This project conducts experimental studies with three objectives: (1) To determine the extent to which variations in knowledge availability and accessibility influence the rate and accuracy of forming inferences. (2) To investigate whether individual difference factors are potential sources of variability in explaining inference making and reading comprehension. (3)To determine whether systematic retrieval practice improves both knowledge availability and accessibility as well as the accuracy and rate of forming inferences using the knowledge-base. Study 1 Overview. Procedures modify those used by Barnes et al. (1996). In Phase 1, students learn a knowledge-base of 14 facts about Gan. In Phase 2, students will read and listen to an informational text that consists of six sections. A computer asks students comprehension questions that tap information from that text section and the taught knowledge base. Phase 3 retests retention of the knowledge-base. One week later, Phase 4 retests the knowledge-base and students take a larger test battery. All students will complete Phases 1-4 individually. Study participants (n = 300) will come from Storm Lake Community School District students in grades 5-8 (n = 657) with both males and females recruited for participation. The school district will provide demographic data, reading scores from the Iowa Assessment for Reading, and Gates MacGinities Reading Test for all participants from which the investigator will randomly select students for the treatment or control group. The investigator will replace students who exit the sample with students from the same grade who consented but not randomly selected for participation. Participation will exclude students with significant cognitive impairments who participate in the Life Skills Class (e.g., severe autism, behavior, cognitive disability) due to task demands. Additional Assessments measure the constructs of reading comprehension, inference making, word reading efficiency, metacognition, working memory, and background knowledge. Multiple measures of each construct allow the study to identify (a) relations among latent abilities; (b) role of these latent abilities in establishing reading comprehension, independent of measure-specific factors or measurement error; and (c) individual differences that lead to variation in performance. Data Analysis. The investigator will screen for missing data and outliers and evaluate the distributional properties of observed variables, making transformations as necessary and appropriate. If possible, the investigator will correlate non-plausible values or list them as missing if not possible, and label extreme data points. Power. General recommendations on sample sizes in structural equation modeling (SEM) suggest minimum samples of 300 (i.e., 75 participants per grades 5-8). This sample size will allow the research team to detect correlations among constructs greater than 0.05 and regression coefficients in latent variable regressions of 0.05 with over 80% power. Consideration of Biological Variables. To sufficiently power the study and generalize findings, this study recruits males and females for participation and collects gender, race, disability status, and socioeconomic status to determine if these child-attributes moderate knowledge accessibility and availability. Analytic Approach. The investigator will model the effects of knowledge availability and accessibility on inference making without dichotomizing the distribution of comprehension. To determine whether reader characteristics (i.e., knowledge availability and accessibility, metacognition, reading comprehension, word reading efficiency, background knowledge, working memory, and grade) make unique contributions to inference making, the investigator will fit mixed effects explanatory item response models to the trial-by-trial inference making accuracy and speed data with item-level accuracy and reading speeds cross-classified between persons and items. The investigator will also nest people within grade, treating grouping variables for people such as grade and items (i.e., literal, word-to-text, and knowledge-to-text inferences) as fixed factors. If limited variability exists in accuracy data, the investigator will only model response time data. The investigator will also evaluate how organization of knowledge relates to knowledge accessibility and inference making. Structural equation modeling will evaluate relations among the constructs (i.e., comprehension, word efficiency, inference making, knowledge accessibility and availability, metacognition, and working memory) to determine the role of knowledge accessibility and availability in reading comprehension and factors influencing knowledge accessibility and availability. Using Cain et al. (2001) framework, the investigator will classify each inference failure as (1) incorrect text-premise recall (i.e., the participant fails to retrieve the relevant premise from the text); (2) incorrect knowledge-based recall (i.e., the participant fails to retrieve the relevant premise from the knowledge-base); and (3) integration failure (i.e., the participant fails to integrate text content with knowledge). A traditional two-way ANOVA (skill group X inference type) on the incorrect inference response will occur with skilled readers defined as achieving proficiency on the Iowa Assessment for Reading and performance above a standard score of 90 on the Gates MacGinitie Reading Test. The investigator will only compare the means for skilled versus less skilled readers. The investigators hypothesize that: (1) less skilled middle grade readers will make more errors than skilled readers; (2) less skilled