MerQuR+Statistics Glossary

Statistics Glossary → Simple effects

Simple effectsconcept

Testing the effect of one factor separately at each level of the other factor; the basic way to unpack an interaction.

In a two-factor design with a significant interaction, the main effect of a factor can mislead on its own, because the effect changes with the level of the other factor. A simple effects analysis 'slices' the data by the levels of the other factor and tests the factor of interest within each slice with an F test. The error mean square of the full model is usually used; in significant slices, pairwise comparisons follow, adjusted for multiple testing.

Simple effects do not test the interaction itself: an effect that is significant in one slice and not in another does not prove that the two effects differ; the interaction term tests that. Simple effects are therefore usually interpreted after a significant interaction, to describe its pattern.

An everyday example

A school compares two teaching methods (traditional and interactive) among girls and boys, and the method × sex interaction is significant. Simple effects show that the interactive method raises the mean by 12 points for girls but only 2 points for boys, a significant difference for girls and a non-significant one for boys.

In MerQur

The 'Simple effects' option in Two-Way ANOVA (factor 1 at each level of factor 2, the reverse, or both); 'Simple effects and pairwise comparisons' on the multi-factor path of Repeated Measures ANOVA; the 'Simple effects and pairwise comparison' checkbox in Mixed-Design ANOVA; and in ANCOVA, through the adjusted means, when a '2nd factor (optional)' is added.

Analyses where this concept appears

Two-Way ANOVARepeated Measures ANOVAMixed-Design ANOVA, Split-PlotANCOVA

Related concepts

Interaction effectEstimated marginal means (EMM)