Modeling (In)Congruence in Groups: A How-To Guide for Applying Polynomial Regression and Response Surface Method to Multilevel Data

• Published in: Group dynamics 2023-09, Vol.27 (3), p.154-170
• Main Authors: Berg, Ann-Kathleen, Bakaç, Cafer, Kauffeld, Simone
• Format: Article
• Language: English
• Subjects: Female; Group Dynamics; Group research; Human; Hypotheses; Hypothesis Testing; Male; Responses; Statistical Regression; Teams
• ISSN: 1089-2699; 1930-7802
• DOI: 10.1037/gdn0000200

Objective: Group scholars often question whether the fit between group members and their respective teams or leaders is associated with an outcome variable. For example, do team members perceive the same team climate as their peers, and how does this affect subsequent outcomes? Such (in)congruence hypothesis-based research questions have been commonly investigated through the application of difference scores. However, ignoring the multidimensional relationship between two predictors and an outcome leads to severe misinterpretations. Therefore, our objective is to outline a relatively new method to test (in)congruence in groups. Method: Polynomial regression (PR) and response surface methodology (RSM) in combination can be used to test the (in)congruence hypothesis in psychological phenomena and provide explanatory power. This article outlines an introduction to the method and offers a how-to guide with an illustrated example of innovation teams. Results: The illustrative example showcases how multilevel PR and RSM can be used to analyze, report, and interpret data of (in)congruence hypothesis in group research. Conclusions: Two interpretation approaches are discussed, and common problems are considered when interpreting results. Some future directions for testing a(n) (in)congruence hypothesis with nested data are highlighted. R syntax is provided in the Supplemental Materials. Highlights and Implications * The commonly used statistical approaches fail to capture congruence or incongruence patterns between group members. * This article introduces the application of multilevel polynomial regression and response surface methodology as a sound method for testing (in)congruence hypotheses. * With an illustrative example, we provide an accessible how-to guide for investigating (in)congruence hypotheses in group-based data structures. * By providing two approaches for interpreting the results, we enable researchers to make informed decisions upon their analysis.