Teaching children to write: A meta-analysis of writing intervention research

Writing intervention studies in regular school settings, grades 4-6; effect sizes pooled into 10 intervention categories

Grade 4-6 pupils (roughly ages 9-12) in regular school settings

Meta-analysis (Journal of Writing Research 7(2), 249-274) of writing interventions aimed at grades 4-6, with pairwise comparison between the categories that reached significance. Outcomes are researcher-scored writing performance measures throughout. The effect sizes here should be read as a caution rather than a menu: an ES of 2.03 for goal setting is not a credible causal parameter for a cheap classroom manipulation, and its presence at the top of the table is a good illustration of METHODOLOGY's rule that d > 0.60 from small studies with treatment-aligned measures is a red flag, not a triumph.

Ten intervention categories were tested - strategy instruction, text structure instruction, pre-writing activities, peer assistance, GRAMMAR INSTRUCTION, feedback, evaluation, process approach, goal setting, and revision. Only five reached statistical significance, and grammar instruction was not among them. Of those that did: goal setting ES = 2.03, strategy instruction ES = 0.96, feedback ES = 0.88, text structure instruction ES = 0.76, peer assistance ES = 0.59. This is the third independent meta-analysis (after Hillocks 1986 and Graham & Perin 2007, and alongside Graham et al. 2012) to test grammar instruction against alternative writing instruction in a general-population sample and find nothing.

Low for direction; the corpus is experimental and quasi-experimental. Magnitudes are not trustworthy - the 2.03 for goal setting implies a manipulation moving pupils two standard deviations, which no independently measured writing intervention has ever produced.