A Meta‐Analysis of Research on the Read Aloud Accommodation

Abstract

Read aloud is a testing accommodation that has been studied by many researchers, and its use on K‐12 assessments continues to be debated because of its potential to change the measured construct or unfairly increase test scores. This study is a summary of quantitative research on the read aloud accommodation. Previous studies contributed information to compute average effect sizes for students with disabilities, students without disabilities, and the difference between groups for reading and mathematics using a random effects meta‐analytic approach. Results suggest that (1) effect sizes are larger for reading than for math for both student groups, (2) the read aloud accommodation increases reading test scores for both groups, but more so for students with disabilities, and (3) mathematics scores gains due to the read aloud accommodation are small for both students with and without disabilities, on average. There was some evidence to suggest larger effects in elementary school relative to middle and high school and possible mode effects, but more studies are needed within levels of the moderator variables to conduct statistical tests.

Comment on Buzick and Stone: Two Tales of One City

“…In my meta‐analysis, whether or not I controlled for other predictors (e.g., subject area, delivery method, grade level), the effect size difference between students with disabilities and students without disabilities was .13 ( p < .05). In Buzick and Stone (), the effect size difference was reported as .35 for reading tests ( p < .05) and .03 ( p > .05) for math tests. Therefore, in regard to math tests, Buzick and Stone found no significant difference in the score gains between students with disabilities and students without disabilities.…”

“…Buzick and Stone () and I (Li, ) each independently conducted a meta‐analysis of read aloud accommodations. Although the two meta‐analyses adopted different methods, they revealed some similar patterns.…”

“…First, the two meta‐analyses adopted slightly different inclusion criteria. As shown in Table , my meta‐analysis included seven articles not included in Buzick and Stone (), whereas Buzick and Stone included three articles that I did not include. For instance, Buzick and Stone excluded four of the articles that I included because the standard and accommodated conditions were not counterbalanced.…”

“…*Indicates articles included in Li () but not in Buzick and Stone (); + indicates articles included in Buzick and Stone () but not in Li ().…”

A Comment on Li: The Same Side of a Different Coin

“…The general conclusions about the efficacy of read aloud accommodations derived from Li's () meta‐analysis were consistent with ours (Buzick & Stone, )—that (a) effect sizes are higher for reading assessments than for math assessments for both students with and without disabilities, (b) students with disabilities do not receive a differentially higher benefit from read aloud accommodations on math assessments relative to students without disabilities, (c) read aloud mode can explain some of the differences in effect sizes, (d) there is significant variability in effect sizes that we are not able to explain given information in available studies, and (e) there is an interaction among moderator variables that we cannot quantify without additional effect sizes from new studies. We both concluded that because students without disabilities benefit from read aloud accommodations on reading assessments, as well as students with disabilities, additional validity evidence would be needed to address fairness in relation to specific uses of reading test scores (e.g., as a graduation requirement).…”