The relation between retell and reading comprehension: a meta-analysis.

The relation between retell and reading comprehension: a meta-analysis.

First Author: Yucheng Cao -- University of California, Irvine
Additional authors/chairs: 
Young-Suk Grace Kim
Keywords: Reading comprehension, Meta-analysis, Grades K-12
Abstract / Summary: 

Abstract
Purpose
This study examined the relation between retell and reading comprehension using a meta-analysis. Specifically we examined 1) the overall correlation between retell and comprehension; 2) the potential factors that may moderate the relation, including reading comprehension assessment methods (e.g., multiple-choice, short open-ended), reading modes prior to retell (i.e., oral or silent reading), number of prompts provided during retell tests, retell evaluation methods (e.g., word counts, idea units), text genres, and grade level.
Method
Literature was searched through five databases yielding a total of 1699 articles of which 23 studies (N = 4705) met the inclusion criteria. Bivariate correlation between retell and reading comprehension and other information were coded. Data were analyzed in R (metafor) using random-effects models with maximum likelihood estimation. Sensitivity analyses were also conducted in terms of influential outliers, publication bias, study quality, and robust variance estimation.
Results
A moderate correlation was found between retell and reading comprehension: r = .46. Moderation analyses revealed that the strength of relation varied as a function of students’ grade level (b = -0.03) and number of prompts (b = 0.12) such that the magnitude of the relation was slightly weaker in higher grades and the relation was stronger with greater number of prompts.
Conclusions
These results suggest that retell is moderately related to reading comprehension, but the relation varies as a function of grade level and number of prompts. Implications for the use of retell as a measure of reading comprehension will be discussed.