Download poster presented at the State Politics and Policy Conference 2018 here
Extant research has established the notion that political attitudes on a range of issues differ significantly from state to state. Some scholars have argued for the existence of state-specific causes of ideology as a result of their history and culture (Erikson et al. 1993). Empirical testing of this premise faces two challenges: One, culture is an ill-defined concept that is notoriously difficult to measure. Two, conducting causal inference on a theoretical construct assumed to be the product of subtle and long-ranging historical processes is highly problematic. I approach this problem by exploiting a specific type of natural experiment - a geographic regression discontinuity design. Under this scheme, the treatment is assigned by the state border, and the populations of interest are the residents living directly on either side. In order to make full use of this approach, very fine-grained data is required. To this end, I turn to geo-tagged Twitter data, which I mine for information via sentiment analysis. Furthermore, I test culture through spatial densities of two types of establishments at the core of the American psyche: churches and gun stores.