Distinguishing Types of Poor Comprehenders of Intermediate-Grade Readers

• In Event: 68.018 - Expanding Our Understanding About Comprehension: Theory, Assessment, and Intervention

Abstract

Previous research has established two types of struggling readers: those who struggle with lower-level reading skills and those who struggle with higher-level reading skills (Cain & Oakhill, 2006; Perfetti, 2007). The latter group is commonly termed poor comprehenders: readers exhibit poor comprehension compared to peers with similar word-reading skills and vocabulary (e.g., Authors, 2014; Cain & Oakhill, 2006; Rapp et al., 2007). Moreover, research has revealed that poor comprehenders exhibit specific difficulty with generating causally coherent inferences (e.g., McMaster et al., 2012; Trabasso & van den Broek, 1985).
Causally coherent inferences require synthesis of why an event occurs based on relevant goals and subgoals previously identified in the text and missing information from background knowledge consistent with this synthesis. Although poor comprehenders do make causally coherent inferences, they do not make them as consistently as good comprehenders.
Instead, when poor comprehenders do not make such inferences, they often use other types of comprehension processes that are strategic and useful, but that fail to fill the causal gap in the text. Specifically, poor comprehenders tend either to paraphrase (i.e., rephrase or repeat prior text information) or make lateral connections (i.e., elaborative inferences or personal associations, which use background knowledge but are not causally coherent with the text). These trends have been found repeatedly with intermediate grade readers (i.e., Grades 3-5; e.g., Authors, 2014; McMaster et al., 2012; Rapp et al., 2007).
Given the consistency of such findings, we developed the Multiple-choice Online Causal Comprehension Assessment (MOCCA) to distinguish which processes poor comprehenders rely on when they are unsuccessful at making causally coherent inferences. During the second year of our study, we purposefully recruited a sample of children who very nearly represent the racial and ethnic diversity of students nationwide with the idea that reading comprehension theories and assessments may recreate inequity if they are not based on more heterogeneous sampling methods.
Our assessment utilizes a maze-like approach, but instead of removing a single word, we remove an entire sentence necessary for a passage’s conclusion to be causally coherent. Students must choose among three sentences the one they believe best completes the story. Each sentence represents one of three comprehension processes: a causally coherent inference, a paraphrase, and a lateral connection. We asked about 4500 children—1500 per Grades 3-5—across the U.S. to take our new assessment.
Results demonstrate that our assessment successfully distinguishes among readers on two dimensions: comprehension skill (i.e., poor to good comprehension) and error propensity (i.e., predilection to paraphrase vs. make lateral connections). That is, we reliably identify readers who are good comprehenders and reliably classify readers who do not exhibit good comprehension into one of three groups: paraphrasers, lateral connectors, and indeterminate poor comprehenders who do not exhibit a strong preference for a specific comprehension process. Differential item and test functioning will also be examined. Results have implications not only for further development of our assessment, but also for the generalizability of commonly accepted theories of reading comprehension and poor reading comprehension.

Authors

• Gina Biancarosa, University of Oregon

• Ben Seipel, University of Wisconsin - River Falls

• Sarah Elizabeth Carlson, University of Oregon

• Mark L. Davison, University of Minnesota