Source
Sheeran, Listrom & Gollwitzer (2024), LIT-1488, read in full in NOTE-1131. What the meta-analysis measured is THEORY-918, which this entry extends.
The claim
The review pools 642 experimental tests from 294 reports to September 2019. After a robust Bayesian correction for publication bias it still finds extreme evidence of an effect, and concludes: “Thus, implementation intentions appear to have a robust impact on outcomes though the effect size is small according to conventional guidelines” (p. 170). It reports that planning helps across four kinds of outcome, behaviour (d = .27), cognitive performance (.47), affect regulation (.66) and social judgement (.52), and presents this scope as a finding of the review.
What THEORY-918 holds is narrower on one point. The correction was applied to the overall estimate only; the four outcome estimates are uncorrected, and within behaviour vaccination and voting did not differ reliably from zero. So “reliably”, for each domain, is the authors’ conclusion carried past what their corrected analysis tested.
What would settle it
The promote_when, which QUESTION-310 also asks: an updated or preregistered set of tests estimated with correction for publication bias and still reliably above zero, of the same small order. A corrected estimate per outcome domain would settle the four-domain clause. A larger uncorrected pooled d cannot answer it.
Note of 2026-10-30: what holds it back
Its title is the wide form THEORY-918 was narrowed from; narrowed to the meta-analysis’s own finding it would be THEORY-918, already Active. What it adds, that the effect is reliable in each of the four outcome domains, rests on per-domain estimates the review did not correct for publication bias, and within behaviour vaccination and voting did not differ from zero. It waits on a bias-corrected estimate per domain, or the updated or registered set its promote_when names (QUESTION-310).