Situations that are inherently strategic are often analysed using quantitative data and standard statistical estimators that do not take strategic interaction into account. For example, empirical studies on government repression often test hypotheses using ordinary least squares (OLS) models (see Hill and Jones, 2014, which is also an example). Civil conflict is often analysed using logit models (e.g. Cederman et al., 2010). However, Signorino (1999, 2003) shows in the context of limited discrete outcomes that if the theoretical model that is used to derive empirical expectations is strategic, non-strategic statistical models are going to be misspecified.
Signorino (1999) utilises McKelvey and Palfrey's (1995, 1996, 1998) concept of agent quantal response equilibrium to derive a strategic statistical estimator for discrete outcomes. Initially, this estimator was used for explaining states' interactions in international crises that might lead to full-fledged war (Signorino, 1999)1. But strategic models have also been used in other contexts. For example, Carson (2005) uses a strategic estimator to analyse the competition of candidates in US House and Senate elections.
This chapter illustrates strategic statistical estimation as developed in Signorino (1999, 2003) by analysing the interaction between a government and a domestic challenger that might lead to repression and/or rebellion. ...