehension: Testing the direct and indirect effects model of reading (DIER). Journal of Educational Psychology, 112(4), 667–684. https://doi.org/10.1037/edu0000407</bibtext> </blist> <blist> <bibtext> Kim Y. S. G. (2020b). Toward integrative reading science: The direct and indirect effects model of reading. Journal of Learning Disabilities, 53(6), 469–491. https://doi.org/10.1177/0022219420908239</bibtext> </blist> <blist> <bibtext> Kopatich R. D., Magliano J. P., Millis K. K., Parker C. P., Ray M. (2019). Understanding how language-specific and domain-general resources support comprehension. Discourse Processes, 56(7), 530–552. https://doi.org/10.1080/0163853X.2018.1519358</bibtext> </blist> <blist> <bibtext> Language and Reading Research Consortium, & Logan J. (2017). Pressure points in reading comprehension: A quantile multiple regression analysis. Journal of Educational Psychology, 109(4), 451–464. https://doi.org/10.1037/edu0000150</bibtext> </blist> <blist> <bibtext> Little R. J. (1988). A test of missing completely at random for multivariate data with missing values. Journal of the American Statistical Association, 83, 1198–1202. https://doi.org/10.1080/01621459.1988.10478722</bibtext> </blist> <blist> <bibtext> Lovett M. W., Frijters J. C., Wolf M., Steinbach K. A., Sevcik R. A., Morris R. D. (2017). Early intervention for children at risk for reading disabilities: The impact of grade at intervention and individual differences on intervention outcomes. Journal of Educational Psychology, 109(7), 889–914. https://doi.org/10.1037/edu0000181</bibtext> </blist> <blist> <bibtext> MacKinnon D. P., Lockwood C. M., Hoffman J. M., West S. G., Sheets V. (2002). A comparison of methods to test mediation and other intervening variable effects. Psychological Methods, 7(1), 83–104. https://doi.org/10.1037/1082-989X.7.1.83</bibtext> </blist> <blist> <bibtext> MacGinitie W. H., MacGinitie R. K., Maria K., Dryer L. (2000). Gates-MacGinitie reading tests (4th ed.). Riverside.</bibtext> </blist> <blist> <bibtext> Magliano J. P., Higgs K., Santuzzi A., Tonks S. M., O'Reilly T., Sabatini J., Feller D., Kopatich R., Ray M., Parker C. (2020). Testing the inference mediation hypothesis in a post-secondary context. Contemporary Educational Psychology, 61, Article 101867. https://doi.org/10.1016/j.cedpsych.2020.101867</bibtext> </blist> <blist> <bibtext> McNamara D. S., Floyd R. G., Best R., Louwerse M. (2003). World knowledge driving young readers' comprehension difficulties. In Kafai Y. B., Sandoval W. A., Enyedy N., Nixon A. S., Herrera F. (Eds.), Proceedings of the Sixth International Conference of the Learning Sciences (pp. 326–333). Lawrence Erlbaum.</bibtext> </blist> <blist> <bibtext> McNamara D. S., Magliano J. (2009). Toward a comprehensive model of comprehension. Psychology of Learning and Motivation, 51, 297–384. https://doi.org/10.1016/S0079-7421(09)51009-2</bibtext> </blist> <blist> <bibtext> Miciak J., Roberts G., Taylor W. P., Solis M., Ahmed Y., Vaughn S., Fletcher J. M. (2018). The effects of one versus two years of intensive reading intervention implemented with late elementary struggling readers. Learning Disabilities Research & Practice, 33(1), 24–36. <ulink href="http://doi.org/10.1111/ldrp.12159">http://doi.org/10.1111/ldrp.12159</ulink></bibtext> </blist> <blist> <bibtext> Muthén L. K., Muthén B. (2018). Mplus user's guide (8th ed.).</bibtext> </blist> <blist> <bibtext> Oakhill J., Cain K. (2007). Issues of causality in children's reading comprehension. In McNamara D. S. (Ed.), Reading comprehension strategies: Theories, interventions, and technologies (pp. 47–71). Lawrence Erlbaum.</bibtext> </blist> <blist> <bibtext> Oakhill J., Cain K. (2017). Children with specific text comprehension problems. In Cain K., Compton D., Parrila R. K. (Eds.), Theories of reading development (pp. 359–375). John Benjamins.