ABSTRACT
After conducting an ANOVA, researchers often wish to explore the data further. Contrasts allow means to be compared to investigate more specific hypotheses than are tested by an ANOVA. Contrasts can be planned before the data have been examined, or be unplanned and conducted once the means have been calculated. A variant on contrasts – trend analysis – can be applied when levels of an IV are quantitative rather than qualitative. Trend analysis allows patterns across means to be explored to see whether there is a trend across the levels of the IV.
Simple effects analysis allows the nature of an interaction between two IVs to be explored further. It isolates one level of one IV at a time to see how the levels of the other IV vary. Simple effects can show differences in patterns even when the original interaction F-ratio is not significant. Interaction contrasts allow analysis of more complex contrasts which involve both variables in the interaction.
If there is a significant main effect and no interaction, then, if the main effect has only two levels, the direction in which the result went can be found by inspecting the marginal means for that IV. However, if the main effect has more than two levels, then contrasts need to be conducted to explore the source of the significant result further.
