Correction for Collins et al., Relational diversity in social portfolios predicts well-being

Social Sciences Correction for “Relational diversity in social portfolios predicts well-being,” by Hanne K. Collins, Serena F. Hagerty, Jordi Quoidbach, Michael I. Norton, and Alison Wood Brooks, which published October 17, 2022; 10.1073/pnas.2120668119 (Proc. Natl. Acad. Sci. U.S.A. 119, e2120668119).

The authors note that on page 2, right column, second full paragraph, line 1, “In line with previous research (5, 50), the proportion of episodes that participants spent socializing yesterday was a significant predictor of well-being (SI Appendix, Table S1).” should instead appear as “In line with previous research (5, 50), the proportion of episodes that participants spent socializing yesterday (see SI Appendix for more details on how this variable is calculated) was a significant predictor of well-being (SI Appendix, Table S1).” The online version has been corrected.

The authors note that, in the SI Appendix, page 7, additional text should be added regarding calculation of “Proportion of Episodes Spent Socializing.”

The added text includes: “Participants were able to report the presence of multiple categories of interaction partner (e.g., one friend and one coworker) for any given interaction, allowing two ways of calculating total amount of social interaction across a given number of interactions. Based on the data structure in Studies 1 to 3, we treat the presence of any given relational category as a separate count; as a result, we take the sum of all reported interactions with any given relational category (even if they happened during the same "episode") and use this to calculate the proportion of episodes spent socializing. However, one could also treat the presence of multiple relational categories during a single episode as a single instance—which aligns with the way that this control variable was calculated in Study 4.

To illustrate these two alternative calculations with an example: If someone reported interacting with friend(s) and coworker(s) during Episode 1, family member(s) during Episode 2, a spouse during Episode 3, and being alone during Episode 4, we calculated proportion of episodes spent socializing as follows:

• (1 interaction with friend(s) + 1 interaction with coworker(s) + 1 interactions with family member(s) + 1 interactions with spouse) = 4 social interactions

• (4 social interactions) / (4 social interactions + 1 episode alone) = 0.80

Alternatively, if one considers interactions with multiple relational categories as single instances, the calculation would be as follows:

• (1 interaction with friend(s)/coworker(s) + 1 interactions with family member(s) + 1 interactions with spouse) = 3 social interactions

• (3 social interactions) / (3 social interactions + 1 episode alone) = 0.75

All of our results and interpretations hold when we consider episodes with multiple relational categories as single instances in Studies 1 to 3. To illustrate this, the main results (predicting subjective well-being from relational diversity controlling for proportion of episodes spent socializing) across Studies 1 to 3 for each of these two ways of calculating our control variable are included in the table below.

This table shows the standardized coefficients of relational diversity on each outcome variable, controlling for the proportion of episodes spent socializing. The table compares two different ways of calculating the proportion of episodes spent socializing: The first column shows the calculation included in the manuscript, which considers interactions with multiple relational categories as multiple instances, and the second column shows an alternative calculation, which considers interactions with multiple relational categories as single instances. All results are consistent across both versions of analysis.”

The SI Appendix was also updated to include the table mentioned above. The SI Appendix has been corrected online.

The authors confirm that all of their results hold when considering interactions with multiple relational categories as single instances.

Primary calculation (Results included in manuscript)	Alternative calculation
Study 1
Subjective well-being
β = 0.13, b = 0.54, 95% CI [0.15, 0.92], P = 0.007	β = 0.14, b = 0.55, 95% CI [0.18, 0.91], P = 0.004
Study 2
Quality of life
β = 0.08, b = 0.36, 95% CI [0.28, 0.43], P < 0.001	β = 0.08, b = 0.38, 95% CI [0.31, 0.45], P < 0.001
Subjective health
β = 0.07, b = 0.17, 95% CI [0.13, 0.21], P < 0.001	β = 0.06, b = 0.14, 95% CI [0.12, 0.17], P < 0.001
Study 3
Life satisfaction
β = 0.04, b = 0.07, 95% CI [0.03, 0.11], P < 0.001	β = 0.04, b = 0.07, 95% CI [0.04, 0.11], P < 0.001
Subjective health
β = 0.07, b = 0.13, 95% CI [0.09, 0.18], P < 0.001	β = 0.07, b = 0.13, 95% CI [0.09, 0.17], P < 0.001
Doctor visits
β = −0.04, b = −0.61, 95% CI [−1.04, −0.19], P = 0.004	β = −0.04, b = −0.70, 95% CI [−1.09, −0.32], P < 0.001

Supplementary Material

Appendix 01 (PDF)