</bibtext> </blist> <blist> <bibtext> Oslund E. L., Clemens N. H., Simmons D. C., Simmons L. E. (2018). The direct and indirect effects of word reading and vocabulary on adolescents' reading comprehension: Comparing struggling and adequate comprehenders. Reading and Writing, 31(2), 355–379. https://doi.org/10.1007/s11145-017-9788-3</bibtext> </blist> <blist> <bibtext> Oslund E. L., Clemens N. H., Simmons D. C., Smith S. L., Simmons L. E. (2016). How vocabulary knowledge of middle-school students from low socioeconomic backgrounds influences comprehension processes and outcomes. Learning and Individual Differences, 45, 159–165. https://doi.org/10.1016/j.lindif.2015.11.013</bibtext> </blist> <blist> <bibtext> Peng P., Goodrich J. M. (2020). The cognitive element model of reading instruction. Reading Research Quarterly, 55, S77–S88. https://doi.org/10.1002/rrq.336</bibtext> </blist> <blist> <bibtext> Pike M. M., Barnes M. A., Barron R. W. (2010). The role of illustrations in children's inferential comprehension. Journal of Experimental Child Psychology, 105(3), 243–255. https://doi.org/10.1016/j.jecp.2009.10.006</bibtext> </blist> <blist> <bibtext> Preacher K. J., Rucker D. D., Hayes A. F. (2007). Addressing moderated mediation hypotheses: Theory, methods, and prescriptions. Multivariate Behavioral Research, 42(1), 185–227. <ulink href="http://doi.org/10.1080/00273170701341316">http://doi.org/10.1080/00273170701341316</ulink></bibtext> </blist> <blist> <bibtext> Rapp D. N., Broek P. V. D., McMaster K. L., Kendeou P., Espin C. A. (2007). Higher-order comprehension processes in struggling readers: A perspective for research and intervention. Scientific Studies of Reading, 11(4), 289–312. https://doi.org/10.1080/10888430701530417</bibtext> </blist> <blist> <bibtext> Roberts G., Fletcher J. M., Stuebing K. K., Barth A. E., Vaughn S. (2013). Treatment effects for adolescent struggling readers: An application of moderated mediation. Learning and Individual Differences, 23, 10–21. https://doi.org/10.1016/j.lindif.2012.09.008</bibtext> </blist> <blist> <bibtext> Roberts G. J., Capin P., Roberts G., Miciak J., Quinn J. M., Vaughn S. (2018). Examining the effects of afterschool reading interventions for upper elementary struggling readers. Remedial and Special Education, 39(3), 131–143. https://doi.org/10.1177/0741932517750818</bibtext> </blist> <blist> <bibtext> Ryu E., Cheong J. (2017). Comparing indirect effects in different groups in single-group and multi-group structural equation models. Frontiers in Psychology, 8, 747. <ulink href="http://doi.org/10.3389/fpsyg.2017.00747">http://doi.org/10.3389/fpsyg.2017.00747</ulink></bibtext> </blist> <blist> <bibtext> Satorra A., Bentler P. M. (2010). Ensuring positiveness of the scaled difference chi-square test statistic. Psychometrika, 75(2), 243–248. <ulink href="http://doi.org/10.1007/s11336-009-9135-y">http://doi.org/10.1007/s11336-009-9135-y</ulink></bibtext> </blist> <blist> <bibtext> Scammacca N. K., Roberts G. J., Cho E., Williams K. J., Roberts G., Vaughn S. R., Carroll M. (2016). A century of progress: Reading interventions for students in grades 4–12, 1914–2014. Review of Educational Research, 86(3), 756–800. https://doi.org/10.3102/0034654316652942</bibtext> </blist> <blist> <bibtext> Scammacca N. K., Roberts G., Vaughn S., Stuebing K. K. (2015). A meta-analysis of interventions for struggling readers in grades 4–12: 1980–2011. Journal of Learning Disabilities, 48(4), 369–390. https://doi.org/10.1177/0022219413504995</bibtext> </blist> <blist> <bibtext> Scarborough H. S., Neuman S., Dickinson D. (2009). Connecting early language and literacy to later reading (dis)abilities: Evidence, theory, and practice. In Fletcher-Campbell F., Soler J., Reid G. (Eds.), Approaching difficulties in literacy development: Assessment, pedagogy and programmes (pp. 23–38). SAGE.