Studying Online Activism: The Effects of Sampling Design on Findings

Abstract

*† Social movement scholars are increasingly interested in Internet activism but have struggled to find robust methods for identifying cases, particularly representative samples of online protest content, given that no population list exists. This article reviews early approaches to this problem, focusing on three recent case sampling designs that attempt to address this problem. The first approach purposively samples from an organizationally based sampling frame. The second approach randomly samples from a SMO-based sampling frame. The third approach mimics user routines to identify populations of “reachable” websites on a given topic, which are then randomly sampled. For each approach, I examine the sampling frame and sampling method to understand how cases were selected, outline the assumptions built into the overall sampling design, and discuss an exemplary research project employing each design. Comparisons of findings from these exemplar studies indicate that sampling designs are extremely consequential. I close by recommending best practices. Information communication technologies (ICTs) are increasingly playing an important role in protest and broader social movements. It is critical, therefore, that scholars design and execute rigorous research programs to investigate how ICTs are used by protesters and organizers and how that usage affects social movements substantively and theoretically. This kind of investtigation often quickly turns to studying protest-relevant material that can be found online in search of both descriptive and causal insights. For instance, descriptive questions such as the percentage of protest-related websites that support offline protests and/or offer online avenues for protest participation are important issues, as are the relationships between organizational sponsorship and kinds of activities offered. The standard resolution to the need for population estimates is to identify an excellent sampling frame and then randomly sample from it. Unfortunately, it has proven quite difficult to identify population lists of online protest content or actions that can serve as good sampling frames, which has substantially complicated case selection and made understanding the contours of protest content online much more difficult. In the face of such a daunting methodological dilemma, a variety of approaches to case selection have been pursued. For instance, case studies where cases were selected because of their notoriety, popularity, success, or the importance of the offline organizations sponsoring selected websites have been common (e.g., Bennett and Fielding 1999; Martinez-Torres 2001) because they side-step the need for sampling. Nonetheless, a number of scholars have tried to move beyond single case studies to look at sets of websites, or what one could think of loosely as various types of samples. This article