# Statistical Interaction: More than the Sum of its Parts

Posted July 6, 2017

When you think of an interaction, a social interaction or a drug interaction might come to mind. But did you know that there might be an interaction amongst the variables in your research? In statistics, an interaction is a special property of three or more variables, where two or more variables interact to affect a third variable in a non-additive manner. In other words, the two variables interact to have an effect that is more than the sum of their parts.

To illustrate, think of a drug interaction: barbiturates alone have a depressive effect on your central nervous system. Alcohol alone has a depressive effect on your central nervous system. Taken together, these two drugs strengthen the effect of the other drug, beyond the effect that that drug would normally produce, so much so that you could fall into a coma!

Fortunately, “mixing” two variables in your dataset will not make you lose consciousness. An interaction term is a variable that represents an interaction between two variables. In some analyses, SPSS will create the interaction term for you, such as in a mixed-models ANOVA. In others, such as a moderation analysis, you may have to create an interaction term yourself. To create an interaction term, simply multiply those two variables using the syntax editor or the compute variable window. The naming convention for this interaction term is: Variable 1 x Variable 2. Keep in mind that there are a few more preliminary steps to take if you are creating an interaction term for a moderation, depending on your study and the variable types.

The most common interaction you are likely to see is as part of a mixed-models ANOVA. In this technique, you have two groups (typically a treatment and control group—but it could be gender, junior/senior, low exercise/high exercise) who are both measured on a single continuous dependent variable two or more times (also known as repeated measures). This design is commonly used by researchers seeking to test an intervention. In this design, you have a Group x Time interaction (with time being your repeated measures variable). Your ANOVA output will give you a main effect of group, a main effect of time, and an interaction effect between group and time. A significant main effect of group means that there are significant differences between your groups. You then interpret the means of each group. If your group has more than two levels, you do post hoc testing. A significant main effect of time means that there are significant differences between your repeated measures. You then either interpret means or do post hoc testing. A significant interaction effect means that there are significant differences between your groups and over time. In other words, the change in scores over time is different depending on group membership. If you have a significant interaction, you will want to examine means, a line-plot, or use post hoc testing to determine the exact nature of the interaction.

If you need help with moderation analysis or interpreting interactions, keep an eye out for future blogposts, or contact us to schedule a consultation!

Shares