A basic tenet of representative democracy is that citizens communicate their political preferences to decision-makers by participating in politics. Stimulating political participation is therefore desirable to ensure a tighter link between citizens’ opinions and the policies that politicians enact.
But why do people participate in politics? One factor has traditionally been considered of undisputed importance: education. Through multiple channels—including increased civic skills and knowledge, stronger civic dutifulness, and mobilization through a politically engaged network—education has been argued to stimulate political participation.
In recent years, the conventional wisdom of education stimulating political participation has been challenged by a number of sophisticated studies. They claim that the observed relationship between education and political participation is the result of individuals with high and low levels of education differing in a number of hard-to-observe ways not taken into account by traditional analyses (i.e. these analyses compare “incomparable” high and low educated individuals).
In the absence of randomized experiments in which education is randomly assigned to individuals (which are extremely rare), it is very difficult to rule out that other factors could explain the relationship between education and participation. But how, then, can we assess the relationship between education and participation?
This study takes a—to our knowledge—hitherto unutilized approach within political science: the co-twin control design. The rationale behind this approach is to study the relationship between education and political participation within monozygotic (“identical”) twin pairs to bypass some of the factors, which may confound this relationship in previous studies. Because monozygotic twins share both genes and family environment, we can rule out confounding influences stemming from both. We thus ask if the twin with the longest education also tends to participate most in politics?
We study the relationship between multiple measures of education and political participation based on twin samples from three countries: the United States, Denmark, and Sweden. Two key results stand out from the empirical analyses. First, the relationship between education and political participation is drastically overestimated in models not taking into account potential confounding by genes and/or familial environment in all three countries. Second, while markedly reduced when analyzing within twin-pairs, education has a significant positive impact on political participation in the US and Denmark, but not in Sweden. In the US, political participation increases significantly with years of education, whereas it is high school completion that raises participation in Denmark.
While the co-twin approach enables us to go some way in bypassing factors that likely confound the relationship between education and political participation, it is no panacea. Perhaps the twins had different experiences growing up—or even earlier, in the womb—which set them on different educational paths and also influenced their subsequent political participation? We address a number of such potential influences—both pre- and postnatal—as explanations for the observed significant relationships, but find no evidence for such confounding.
In summary, taking a fresh perspective by utilizing within twin-pair variation in education, this study provides new evidence on the much-discussed effect of education on political participation. Specifically, we show that a limited effect of education on participation seems to exist in the United States and Denmark, but not in Sweden.