The Role of Knowledge Retrieval in Inference-making among Rural, Middle Grade Struggling Readers

点击翻译按钮获取中文摘要

Project Summary/Abstract Recent research shows that inferencing improves across grades 6-12, uniquely accounts for variance in sentence- and passage-level comprehension, and individual differences in inferencing relate in a principled way to variations in reading comprehension for readers of all abilities (Barth et al., 2015; Barnes et al., 2015). These findings suggest that comprehension requires inferencing and that comprehension fails when readers do not possess relevant knowledge or can only slowly retrieve and integrate knowledge from text or semantic memory during reading (Kendeou, 2015). Yet, very little research has examined how inferencing changes as a function of changes in knowledge. To extend this limited body of research, this application requests three years of funding to conduct two experimental studies designed to (a) examine the relationship between knowledge retrieval and inferencing (Aim 1) and (b) determine the effectiveness of an intervention that improves knowledge retrieval and inferencing (Aim 2) among rural, middle grade struggling readers. This proposal also seeks to expand research opportunities for undergraduates at Buena Vista University (Aim 3). The research design uses 316 struggling readers in grades 5-8 who attend a rural school. To test Aim 1, the effects of knowledge retrieval (accuracy and speed) on inferencing will be modeled without dichotomizing the distribution. Linear mixed effect models will be fit to determine whether reader characteristics (i.e., knowledge retrieval, metacognition, reading comprehension, word reading efficiency, background knowledge, working memory, and grade) make unique contributions to inferencing across the posttest and follow-up data collection time points. To test Aim 2, we will first consider several structural models as students may be nested in teachers, schools, and tutors. We will run unconditional models to estimate the intraclass correlation for each level of the study design. If significant interclass correlations emerge, we will fit multilevel models to evaluate the effectiveness of the intervention while controlling for covariates such as pre-test performance on inference-related measures and child-attributes such as English learner status. Our primary analysis plan assumes an intent-to-treat model in which we test the efficacy of two intact conditions. We will also estimate effect sizes to report the magnitude of difference between the two conditions. Expected outcomes include (a) the identification of a method that effectively facilitates knowledge retrieval and the application of relevant knowledge to form inferences among rural, middle grade, struggling readers; (b) the validation of an intervention that teaches rural, struggling middle grade readers how to activate, retrieve, and interweave relevant knowledge with information in the text and accurately form inferences while reading that can be broadly implemented in middle grade classrooms; and (c) expansion of undergraduate research opportunities at Buena Vista University by integrating students on the research team with the knowledge and skills to successfully implement these randomized controlled trials.