</bibtext> </blist> <blist> <bibtext> Segers E., Verhoeven L. (2016). How logical reasoning mediates the relation between lexical quality and reading comprehension. Reading and Writing, 29(4), 577–590. https://doi.org/10.1007/s11145-015-9613-9</bibtext> </blist> <blist> <bibtext> Slavin R. E., Karweit N. L., Madden N. A. (1989). Effective programs for students at risk. Allyn & Bacon.</bibtext> </blist> <blist> <bibtext> Smith S. R. (2013). Testing a multicomponent model of reading comprehension for seventh- and eighth-grade students (Doctoral dissertation). Texas A&M University Libraries.</bibtext> </blist> <blist> <bibtext> Szadokierski I., Burns M. K., McComas J. J. (2017). Predicting intervention effectiveness from reading accuracy and rate measures through the instructional hierarchy: Evidence for a skill-by-treatment interaction. School Psychology Review, 46(2), 190–200. https://doi.org/10.17105/SPR-2017-0013.V46-2</bibtext> </blist> <blist> <bibtext> Tarchi C. (2010). Reading comprehension of informative texts in secondary school: A focus on direct and indirect effects of reader's prior knowledge. Learning and Individual Differences, 20(5), 415–420. https://doi.org/10.1016/j.lindif.2010.04.002</bibtext> </blist> <blist> <bibtext> Torgesen J. K., Wagner R. K., Rashotte C. A. (1999). TOWRE: Test of word reading efficiency. Pro-Ed.</bibtext> </blist> <blist> <bibtext> Van Norman E. R., Maki K. E., Burns M. K., McComas J. J., Helman L. (2018). Comparison of progress monitoring data from general outcome measures and specific subskill mastery measures for reading. Journal of School Psychology, 67, 179–189. https://doi.org/10.1016/j.jsp.2018.02.002</bibtext> </blist> <blist> <bibtext> Vaughn S., Roberts G. J., Miciak J., Taylor P., Fletcher J. M. (2019). Efficacy of a word-and text-based intervention for students with significant reading difficulties. Journal of Learning Disabilities, 52(1), 31–44. https://doi.org/10.1177/0022219418775113</bibtext> </blist> <blist> <bibtext> Vaughn S., Solís M., Miciak J., Taylor W. P., Fletcher J. M. (2016). Effects from a randomized control trial comparing researcher and school-implemented treatments with fourth graders with significant reading difficulties. Journal of Research on Educational Effectiveness, 9(1), 23–44. https://doi.org/10.1080/19345747.2015.1126386</bibtext> </blist> <blist> <bibtext> Vaughn S., Swanson E. A., Roberts G., Wanzek J., Stillman-Spisak S. J., Solis M., Simmons D. (2013). Improving reading comprehension and social studies knowledge in middle school. Reading Research Quarterly, 48(1), 77–93. https://doi.org/10.1002/rrq.039</bibtext> </blist> <blist> <bibtext> Vitale M. R., Romance N. R. (2007). A knowledge-based framework for unifying content-area reading comprehension and reading comprehension strategies. In McNamara D. S. (Ed.), Reading comprehension strategies: Theories, interventions, and technologies (pp. 73–104). Lawrence Erlbaum.</bibtext> </blist> <blist> <bibtext> Völlinger V. A., Spörer N., Lubbe D., Brunstein J. C. (2018). A path analytic test of the reading strategies mediation model: Relating cognitive competences and motivational influences to individual differences in fifth-grade students' reading comprehension. The Journal of Educational Research, 111(6), 733–745. https://doi.org/10.1080/00220671.2017.1412930</bibtext> </blist> <blist> <bibtext> Wagner R. K., Muse A. E., Tannenbaum K. R. (2007). Vocabulary acquisition: Implications for reading comprehension. Guilford Press.</bibtext> </blist> <blist> <bibtext> Wagner R. K., Torgesen J. K., Rashotte C. A. (2010). Test of silent reading efficiency and comprehension (TOSREC). Pro-Ed.