Counter-Intuitive Findings on Affect and Ideology Likely Reflect Collider Bias: Commentary on Turner-Zwinkels et al., 2025

Abstract

A paper recently published by Turner-Zwinkels et al. (2025) contains a peculiar set of findings: using cross-sectional analyses of international survey data, one study found that the more partisans agree with their co-partisans on matters of policy and ideology, the less they like the party they all support, whereas in another study these variables were uncorrelated. These negative and null relationships are counter-intuitive and conflict with prior findings showing that greater perceived ideological similarity tends to increase liking. Turner-Zwinkels et al. suggest these results could be a product of optimal distinctiveness theory. However, I suggest that this result is a statistical artefact caused by collider bias. I explain how collider bias could create this result even if the true relationship between liking and ideological similarity is positive. I demonstrate the plausibility of this explanation using simulations and a re-analysis of Turner-Zwinkels et al.’s data.