Transmitting Obesity through Social Networks: The Underwhelming Influence of Social Norms
In a study published in 2007 in The New England Journal of Medicine, Christakis and Fowler found that obesity is clustered in social networks and that having a close relationship with someone who becomes obese drastically increases an individual’s chance of becoming obese. This study (also mentioned in another blog) begs the question of how social interactions and relationships influence an individual’s Body Mass Index (BMI) and likelihood of becoming obese.
In their 2009 study, published online through The American Journal of Public Health: First Look, Hruschka, Brewis, Wutich, and Morin attempt to answer this question. They investigate the influence of social norms regarding obesity within networks. They identify three norms that may spread through networks to influence individual’s BMIs, and discuss the potential mechanisms through which this influence occurs. The norms they consider are 1) ideal body size, measured through individuals’ choices based on line drawings; 2) antiobesity preference, measured through pairwise preferences between obesity and other socially stigmatized conditions, including depression, herpes, and alcoholism; and 3) antifat stigma, measured by the degree to which individuals agreed with statements such as “Obesity happens when people do not have self control.”
The results of this study support those of Christakis and Fowler, showing that individuals with strong mutual ties that are obese, especially same-gender familial ties, are significantly more likely to be obese. However, the authors find minimal support for the influence of social norms on BMI similarities. The strongest path of influence that they found, accounting for at most 20% of the relationship between individuals’ BMIs and those of their network members, was that a network member’s BMI may influence an individual’s ideal body size, which could in turn influence the individual’s BMI. The authors conclude, “These analyses provide only limited support for the proposition that social norms of acceptable body size account for observed patterns of social clustering in obesity. If shared social norms are not the primary culprit, then this finding has implications for the kinds of interventions that would be most effective at reversing current obesity trends.” (Hruschka et al., e5)
Besides offering insight into the potential effectiveness of norms-based interventions, this study is significant because it offers a model of structural network analysis that attempts to capture the subtlety of network ties. The authors find that the strength and direction of ties matters. The influence of mutually close ties is greater than the influence of ties that are identified as close by one party and not by the other. This replicates Christakis and Fowler’s finding that mutual friendships have the strongest effect on obesity. This suggests that symmetric relationships are stronger in terms of transmitting obesity than asymmetric ones. Furthermore, Hruschka et al. found different effects between family members, friends, and romantic partners or spouses. The differing trends between these groups may provide clues to the different types of interactions that can transmit obesity. This study, building on and supporting that of Christakis and Fowler, indicates the importance of carefully characterizing network ties in order to determine the mechanisms through which influence works in the network.
More generally, and most importantly, the Hruschka et al. study highlights the difficulty of tracing influence in networks and of determining the mechanisms through which that influence operates. The authors mention several potential weaknesses of their study, mostly dealing with the effectiveness of their classifications and measurements of the variables they use. Careful definitions and measurements of variables pose difficult methodological issues, compounded by the potential biases of self-reporting and selection bias using human subjects.
It is clear that social and environmental factors influence the spread of obesity. But more understanding of how these factors operate is needed in order to design and implement the most effective means for combating the epidemic. The spread of obesity through networks needs more analysis, as called for by Hruschka et al., especially in terms of how network members interact and how these interactions ultimately impact individuals’ BMI values.