Association Rule Analysis of the Repression-Dissent Dynamics
This paper attempts to overcome such a methodological deadlock by using a data set of popular contention in Great Britain (BRIT) collected by Charles Tilly. Unlike most event data sets, BRIT records the information about detailed sequences of contentious interactions within each event. This provides scholars with an unprecedented opportunity to examine contentious dynamics quantitatively. This paper applies a new method, “association rule analysis,” developed in the field of text mining of big data, to examine the repression-dissent dynamics. The method enables researchers to detect “rules of associations” or hidden patterns of contentious sequences in the form of probability statement (i.e. “the probability of police beating followed by students’ rioting is 42 %”). This paper uncovers, among others, state agents who are more likely to trigger civil violence, social actors who are more likely to resort to violence after state repression, and repression strategies which tend to evoke strong civil resistance. The association rule analysis will advance theories and methodologies concerning the dynamics of social movements.