Using linear mixed models to analyze learning processes within sessions improves detection of treatment effects: An exemplary study of chronometric mental rotation

Abstract

Practice effects occur for many cognitive tasks. They are observed not only between repeated tests, but also within sessions. They can confound the detection of treatment effects, even when compared with control groups. We present an approach to reduce the impact of within-session practice effects through inclusion and manipulation of time in linear mixed models. With an exemplary mental rotation task, we demonstrate the possibility to investigate performance within sessions. We show how controlling for practice effects can improve comparisons between treatments. Through simulations, we demonstrate the impact of within-session practice effects and the usefulness of the presented approach. • Practice effects are prevalent in cognitive tests. • Concern: impact detection of treatment effects in repeated tests. • Inclusion in analysis can separate practice and treatment effects.