</bibtext> </blist> <blist> <bibtext> Wijekumar K. K., Meyer B. J., Lei P. (2012). Large-scale randomized controlled trial with 4th graders using intelligent tutoring of the structure strategy to improve nonfiction reading comprehension. Educational Technology Research and Development, 60(6), 987–1013. https://doi.org/10.1007/s11423-012-9263-4</bibtext> </blist> <blist> <bibtext> Woodcock R. W., McGrew K., Mather N. (2007). Woodcock-Johnson tests of achievement (3rd ed.). Riverside.</bibtext> </blist> <blist> <bibtext> Zargar E., Adams A. M., Connor C. M. (2020). The relations between children's comprehension monitoring and their reading comprehension and vocabulary knowledge: An eye-movement study. Reading and Writing, 33(3), 511–545. https://doi.org/10.1007/s11145-019-09966-3</bibtext> </blist> </ref> <ref id="AN0154068092-40"> <title> Footnotes </title> <blist> <bibtext> The opinions expressed in this article are the authors' and do not represent the views of the National Institute of Child Health and Human Development.</bibtext> </blist> <blist> <bibtext> The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.</bibtext> </blist> <blist> <bibtext> This work was supported by the Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD) under Grants R01HD096262 and P50HD052117.</bibtext> </blist> <blist> <bibtext> Supplementary material for this article is available on the <emph>Journal of Learning Disabilities</emph> website with the online version of this article.</bibtext> </blist> </ref> <aug> <p>By Yusra Ahmed; Jeremy Miciak; W. Pat Taylor and David J. Francis</p> <p>Reported by Author; Author; Author; Author</p> </aug> <nolink nlid="nl1" bibid="bib35" firstref="ref1"></nolink> <nolink nlid="nl2" bibid="bib12" firstref="ref2"></nolink> <nolink nlid="nl3" bibid="bib18" firstref="ref3"></nolink> <nolink nlid="nl4" bibid="bib76" firstref="ref4"></nolink> <nolink nlid="nl5" bibid="bib66" firstref="ref5"></nolink> <nolink nlid="nl6" bibid="bib65" firstref="ref6"></nolink> <nolink nlid="nl7" bibid="bib30" firstref="ref9"></nolink> <nolink nlid="nl8" bibid="bib29" firstref="ref10"></nolink> <nolink nlid="nl9" bibid="bib21" firstref="ref11"></nolink> <nolink nlid="nl10" bibid="bib19" firstref="ref13"></nolink> <nolink nlid="nl11" bibid="bib20" firstref="ref14"></nolink> <nolink nlid="nl12" bibid="bib13" firstref="ref15"></nolink> <nolink nlid="nl13" bibid="bib51" firstref="ref17"></nolink> <nolink nlid="nl14" bibid="bib32" firstref="ref20"></nolink> <nolink nlid="nl15" bibid="bib27" firstref="ref22"></nolink> <nolink nlid="nl16" bibid="bib11" firstref="ref23"></nolink> <nolink nlid="nl17" bibid="bib61" firstref="ref26"></nolink> <nolink nlid="nl18" bibid="bib23" firstref="ref28"></nolink> <nolink nlid="nl19" bibid="bib39" firstref="ref31"></nolink> <nolink nlid="nl20" bibid="bib60" firstref="ref32"></nolink> <nolink nlid="nl21" bibid="bib67" firstref="ref33"></nolink> <nolink nlid="nl22" bibid="bib42" firstref="ref34"></nolink> <nolink nlid="nl23" bibid="bib50" firstref="ref35"></nolink> <nolink nlid="nl24" bibid="bib62" firstref="ref38"></nolink> <nolink nlid="nl25" bibid="bib17" firstref="ref41"></nolink> <nolink nlid="nl26" bibid="bib81" firstref="ref45"></nolink> <nolink nlid="nl27" bibid="bib333" firstref="ref51"></nolink> <nolink nlid="nl28" bibid="bib317" firstref="ref52"></nolink> <nolink nlid="nl29" bibid="bib340" firstref="ref53"></nolink> <nolink nlid="nl30" bibid="bib205" firstref="ref56"></nolink> <nolink nlid="nl31" bibid="bib77" firstref="ref63"></nolink> <nolink nlid="nl32" bibid="bib82" firstref="ref64"></nolink> <nolink nlid="nl33" bibid="bib47" firstref="ref65"></nolink> <nolink nlid="nl34" bibid="bib83" firstref="ref67"></nolink> <nolink nlid="nl35" bibid="bib73" firstref="ref68"></nolink> <nolink nlid="nl36" bibid="bib37" firstref="ref69"></nolink> <nolink