Relational diversity in social portfolios predicts well-being

Abstract

We document a link between the relational diversity of one's social portfolio-the richness and evenness of relationship types across one's social interactions-and well-being. Across four distinct samples, respondents from the United States who completed a preregistered survey (n = 578), respondents to the American Time Use Survey (n = 19,197), respondents to the World Health Organization's Study on Global Aging and Adult Health (n = 10,447), and users of a French mobile application (n = 21,644), specification curve analyses show that the positive relationship between social portfolio diversity and well-being is robust across different metrics of well-being, different categorizations of relationship types, and the inclusion of a wide range of covariates. Over and above people's total amount of social interaction and the diversity of activities they engage in, the relational diversity of their social portfolio is a unique predictor of well-being, both between individuals and within individuals over time.

Keywords: conversation; diversity; relationships; well-being.

Fig. 1.

Schematic representing prototypical respondents low and high in: level of social interaction (one SD below vs. above the sample mean: low = 36%, high = 79%) and level of social portfolio diversity (one SD below vs. above the sample mean: H_low = 0.78, H_high = 1.71) in the French experience-sampling dataset (Study 4).

Fig. 2.

Standardized relationship between social portfolio diversity and subjective well-being in all four datasets. This figure shows the relationship between portfolio diversity and global well-being in the preregistered dataset (β = 0.13, b = 0.54, 95% CI [0.15, 0.92], P = 0.007, n = 576; controlling for proportion of time spent socializing), quality of life in the ATUS dataset (β = 0.08, b = 0.34, 95% CI [0.27, 0.42], P < 0.001, n = 19,195; controlling for proportion of time socializing and activity diversity), life satisfaction in the SAGE dataset (β = 0.05, b = 0.08, 95% CI [0.04, 0.12], P < 0.001, n = 8,824; controlling for proportion of time socializing and activity diversity), and global happiness in the French experience-sampling dataset (β = 0.04, b = 1.33, 95% CI [0.85, 1.81], P < 0.001, n = 21,645, edf = 1.72; controlling for proportion of time socializing and activity diversity).

Fig. 3.

Results from SCA for each dataset, showing the standardized regression coefficient for each model specification (x axes show the model numbers, which are sorted from smallest to largest effect size). (A) SCA results on our preregistered dataset, with the raw association (no controls) between portfolio diversity and global well-being in red. (B) SCA results for the ATUS data, with the raw association (no controls) between portfolio and quality of life in red, and subjective ratings of health in orange. (C) SCA results the SAGE dataset, with the raw association (no controls) between portfolio diversity and life satisfaction in red, subjective health in orange, and number of overnight hospital visits in the past 12 mo in yellow. (D) SCA results for the French experience-sampling dataset, with the raw association (no controls) between portfolio diversity and global happiness in red, positive emotion in orange, and negative emotion (reverse coded) in yellow.