Learning Styles: Concepts and Evidence

Commissioned review of the learning-styles literature; a handful of studies were found to use the required crossover design

Learners from kindergarten to graduate and adult education, across the whole learning-styles research base

Psychological Science in the Public Interest review. Its durable contribution is a design criterion rather than a result: to validate matching, learners must be classified by style, then randomly assigned across instructional methods, then given an identical test, and the data must show a CROSSOVER interaction (the method that is best for style A is not the method that is best for style B). Anything short of that design cannot bear on the meshing hypothesis no matter what it reports.

Ample evidence that people will express presentation preferences if asked, and ample evidence that people differ in aptitude; essentially no evidence of the crossover interaction the meshing hypothesis requires. Of the very few studies using an adequate design, the results contradicted meshing. The authors conclude there is no adequate evidence base for adopting learning-styles-based instruction, and note the opportunity cost of the commercial industry built on it.

Not applicable in the usual direction, and this is why the topic is unusually clean. The meshing hypothesis predicts an aptitude-by-treatment INTERACTION, not a main effect, so heritable between-person differences cannot manufacture it and selection into treatment cannot fake it. The absence of the crossover is therefore not explicable by confounding.