LAUSR

Statistical interactions between two continuous variables in linear regression are common in psychological science. As a follow-up analysis of how the moderator impacts the predictor-outcome relationship, researchers often use the pick-a-point simple slopes method. The simple slopes method requires researchers to make two decisions: (a) which moderator values should be used for plotting and testing simple slopes, and (b) which predictor should be considered the moderator. These decisions are meant to be driven by theory, but in practice researchers may use arbitrary conventions or theoretical reasons may not exist. Even when done thoughtfully, simple slopes analysis omits important information about the interaction. Consequently, it is problematic that the simple slopes approach is the primary basis for interpreting interactions. A more nuanced alternative is to utilize the Johnson-Neyman technique in conjunction with a regression plane depicting the interaction effect in three-dimensional space. This approach does not involve picking points but rather shows the slopes at all possible values of the predictor variables and gives both predictors equal weight instead of selecting a de facto moderator. Because this approach is complex and user-friendly implementation tools are lacking, we present a tutorial explaining the Johnson-Neyman technique and how to visualize interactions in 3-D space along with a new open-source tool that completes these procedures. We discuss how this approach facilitates interpretation and communication as well as its implications for replication efforts, transparency, and clinical applications. (PsycInfo Database Record (c) 2020 APA, all rights reserved).