The Center for Data Innovation spoke with Dr. Rose McDermott, a professor of Political Science at Brown University who recently led a study that models divorce as a function of network structure. Along with network scientists Nicholas Christakis and James Fowler, McDermott found that study participants were 75% more likely to get divorced if a friend is divorced. The research is published in the journal Social Forces.
Travis Korte: Tell me about the early stages of this research question. What gave you the inkling that divorce might be modeled as a social contagion?
Rose McDermott: My own divorce. I noticed a bunch of people I knew getting divorced around the same time and I wondered if it might be true in a research as me-search kind of way. I happened to mention my speculation to James and he said he and Nicholas had a way to test for that in their Framingham Heart Study data.
TK: Other than the finding that participants were 75% more likely to get divorced if a friend is divorced, what would you consider to be the study’s other major findings?
RM: That the effect extends to 2 degrees of separation, so you are affected by the friends of your friends, even if you don’t know them. That popular people are less likely to get divorced. That divorced people are less popular (I think this really matters since I think it is because people, especially couples, often stop being friends with divorced people and that social isolation really exacerbates the bad effects of divorce, particularly when recognized in combination with the data on how single/divorced men aren’t as healthy and don’t live as long).
TK: What other social variables, besides divorce and obesity, do you suspect might be able to be studied with networks in the future?
RM: James and Nicholas have looked at a lot of things, including smoking and several aspects of political behavior. I want to be clear that I do not do this work with them. My other work on divorce, with other colleagues, looks at the possible genetic contribution to divorce using an entirely different data set. So my primary interest in this study was divorce, not social networks in particular, while James and Nicholas were more interested in social networks.
TK: Talk to me about re-using data from the Framingham Heart Study. Would your research have been possible without it?
RM: Not this study in this way, although it might have been possible to do a similar longitudinal study but it would have taken 30 years or more in order to create the panel and follow them for as long as we had data from the Framingham study. This study is just a huge benefit and gift to science in many ways across many domains of study, encompassing widespread areas of investigation from medical to social issues.