nlid="nl37" bibid="bib58" firstref="ref70"></nolink> <nolink nlid="nl38" bibid="bib15" firstref="ref72"></nolink> <nolink nlid="nl39" bibid="bib16" firstref="ref73"></nolink> <nolink nlid="nl40" bibid="bib52" firstref="ref74"></nolink> <nolink nlid="nl41" bibid="bib36" firstref="ref75"></nolink> <nolink nlid="nl42" bibid="bib64" firstref="ref76"></nolink> <nolink nlid="nl43" bibid="bib63" firstref="ref77"></nolink> <nolink nlid="nl44" bibid="bib59" firstref="ref78"></nolink> <nolink nlid="nl45" bibid="bib46" firstref="ref80"></nolink> <nolink nlid="nl46" bibid="bib26" firstref="ref81"></nolink> <nolink nlid="nl47" bibid="bib14" firstref="ref82"></nolink> <nolink nlid="nl48" bibid="bib252" firstref="ref85"></nolink> <nolink nlid="nl49" bibid="bib44" firstref="ref86"></nolink> <nolink nlid="nl50" bibid="bib25" firstref="ref88"></nolink> <nolink nlid="nl51" bibid="bib54" firstref="ref89"></nolink> <nolink nlid="nl52" bibid="bib84" firstref="ref91"></nolink> <nolink nlid="nl53" bibid="bib78" firstref="ref92"></nolink> <nolink nlid="nl54" bibid="bib10" firstref="ref94"></nolink> <nolink nlid="nl55" bibid="bib49" firstref="ref95"></nolink> <nolink nlid="nl56" bibid="bib22" firstref="ref97"></nolink> <nolink nlid="nl57" bibid="bib40" firstref="ref99"></nolink> <nolink nlid="nl58" bibid="bib41" firstref="ref100"></nolink> <nolink nlid="nl59" bibid="bib48" firstref="ref102"></nolink> <nolink nlid="nl60" bibid="bib53" firstref="ref103"></nolink> <nolink nlid="nl61" bibid="bib56" firstref="ref104"></nolink> <nolink nlid="nl62" bibid="bib55" firstref="ref105"></nolink> <nolink nlid="nl63" bibid="bib68" firstref="ref106"></nolink> <nolink nlid="nl64" bibid="bib70" firstref="ref107"></nolink> <nolink nlid="nl65" bibid="bib72" firstref="ref108"></nolink> <nolink nlid="nl66" bibid="bib79" firstref="ref109"></nolink> <nolink nlid="nl67" bibid="bib38" firstref="ref122"></nolink> <nolink nlid="nl68" bibid="bib80" firstref="ref128"></nolink> <nolink nlid="nl69" bibid="bib31" firstref="ref139"></nolink> <nolink nlid="nl70" bibid="bib75" firstref="ref145"></nolink> <nolink nlid="nl71" bibid="bib43" firstref="ref146"></nolink> <nolink nlid="nl72" bibid="bib57" firstref="ref147"></nolink> <nolink nlid="nl73" bibid="bib74" firstref="ref148"></nolink> <nolink nlid="nl74" bibid="bib69" firstref="ref149"></nolink> <nolink nlid="nl75" bibid="bib71" firstref="ref150"></nolink> <nolink nlid="nl76" bibid="bib45" firstref="ref151"></nolink> <nolink nlid="nl77" bibid="bib33" firstref="ref153"></nolink> <nolink nlid="nl78" bibid="bib24" firstref="ref157"></nolink> <nolink nlid="nl79" bibid="bib28" firstref="ref159"></nolink> <nolink nlid="nl80" bibid="bib34" firstref="ref160"></nolink>
Header DbId: eric
DbLabel: ERIC
An: EJ1321503
AccessLevel: 3
PubType: Academic Journal
PubTypeId: academicJournal
PreciseRelevancyScore: 0
IllustrationInfo
Items – Name: Title
  Label: Title
  Group: Ti
  Data: Structure Altering Effects of a Multicomponent Reading Intervention: An Application of the Direct and Inferential Mediation (DIME) Model of Reading Comprehension in Upper Elementary Grades
– Name: Language
  Label: Language
  Group: Lang
  Data: English
– Name: Author
  Label: Authors
  Group: Au
  Data: <searchLink fieldCode="AR" term="%22Ahmed%2C+Yusra%22">Ahmed, Yusra</searchLink><br /><searchLink fieldCode="AR" term="%22Miciak%2C+Jeremy%22">Miciak, Jeremy</searchLink><br /><searchLink fieldCode="AR" term="%22Taylor%2C+W%2E+Pat%22">Taylor, W. Pat</searchLink><br /><searchLink fieldCode="AR" term="%22Francis%2C+David+J%2E%22">Francis, David J.</searchLink>
– Name: TitleSource
  Label: Source
  Group: Src
  Data: <searchLink fieldCode="SO" term="%22Journal+of+Learning+Disabilities%22"><i>Journal of Learning Disabilities</i></searchLink>. Jan-Feb 2022 55(1):58-78.
