Humans are social creatures in nature. Through interactions, we form and maintain relationships, which provide benefits to our physical and mental well-being. However, not all interactions are beneficial as they inevitably involve both positive and negative experiences. Research has shown that while positive interactions (those that are pleasant and enjoyable) are associated with better well-being, negative interactions (containing conflicts or poor treatment) can bring more harm. However, literature has not extensively examined what contributes to a positive and negative interaction, other than partner type and social support behaviors. Therefore, this thesis studies the factors that make an interaction positive or negative and whether they have any impact on one's well-being.

This thesis approaches this by examining the effect of interaction details, i.e., what happens in a social interaction such as who is involved, what joint activities are done, where the interaction occurs, whether there are exchanges of support behaviors, and etc. Specifically, the thesis queries how these interaction details affect the positive or negative experience of the interaction and well-being. Using Ecological Momentary Assessments, I conducted three separate longitudinal studies with a total of over 800 local and national participants. The data showed that interactions that involve close partners or contain joint activities and exchanges of support are rated more positively than their counterparts. More importantly, these interaction details have both direct and indirect impact on well-being. For example, interactions where people provide or receive support have direct associations with better well-being at the end of the day. Interactions that involve close ties and doing joint activities have indirect associations with better well-being by contributing to more positive interactions. In addition, the studies show that the interaction details do not explain the negative interactions.

This theoretical contribution, i.e., what happens in a social interaction can impact well-being, has both practical and technical implications. One benefit is its potential to lead to actionable recommendations for people who are willing to make changes to their social lives for a more thriving life. In addition, examining interaction details provides a tangible way to measure aspects of one's social life that matter for well-being. The thesis explores the possibility of using sensors embedded in mobile phones to automatically predict occurrences of social interactions and what happens in them. While the prediction performance did not work as well as hoped, it performed better than change, suggesting that there is useful information in the mobile sensed data to predict the medium of an interaction, whether it involves a close tie, and what activity is done. Based on the prediction results, the work discusses barriers and challenges to practical deployment of such systems in the near-term.

In summary, this work contributes: 1) theoretical understanding of how interaction details affect the experience of the interactions and well-being; 2) practical and actionable recommendations on changes one can make to their social lives for better well-being; and 3) technical implications on how to use mobile sensors to passively measure one's social life and what to measure.

150 pages

Thesis Committee:
Jason I. Hong (Co-Chair)
Robert E. Kraut (Co-Chair)
Mayank Goel
Eden Litt (Facebook)

Jodi Forlizzi, Head, Human-Computer Interaction Institute
Martial Hebert, Dean, School of Computer Science

Return to: SCS Technical Report Collection
School of Computer Science homepage

This page maintained by reports@cs.cmu.edu