readers will not always recognize the need to form inferences during reading relative to skilled readers; and (3) no significant differences will exist across the middle grades (Ahmed et al., 2016; Barth et al., 2015; Barnes et al., 2015). Results of Aim 1 will tell us if less skilled readers need the duration of retrieval practice increased to achieve similar benefits as more skilled readers and if less skilled readers need help organizing the taught knowledge-base so that it is more available and accessible when reading informational texts, which will inform Aim 2 - To examine the extent to which retrieval practice (i.e., spaced practice testing) increases knowledge availability and accessibility and improves the accuracy and rate of forming inferences using that knowledge-base among middle grade readers. Study 2 uses procedures modified from those used by Butler (2010). In Phase 1, students learn a knowledge-base of 14 facts about Gan. In Phase 2, students read and study the six-section informational texts with reassessment of the student's retention of the knowledge base immediately after studying these 6 texts. In Phase 3, students restudy two of the text sections (restudy passages), repeatedly take the same test on another two text sections (same test), and repeatedly take different tests on the other two text sections (variable test). Immediately after interacting with all six text sections, the project reassesses the retention of the knowledge base. Phase 4 occurs one week later and again reassesses the students' retention of the knowledge-base and administers the same additional test battery as in Aim 1. Study participants (n = 180) will come from Storm Lake Community School District students in grade 5-8. The investigators anticipate that students will come from diverse socio-economic and ethnic backgrounds. Student Selection and Exclusions. In the fall of 2019, laboratory assistants will recruit students and screen them with the Gates-MacGinitie Reading Test. Students selected for participation: (1) perform below a standard score of 90 on the Gates-MacGinitie Reading Test; (2) did not participate in Aim 1; and (3) are not enrolled in life skills classes (i.e., severe Autism, behavior, cognitive impairments) due to task appropriateness. The investigator will document any additional services that students receive (i.e., Title, special education, and reading support) and included them as contextual variables in later analyses. Power and Design. The investigators estimate the sample associated with an effect size of 0.25, alpha 0.05, and with over 80% power at 136 participants for between group differences, 36 for within group differences, and 180 for the interaction. Randomization will occur within grade, with all qualified students in grade 5 available for randomization because these students did not participate in Aim 1. The investigators estimate attrition to be 20% or 36 participants total and will replace participants who exit the pool of consented students not randomly selected for participation. Consideration of Biological Variables. To sufficiently power the study and generalize findings, the study recruits males and females for participation and collects gender, race, disability status, and socioeconomic status to determine if these child-attributes moderate knowledge accessibility and availability. Data Analysis. As in Aim 1, the investigators will screen for missing data and outliers and evaluate the distributional properties of observed variables, making transformations as necessary and appropriate. When possible, investigators will correct non-plausible values or listed them as missing if investigators cannot correct them and label extreme data points. Analytic Approach. Data analysis will use repeated measures analysis of variance models to examine the effect of the experimental condition on knowledge retention. Separate ANOVAs will analyze scores obtained on the question types (i.e., explicit text-, word-to-text- and inference-questions) and the different error types by condition. Results of Aim 2 (and Aim 1) and additional analyses will inform the independent research projects conducted by undergraduate laboratory assistants (Aim 3). Aim 3. To integrate investigative research into an undergraduate Honors Research Program. This proposal will support the development of a laboratory component for the Honors Proposal Seminar. This sophomore-level seminar focuses on the organizational structure of their research proposal (i.e., introduction, hypotheses, and method section). The laboratory component will provide students hands-on experiences in the practices of conducting applied research in the schools. Students will also have access to extant data to examine interesting questions related to the reading and reading-related skills of adolescents which should heighten student interest and enthusiasm for conducting research among adolescents. Laboratory Outline. This project will create a two-semester laboratory course (e.g., 32 sessions) that meets weekly for two hours. Lab sessions will address Institutional Review Board applications; appropriate engagement with school personnel; subject recruitment; parent/guardian consent; child assent; formation of research questions and hypotheses; fidelity of implementation; experimental design; inputting and verifying data; analyzing data; and discussing, reporting, and publishing findings (i.e., manuscripts, presentations, and spoken posters). Throughout the year, students will also engage in the primary literature, culminating with the eventual submission of a primary journal article that includes appropriate introduction, method, results, and discussion.
