Strategies for Teaching Students to Think Critically

867 studies screened from 1930-2009; 341 effect sizes retained after methodological filtering

Elementary school through adult; a majority of the primary literature on teaching critical thinking is post-secondary. Within school ages the review distinguishes only 11-15 and 16-18.

The boundary case for this topic — not science-specific, but it is the largest synthesis of whether "teaching thinking" produces measurable thinking. Two-stage design: a methodological screen that removed weaker designs and instruments (the authors report that study design and instrument type moderated results, i.e. bad designs gave bigger numbers), then a content analysis on 341 higher-quality effects. Outcomes are standardised critical-thinking instruments plus content-specific CT and CT disposition measures. Crucially for this archive: measuring critical thinking after teaching critical thinking is a gain on the trained construct, not far transfer. The review reports no attainment outcome in an untaught school subject, so it cannot speak to whether CT instruction moves maths or English. STEM vs non-STEM subject moderation was tested and found no significant difference — teaching thinking through science is not special.

Critical thinking is teachable, at g = 0.30 overall (341 effects), and the effect is stable across educational level, subject and intervention duration. Content-specific CT g = 0.57 (97 effects), CT disposition g = 0.23 (25 effects), and subject-content learning g = 0.33 (140 effects). Combining application-oriented instruction, dialogue and mentoring gave the largest effect, g = 0.57 (19 effects). Middle school g = 0.37 vs high school g = 0.25, not significantly different. Two things this does NOT establish, and both matter here: it does not show transfer to attainment in an untaught subject, and it contains no counterweight to the observation that d ≈ 0.30 on a construct-aligned measure is roughly what a year of ordinary schooling produces on a standardised one.

Medium: pools quasi-experimental with true experimental designs; the authors' own moderator analysis shows design quality changes the answer.

The trainability of critical thinking is consistently reproduced. What has not been shown anywhere in this literature is that CT instruction moves an independent standardised measure of achievement in a subject that was not taught.