Contributed paper list

   (Tuesday 13th, 08:20-09:20)

Statistics education in the social and behavioural sciences: from dichotomous thinking to estimation thinking and meta-analytic thinking


Geoff Cumming (Australia)


Null hypothesis significant testing (NHST) dominates in the social and behavioural sciences, despite strong evidence of its disadvantages. Worst may be its reinforcement of dichotomous thinking (DT), which focuses on impoverished reject or don’t-reject decisions. By contrast, estimation thinking (ET) and meta-analytic thinking (MAT) focus on sizes of effects, and cumulation of evidence to increase precision. A shift from DT to ET and MAT is highly desirable. Statistics education for ET and MAT will emphasise effect sizes, confidence intervals and metaanalysis, starting with the introductory course. Interpretation of confidence intervals should emphasise estimation, not NHST. Students and researchers, when specifying research goals, or discussing and interpreting findings, should use language that reflects ET and MAT. The outcome should be more quantitative theories, more sophisticated disciplines and better research progress.