– Name: Avail
  Label: Availability
  Group: Avail
  Data: SAGE Publications and Hammill Institute on Disabilities. 2455 Teller Road, Thousand Oaks, CA 91320. Tel: 800-818-7243; Tel: 805-499-9774; Fax: 800-583-2665; e-mail: journals@sagepub.com; Web site: http://sagepub.com
– Name: PeerReviewed
  Label: Peer Reviewed
  Group: SrcInfo
  Data: Y
– Name: Pages
  Label: Page Count
  Group: Src
  Data: 21
– Name: DatePubCY
  Label: Publication Date
  Group: Date
  Data: 2022
– Name: SourceSuprt
  Label: Sponsoring Agency
  Group: SrcSuprt
  Data: Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD) (DHHS/NIH)
– Name: NumberContract
  Label: Contract Number
  Group: NumCntrct
  Data: R01HD096262<br />P50HD052117
– Name: TypeDocument
  Label: Document Type
  Group: TypDoc
  Data: Journal Articles<br />Reports - Research
– Name: Audience
  Label: Education Level
  Group: Audnce
  Data: <searchLink fieldCode="EL" term="%22Elementary+Education%22">Elementary Education</searchLink><br /><searchLink fieldCode="EL" term="%22Early+Childhood+Education%22">Early Childhood Education</searchLink><br /><searchLink fieldCode="EL" term="%22Grade+3%22">Grade 3</searchLink><br /><searchLink fieldCode="EL" term="%22Primary+Education%22">Primary Education</searchLink><br /><searchLink fieldCode="EL" term="%22Grade+4%22">Grade 4</searchLink><br /><searchLink fieldCode="EL" term="%22Intermediate+Grades%22">Intermediate Grades</searchLink><br /><searchLink fieldCode="EL" term="%22Grade+5%22">Grade 5</searchLink><br /><searchLink fieldCode="EL" term="%22Middle+Schools%22">Middle Schools</searchLink>
– Name: Subject
  Label: Descriptors
  Group: Su
  Data: <searchLink fieldCode="DE" term="%22Reading+Instruction%22">Reading Instruction</searchLink><br /><searchLink fieldCode="DE" term="%22Reading+Comprehension%22">Reading Comprehension</searchLink><br /><searchLink fieldCode="DE" term="%22Elementary+School+Students%22">Elementary School Students</searchLink><br /><searchLink fieldCode="DE" term="%22Reading+Difficulties%22">Reading Difficulties</searchLink><br /><searchLink fieldCode="DE" term="%22Grade+3%22">Grade 3</searchLink><br /><searchLink fieldCode="DE" term="%22Grade+4%22">Grade 4</searchLink><br /><searchLink fieldCode="DE" term="%22Grade+5%22">Grade 5</searchLink><br /><searchLink fieldCode="DE" term="%22Intervention%22">Intervention</searchLink><br /><searchLink fieldCode="DE" term="%22Instructional+Effectiveness%22">Instructional Effectiveness</searchLink><br /><searchLink fieldCode="DE" term="%22After+School+Programs%22">After School Programs</searchLink>
– Name: SubjectThesaurus
  Label: Assessment and Survey Identifiers
  Group: Su
  Data: <searchLink fieldCode="SU" term="%22Gates+MacGinitie+Reading+Tests%22">Gates MacGinitie Reading Tests</searchLink><br /><searchLink fieldCode="SU" term="%22Woodcock+Johnson+Tests+of+Achievement%22">Woodcock Johnson Tests of Achievement</searchLink><br /><searchLink fieldCode="SU" term="%22Kaufman+Brief+Intelligence+Test%22">Kaufman Brief Intelligence Test</searchLink>
– Name: DOI
  Label: DOI
  Group: ID
  Data: 10.1177/0022219421995904
– Name: ISSN
  Label: ISSN
  Group: ISSN
  Data: 0022-2194
– Name: Abstract
  Label: Abstract
  Group: Ab
