Methods | Special Issue: Acculturation Reimagined

Comparing bivariate and multivariate approaches to testing individual-level interaction effects in meta-analyses: The case of the integration hypothesis

Dinh-Hung Vu

https://doi.org/10.56296/aip00038
Published: July 7, 2025
Copyright: The authors (CC BY 4.0)

Vu, D. & Bierwiaczonek, K. (2025). Comparing bivariate and multivariate approaches to testing individual-level interaction effects in meta-analyses: The case of the integration hypothesis. advances.in/psychology, 2, e919144. https://doi.org/10.56296/aip00038

Vu, Dinh-Hung, and Kinga Bierwiaczonek. "Comparing bivariate and multivariate approaches to testing individual-level interaction effects in meta-analyses: The case of the integration hypothesis." advances.in/psychology, vol. 2, no. 1, 2025, e919144. https://doi.org/10.56296/aip00038.

Vu, Dinh-Hung, and Kinga Bierwiaczonek. 2025. "Comparing bivariate and multivariate approaches to testing individual-level interaction effects in meta-analyses: The case of the integration hypothesis." advances.in/psychology 2 (1): e919144. https://doi.org/10.56296/aip00038.

Vu D, Bierwiaczonek K. Comparing bivariate and multivariate approaches to testing individual-level interaction effects in meta-analyses: The case of the integration hypothesis. advances.in/psychology. 2025;2(1):e919144. doi:10.56296/aip00038.

Vu, D. and Bierwiaczonek, K. (2025) 'Comparing bivariate and multivariate approaches to testing individual-level interaction effects in meta-analyses: The case of the integration hypothesis', advances.in/psychology, 2(1), e919144. Available at: https://doi.org/10.56296/aip00038.

Download .RIS Download .bib

Methods for testing individual-level interactions in meta-analyses have not existed until recently, and past meta-analyses often attempted to approximate interaction tests using bivariate approaches that yield results of unknown accuracy. Focusing on one of the most prominent interaction-based hypotheses in psychology, the integration hypotheses, we test to what extent results from meta-analyses using four such bivariate approximations (the summative approach, the multiplicative approach, Euclidean distance, the midpoint split approach) diverge from a state-of-the art meta-analytical test of individual-level interaction (multivariate approach). A re-analysis of two datasets previously used in meta-analyses (total k = 57, total N = 7,512) revealed that variance explained by interaction proxies from bivariate approaches oscillates around 2%, while variance explained by a correct test of interaction tends toward zero, with f2 < .009 (average effect size for interaction in psychology). Thus, results from bivariate approximations of an interaction test, employed in past meta-analyses of the integration hypothesis, are largely inflated.

Christine Emmer, Frauke Mann, Ksenija Edelinski, Jutta Mata (2026). Well-being and cultural identity in migrant acculturative stress: a daily diary study. Psychology and Health. https://doi.org/10.1080/08870446.2026.2677233

Download PDF Back to article