This week I am attending the Joint Statistical Meetings (JSM), held here in Seattle, WA. On Wednesday I will be presenting “Social Interaction in Activity-Based Online Communities” joint work with Prof. Zack W. Almquist at the Networks and Social session.
Title: Social Interaction in Activity-Based Online Communities
Abstract: People adjust not only their opinions, but also their behaviors based on both direct and indirect interaction. Questions about influence are particularly salient for activity-based behaviors, (e.g. biking); indeed, much attention has been paid to promoting healthy habits through social interaction. We explore the relationship between interaction and networks utilizing a large-scale dataset from a social fitness community over several months. Ideally, one would like to draw a random sample of users, however, in practice this is rarely feasible; in our case, the platform does not allow queries based on a user identifier. Instead we employ a rejection sampling strategy on the set of all activities logged by users. One implication of this sampling strategy is that from a user perspective the resulting dataset will be biased towards highly active users. We derive unbiased mean and variance estimators and sample weights. We explore how social characteristics, such as gender, are related to activity, interaction, and social structure. These results have implications for studies of peer influence and offer insight into the social structures that develop in activity-based communities.