A study explores predictors of the quality of romantic relationships. Research on the predictors of the quality of romantic relationships is often limited in scope and scale. Samantha Joel and colleagues used a machine-learning approach to analyze 43 datasets that included 2,413 mostly self-reported measures collected from 11,196 couples recruited by 29 laboratories. The authors found that relationship-specific variables, such as perceived-partner commitment, sexual satisfaction, perceived-partner satisfaction, and lack of conflict, accounted for approximately 45% of variance in relationship quality. The top individual characteristics with the strongest predictive power for relationship quality were life satisfaction, negative affect, depression, attachment avoidance, and attachment anxiety. However, individual characteristics explained only 21% of variance in relationship quality. Similarly, variables capturing the partners' views about the relationship explained only 15% of variance in relationship satisfaction. Individual characteristics and relationship-related judgments reported by partners did not provide additional value over one's own relationship-specific opinions in predicting relationship quality. According to the authors, the machine-learning approach could provide a model for future research in the field of relationship science.
Article #19-17036: "Machine learning uncovers the most robust self-report predictors of relationship quality across 43 longitudinal couples studies," by Samantha Joel et al.
MEDIA CONTACT: Samantha Joel, Western University, London, CANADA; e-mail: samantha.joel@uwo.ca
###
Journal
Proceedings of the National Academy of Sciences