Autocorrelation, the similarity of linked pairs in a network, is a ubiquitous property in social networks and its prevalence is the foundation of relational learning techniques. However, the exact reason why autocorrelation appears or what processes generate it in networks is unknown. Social psychology has presented two possible sources: social influence, where individuals adopt the properties of their friends over time, and homophily, where individuals select friends that are already similar to themselves. My work has been to develop hypothesis test techniques to detect whether significant homophily or social influence is present in an evolving social network, and to attempt to compare the relative strength of each of the effects as a source of autocorrelation.