  Data: We evaluate the direct and inferential mediation (DIME) model for reading comprehension with a sample of struggling readers in Grades 3 to 5 (N = 364) in the context of a large-scale randomized controlled trial (RCT) investigating two theoretically distinct reading interventions (text processing + foundational skills [n = 117] or text processing only [n = 120]) and a control condition (n = 127). We investigate whether the intervention affects not just reading comprehension levels, but also how variables within the reading system interrelate. This approach allows the focus to shift from intervention as influencing a change in reading comprehension status to a complex set of processes. We fit structural equation models (SEMs) to evaluate the DIME model at baseline and a change model that included reading comprehension and word reading at posttest. There were no significant mean differences between groups in reading comprehension. However, significant differences emerged on the direct and indirect effects of background knowledge, vocabulary, word reading, strategies, and inferencing on comprehension across grade levels and treatment conditions. Related to treatment groups, background knowledge, vocabulary, and inferencing were significantly related to comprehension at posttest for students who received text processing and/or foundational skills interventions. The results have implications for the direct instruction of higher-order reading skills in the context of multicomponent interventions.
– Name: AbstractInfo
  Label: Abstractor
  Group: Ab
  Data: As Provided
– Name: DateEntry
  Label: Entry Date
  Group: Date
  Data: 2022
– Name: AN
  Label: Accession Number
  Group: ID
  Data: EJ1321503
PLink https://search.ebscohost.com/login.aspx?direct=true&site=eds-live&db=eric&AN=EJ1321503
RecordInfo BibRecord:
  BibEntity:
    Identifiers:
      – Type: doi
        Value: 10.1177/0022219421995904
    Languages:
      – Text: English
    PhysicalDescription:
      Pagination:
        PageCount: 21
        StartPage: 58
    Subjects:
      – SubjectFull: Reading Instruction
        Type: general
      – SubjectFull: Reading Comprehension
        Type: general
      – SubjectFull: Elementary School Students
        Type: general
      – SubjectFull: Reading Difficulties
        Type: general
      – SubjectFull: Grade 3
        Type: general
      – SubjectFull: Grade 4
        Type: general
      – SubjectFull: Grade 5
        Type: general
      – SubjectFull: Intervention
        Type: general
      – SubjectFull: Instructional Effectiveness
        Type: general
      – SubjectFull: After School Programs
        Type: general
      – SubjectFull: Gates MacGinitie Reading Tests
        Type: general
      – SubjectFull: Woodcock Johnson Tests of Achievement
        Type: general
      – SubjectFull: Kaufman Brief Intelligence Test
        Type: general
    Titles:
      – TitleFull: Structure Altering Effects of a Multicomponent Reading Intervention: An Application of the Direct and Inferential Mediation (DIME) Model of Reading Comprehension in Upper Elementary Grades
        Type: main
  BibRelationships:
    HasContributorRelationships:
      – PersonEntity:
          Name:
            NameFull: Ahmed, Yusra
      – PersonEntity:
          Name:
            NameFull: Miciak, Jeremy
      – PersonEntity:
          Name:
            NameFull: Taylor, W. Pat
      – PersonEntity:
          Name:
            NameFull: Francis, David J.
    IsPartOfRelationships:
      – BibEntity:
          Dates:
            – D: 01
              M: 01
              Type: published
              Y: 2022
          Identifiers:
            – Type: issn-print
              Value: 0022-2194
          Numbering:
            – Type: volume
              Value: 55
            – Type: issue
              Value: 1
          Titles:
            – TitleFull: Journal of Learning Disabilities
              Type: main
ResultId 1