A meta-analysis of writing in social studies, science and maths

Is writing about classroom content an effective way to learn? Arizona State University’s Steven Graham and colleagues at the University of Utah recently performed a meta-analysis on the effects of writing about classroom content in social studies, science and maths. Specifically, they examined if writing increased pupil achievement, if the results differed among subjects, and if any relationships existed by year level, activity type, or any other factors.

To be included, studies had to meet quality criteria including true or quasi-experimental research design, reliability of measures, controlling for teacher effects, multiple classes in the experimental and control conditions, experimental and control group pre-test equivalence, and both groups experiencing equal amounts of time learning the same topics.

This search yielded 56 studies in 53 documents meeting criteria for inclusion, involving 6,235 pupils in grades 1-11 (Years 2-12). Pupils in experimental groups wrote about classroom content, while most controls did not write at all. Forty-six percent of the studies assessed the impact of writing on science, 38% on maths and 14% on social studies. Thirty-four percent examined elementary (primary) pupils, and 32% each examined middle and high school (secondary) pupils. The types of writing activities for the experimental groups included writing informational text, such as summarising information or writing a report (34%); journal writing (32%); argumentative writing (13%); and narrative writing, such as creating a word problem in maths lessons (5%). These were coded to determine which, if any, were more effective than others.

Results showed that writing about content increased pupil achievement when compared to equivalent peers in non-writing control groups. Average weighted effect sizes were statistically significant in science (+0.31), social studies (+0.31) and maths (+0.32), as they were when broken down by elementary (+0.29), middle (+0.30) and high school (+0.30) levels. No correlation was found with number of treatment days, type of writing task, or type of assessment.

Source: The effects of writing on learning in science, social studies, and mathematics: A meta-analysis (March 2020), Review of Educational Research