Information exchange and multiple peer groups: A natural experiment in an online community

We utilise a quasi-experimental setup to identify causal effects of having additional peer groups on information exchange in a large online maternity community. The information exchange is a key performance indicator for the community as well as a public good among users. Pregnant users join default peer groups based on estimated due date (EDD). Natural uncertainties of EDD can lead to multiple peer groups. Using EDD as an instrumental variable, we find that additional peer group(s) reduces information exchange in both default peer group and total peer groups. Having more advanced groups mitigates the reduction, likely due to information spillovers.