Computational procedures for probing interactions in OLS and logistic regression: SPSS and SAS implementations

Area of Science:

• Social Sciences
• Psychology
• Biostatistics

Background:

• Hypothesizing moderated effects, where an independent variable's impact varies with a moderator, is common in research.
• Statistical interactions between independent and moderator variables represent these moderated effects in outcome models.

Purpose of the Study:

• To describe methods for probing significant interaction effects in statistical models.
• To introduce the Johnson-Neyman technique as an alternative to the pick-a-point approach for interaction probing.
• To provide computational tools (SPSS and SAS macros) for simplifying interaction analysis in regression.

Main Methods:

• Description of the pick-a-point method for interaction probing.
• Detailed explanation of the Johnson-Neyman technique for analyzing conditional effects.
• Development and presentation of SPSS and SAS macros for implementing these techniques in ordinary least squares and logistic regression.

Main Results:

• The study facilitates the probing of interactions by offering accessible computational tools.
• The Johnson-Neyman technique provides a more comprehensive understanding of conditional effects compared to pick-a-point.
• Macros simplify the application of these advanced statistical methods for researchers.

Conclusions:

• Accurate probing of statistical interactions is essential for testing specific theoretical predictions.
• The Johnson-Neyman technique and provided macros enhance the ability of researchers to analyze moderated effects.
• These tools support more nuanced interpretations of independent and moderator variable relationships in regression analyses.