Social Cognition and Decision Sciences

Predicted values with interactions in regression analysis

 

Prof. Dr. René Ziegler

(new version, July 2015)

It has become very common in psychological research to run moderated regression analyses to test for interactive effects of

(a) two or more continuous variables or

(b) one or more continuous variables with one or more categorical variables

on some dependent measure.


Often, when an interaction is found significant with such an analysis, one would like to know what this interaction "looks like". In the case of analysis of variance (ANOVA), output from software programs delivers pertaining means. However, such output is not readily available in the case of interactions involving one or more continuous independent (or predictor) variables. Indeed, what is of interest in such cases is the predicted values.

For this reason, I have programmed a tool that calculates predicted values based on the results of a moderated regression analysis. Click on the LINK, save the zip-file on your computer, double-click it, and install it by running setup. The readme-file shows in detail how the program can be used.

Features of the tool:

Given this last feature, the program may be helpful, for instance, in those cases in which the same regression equation (i.e., containing the same predictors) is tested repeatedly (e.g. with different dependent variables).

For two rather simple cases, you may also download one of two Excel files in which you can enter the necessary information and directly use the 'Read data from file' Mode of the tool. To do so, first click on the Excel file and save it on your computer. Also install the tool on your computer. Then open the Excel file, enter the necessary information, and run the read-from-file mode of the tool.
One Excel file is prepared for the case of one continuous independent variable and one categorical independent variable with two levels (2-way).
The other Excel file is prepared for the case of one continuous independent variable and two categorical independent variables each of which having two levels (3-way).

If you like, send me an e-mail as to your thoughts about and experience with the tool. In order to use the tool, save the zip-file on your computer and install it by running the setup file.

You may also be interested in the pages by Jeremy Dawson, Andrew Hayes, Wilhelm Hofmann, David Kenny, and Kristopher Preacher.

DISCLAIMER: I have aimed to assure that the computations done by the tool are correct. Nonetheless, I cannot accept any responsibility for any negative consequences that might result from it's use. I offer the tool to you 'as is'.