Abstract
Post-conversation evaluations are partly shared between partners and partly specific to each individual. Using a mixed-effects actor-partner interdependence (APIM) framework, we analyzed 1,602 first-encounter conversations, modeling a composite of enjoyment, liking, and perceived conversational skill. Variance in post-conversation positivity was distributed across stable perceiver tendencies (17.4%), stable target tendencies (19.9%), a shared conversation-level component (18.5%), and residual variation (44.2%). Partners showed modest within-dyad similarity (label-invariant ICC = .201), while absolute within-dyad gaps averaged nearly one standard deviation (mean |D| = 0.93). Personality traits and demographics jointly accounted for 9.0% of variance in positivity and 5.3% of the directional asymmetry between partners’ ratings. Agreeableness, Openness, and age were associated primarily with more positive perceiver-side evaluations, whereas Extraversion, female sex, and higher educational attainment were associated more strongly with partner-rated positivity. At the dyad level, fixed effects accounted for 11.1% of between-dyad variation in dyadic mean positivity and reproduced 24.0% of the observed cross-partner covariance. A descriptive label-invariant covariance decomposition further suggested that the modest within-dyad covariance attributable to measured characteristics arose mainly from additive perceiver-target structure rather than assortative pairing. The same predictors, however, did not capture the magnitude of within-dyad disagreement: predicted gaps were uniformly small while observed gaps varied widely. Stable individual differences therefore mattered in two ways, both as a lens through which conversations were evaluated and as part of the impression left on partners, while much of dyadic gap size remained unexplained by the measured stable characteristics.
Funding Information
  • Fundação de Amparo à Pesquisa do Estado de São Paulo (#2023/11070-1, #2024/06797- 2, #22/05848-7)
  • Vetenskapsrådet (#2020-03812)
  • Norges Forskningsråd (#343370)