ABSTRACT
To analyse the daily interactions of the repression-protest nexus I employ vector autoregression (VAR) (Freeman 1983; Freeman et al. 1989; Sims 1987).1
VAR models can accommodate dynamics, specific concepts of causality and reciprocal relationships, and are therefore particularly well suited to the empirical investigation of the two-way relationship between protest and repression. The advantages of VAR models are that they are based on weak theoretical assumptions and that all variables are treated as endogenous. Hence, it does not have to be predetermined whether, and in what way, protest affects repression or, rather, whether repression affects protest in certain ways (Davis and Ward 1990). The researcher selects certain variables without having to specify the type and direction of the relationship between them. The goal of VARs is ‘to find important interrelationships among the variables’ (Enders 1995: 301). Although VAR models have been criticized for being atheoretical, the weak theoretical assumptions and the flexibility are an advantage in this context, bearing in mind the different and contradicting theories in this area of research.
