UNDERSTANDING DATA WHEN INTERACTIONS ARE PRESENT OR HYPOTHESIZED

The traditional approach to interpreting data when an interaction is p resent is to interpret cell means as reflecting a difference between d ifferences. An alternative is to interpret any main effects and separa tely interpret interaction residuals. Interpretation of interaction re siduals can sometimes lead to nonsensical conclusions. Thus, in decidi ng between the approaches, researchers should consider (a) the concept ual nature of the variables involved, (b) relevant theories, and (c) t he additional data that might be necessary to distinguish among compet ing plausible representations. Issues have also arisen regarding how t o best test a hypothesis that involves interaction variance. Some rese archers have argued that the use of a focused contrast designed to tes t a specific ordering of cell means (e.g., a lineal contrast) is often the best strategy. We note potential problems with the use of such co ntrasts and discourage researchers from blanket use of contrasts that combine main effect and interaction variance.