What the study found
The study found that polyvariate regression, an analysis method that tests how a predictor relates separately to multiple outcomes, performed better than using difference scores to study affective polarization. The authors also conclude that difference score measures can obscure potentially interesting findings.
Why the authors say this matters
The authors argue that researchers studying affective polarization should avoid difference scores because of their conceptual, statistical, and theoretical problems. They suggest polyvariate regression offers a better way to study how predictors relate to affective polarization without relying on subtraction-based measures.
What the researchers tested
The paper reviews problems with difference scores in measurement research and provides a tutorial for polyvariate regression. It then applies both approaches to U.S. data by testing whether people’s stances on government defense spending are associated with feeling thermometer ratings of the Republican and Democratic parties.
What worked and what didn't
The comparison showed better results for polyvariate regression than for a difference score measure of affective polarization. The abstract says the same association was tested both ways and that the difference-score version may have hidden findings that the polyvariate approach detected.
What to keep in mind
The abstract says polyvariate regression is discussed for multiparty contexts and other conditions, and it notes limitations of the method. It also says polynomial regression is recommended when polarization is the predictor rather than the outcome, but the abstract does not give further detail on those limitations.
Key points
- The paper criticizes difference scores for measuring affective polarization.
- Polyvariate regression is presented as an alternative analytic approach.
- In a U.S. example, defense spending stance was linked to ratings of the Republican and Democratic parties.
- The comparison of methods favored polyvariate regression over a difference-score outcome.
- The abstract says difference scores may obscure interesting findings.
Disclosure
- Research title:
- Polyvariate regression outperformed difference scores in affective polarization analysis
- Authors:
- Lukas K. Sotola
- Publication date:
- 2026-02-23
- OpenAlex record:
- View
Get the weekly research newsletter
Stay current with peer-reviewed research without reading academic papers — one filtered digest, every Friday.