Data from social media sites provide exceptional opportunities as well as incredible risks for researchers. They provide a window into peoples’ everyday lives, are untainted by researcher intervention, and provide a depth of data that allows new sorts of analyses for research. At the same time, however, we know little about where social media data come from, both in terms of the set of users who are posting and the social and psychological processes underlying the generation of the data. In this talk, I discuss an ongoing project to understand what the mass of data researchers receive implies about the processes that lead to data generation and work through some of the benefits and perils of various techniques for social media data analysis.