Religion, Identity, and Political Preferences

Scholars have long argued that religion is a major driver of political preferences. This effect is hypothesized to be both direct—religiosity itself making people favor conservative social policies—and indirect—religious leaders influencing their followers by endorsing specific candidates or policies. Notwithstanding the empirical correlation between religious affiliation and right-wing voting, causal evidence remains scarce. The main empirical problem is that religion is not randomly assigned. Instead, people choose their religious affiliation based on observed and unobserved factors such as income, geographic location, personal experiences, and cultural fit. Because these factors are also likely to affect people’s political preferences, it remains unclear whether the observed relationship between religion and politics is a spurious consequence of omitted variables, or a genuine causal connection. Complicating matters further is the possibility of reverse causality, whereby conservative ideologies may also affect church attendance. In the current paper, we examine the causal effect of religiosity on political preferences. To address the endogeneity problem, we use clergy scandals as a source of exogenous variation in religious affiliation (Bottan and Perez-Truglia, 2015). We present three sets of results. First, using data from the Freshman Survey and new developments in the difference-in-differences literature, we replicate the finding by Bottan and Perez-Truglia (2015) that scandals reduce Catholicism. On average, following a clergy scandal in one’s home zip code, the probability of identifying as Catholic decreases by 1.5 percentage points. Scandals additionally cause a decrease in the number of Catholic schools and registered adherents at the county-level. Based on these results, we can use an instrumented difference-indifferences analysis that only considers scandal-induced variation in Catholicism to study the causal effect of religion on political preferences and behavior. Our second set of results shows that Catholicism causes adherents to support the Republican party. The effect size is large, as we document that converting from Catholicism to Atheism shifts one’s political orientation leftward by approximately 1 point on a five-point scale. When we consider aggregate voting patterns, we observe similar effects, with scandals causing a significant increase in Democrat vote shares in presidential, Senate, and House of Representatives elections. We furthermore find that the progressive shift in voting patterns is mirrored by a corresponding increase in county-level political donations to Democrat candidates, as well as a decline in Republican contributions. Our third set of results examines the effect of Catholicism on broader set of societal beliefs and policy preferences. To our best knowledge, we are the first to show that Catholics are not only less favorable towards homosexuality, abortion, and taxing the rich, but that these beliefs are caused by Catholicism. Using the same instrumented diff-in-diff as before, we find a consistent shift towards more conservative beliefs in all these domains. At the same time, our findings indicate that Catholicism leads to more progressive beliefs about the government’s role in protecting the environment and the need for national health care. Taken together, we find that religion has a profound effect on people’s policy preferences.