The empathy perception paradox: how women’s generous partner perceptions create apparent gender gaps in romantic relationships

Hyde’s [1] gender similarities hypothesis suggests that gender differences in relationship perceptions are often exaggerated, yet prior work has rarely modelled dyadic non-independence or formally decomposed self–partner discrepancies into their sensitivity and bias components. Data were collected from 545 Slovak couples ( N = 1,090); the dyadic analyses presented here focus on 528 heterosexual dyads ( N = 1,056). Measures of satisfaction, attraction, attachment behaviours, and person-centred relational qualities were administered using both self- and partner-report forms. Gender comparisons were estimated with generalised estimating equations (GEE) with an exchangeable working correlation and robust (sandwich) standard errors, clustered by couple. The focal empathy variable was further decomposed using the Truth and Bias model [2], which separates the components labelled tracking accuracy, directional bias, and assumed similarity in the original model and referred to here, for the reasons given in the Introduction, as perceptual sensitivity, directional bias, and perceptual projection. Within-couple correlations were substantial across all relationship variables ( r = .55–0.71; ICC = 0.53–0.70), reflecting the nested design in which partners were sampled together and cannot be treated as independent observations. Consistent with the gender similarities hypothesis, GEE estimates of gender differences were predominantly small (d = 0.05–0.26). For empathy, the decomposition showed equal perceptual sensitivity for men and women (β = 0.43, p < .001, for both) and greater perceptual projection among men (β = 0.77 vs. 0.60, p = .030). Directional bias was small and asymmetric: men underestimated women’s empathy by 1.38 points, or 1.9% of the scale range, a bias that differed from zero ( p = .008), whereas women overestimated men’s empathy by only 0.23 points, or 0.3% of the range, a bias indistinguishable from zero ( p = .592). The gender × satisfaction interaction was specific to empathy: in the lowest relationship-satisfaction tertile men’s calculated self–partner discrepancy scores were large and positive (M = + 8.33) compared with women (M = + 1.12), whereas in satisfied couples the gap closed (Ms = − 1.96 vs. −4.15). Once dyadic dependence is properly modelled and judgments are decomposed via the Truth and Bias framework, the apparent “empathy perception paradox” is more nuanced than individual-level discrepancy scores suggest. Both partners’ judgments are equally sensitive to their partner’s self-reported empathy; the asymmetry is concentrated in directional bias (men underestimate women) and perceptual projection (men more than women). Clinical work with couples should attend to these gender-patterned perceptual processes rather than assume self-perception bias.