Abstract
Background
Obesity is a known risk factor for cardiovascular disease risk factors, including hypertension and type II diabetes. Although numerous weight loss interventions have demonstrated efficacy, there is considerably less evidence about the theoretical mechanisms through which they work. Delivering lifestyle behavior change interventions via social media provides unique opportunities for understanding mechanisms of intervention effects. Server data collected directly from web-based platforms can provide detailed, real-time behavioral information over the course of intervention programs that can be used to understand how interventions work.
Objective
The objective of this study was to demonstrate how social network analysis can facilitate our understanding of the mechanisms underlying a social media–based weight loss intervention.
Methods
We performed secondary analysis by using data from a pilot study that delivered a dietary and physical activity intervention to a group of participants via Facebook. We mapped out participants’ interaction networks over the 12-week intervention period and linked participants’ network characteristics (eg, in-degree, out-degree, network constraint) to participants’ changes in theoretical mediators (ie, dietary knowledge, perceived social support, self-efficacy) and weight loss by using regression analysis. We also performed mediation analyses to explore how the effects of social network measures on weight loss could be mediated by the aforementioned theoretical mediators.
Results
In this analysis, 47 participants from 2 waves completed the study and were included. We found that increases in the number of posts, comments, and reactions significantly predicted weight loss (β=–.94, P=.04); receiving comments positively predicted changes in self-efficacy (β=7.81, P=.009), and the degree to which one’s network neighbors are tightly connected with each other weakly predicted changes in perceived social support (β=7.70, P=.08). In addition, change in self-efficacy mediated the relationship between receiving comments and weight loss (β=–.89, P=.02).
Conclusions
Our analyses using data from this pilot study linked participants’ network characteristics with changes in several important study outcomes of interest such as self-efficacy, social support, and weight. Our results point to the potential of using social network analysis to understand the social processes and mechanisms through which web-based behavioral interventions affect participants’ psychological and behavioral outcomes. Future studies are warranted to validate our results and to further explore the relationship between network dynamics and study outcomes in similar and larger trials.