Who Makes Whom Happy in a Romantic Relationship? Contributions to the Predictive Validity of Personality Questionnaires in Mating Contexts

Abstract

Whose romantic relationships last happily? The present work aimed to contribute to the predictive validity of personality questionnaires in mating contexts. Thereby, it refined the understanding of how personality and romantic relationships - particularly their quality - interplay. Five studies that utilized questionnaires for self-assessing both relationship-related and general personality traits were conducted.

Study I and IV focused on predicting relationship quality with the help of methods from machine learning: in both, linear additive models were developed and cross-validated based on different sets of personality variables as predictors. They reproducibly predicted relationship quality to an unprecedented extent. In the four-year longitudinal Study I, variables of relationship-related personality better than those of general personality predicted future relationship quality, while actor effects were better predictors than partner, similarity, and other interaction effects. In contrast to Study I’s findings, interaction effects (fittings, similarities, other moderators) outperformed actor and partner effects in the cross-sectional Study IV; this may be due to the usage of a novel matching test which was developed and validated successfully in Study III, thus paving the way to be applied in Study IV and V. The novel economic questionnaire measures one’s own and preferred partner characteristics in a parallel design to enable real-ideal fittings with a current (potential) partner. Just as similarity scores, fitting scores showed high predictive validity in Study III, but were not able to incrementally contribute to prediction of relationship quality in the cross-sectional Study IV.

Requesting transferability to singles in mating contexts, Study II and V contributed to assess the cross-contextual applicableness of prediction models, which are typically developed on couple’s datasets in anonymous settings. During the four-year-longitudinal Study II, relevant changes of relationship-related personality facets over time and different relationship statuses were tested and described. Unsurprisingly, but still pioneering, many facet changes were relevantly affected by current partnerships and their outcome, while others were not. The findings partly limit the applicableness of models from Study I, which are based on the same personality test. In Study V, answer distortions between an anonymous and a fake online dating setting were tested and described. Patterns of self-exaggeration, lower retest-correlations for preferred than for own characteristics, and setting-related differences in predictive validity regarding relationship quality were found. These outcomes partly limit the applicableness of models from Study IV, which are based on the same personality test.