Alan S. Gerber, Gregory A. Huber, Daniel R. Biggers, and David J. Hendry
Proponents of participatory democracy frequently invoke the “transformative” effect of political participation. Such engagement, they argue, encourages the development of a democratic character that leads to a heightened political efficacy, less anti-social behavior, a realization of one’s stake in the existing system, and an improved respect for the democratic process and legal institutions. One anticipated consequence of this new character is a reduction in the inclination to partake in criminal activities. Prior studies provide modest evidence supporting this expectation, finding that those who report previously voting are less likely to report being convicted of a crime, and that those who vote after previously engaging in criminal activity are less likely to recidivate than those who do not.
Our recent article in Political Behavior challenges those findings through an improved empirical approach. While those who vote are much less likely than those who do not vote to be convicted of crimes, this does not demonstrate that participation causes a reduction in criminal tendencies. Instead, there are many ways in which those who choose to vote differ from those who do not, and those differences (such as greater engagement in community life or greater educational attainment) might also be directly related to the propensity to engage in criminal behavior. This concern is common with research based on observational data, as the treatment (voting) is not randomly assigned to individuals, and scholars are often unable to control for the myriad factors that influence both the decision to vote and the decision to commit a crime. As such, it is impossible to know whether the observed large, negative effect of participation on one’s criminal inclination derives from the act of voting or those other, unobserved factors.
We address this concern by exploiting a voter mobilization field experiment prior to the 2010 midterm election involving over 550,000 non-white individuals aged 18-20, in which some subjects were sent a non-partisan registration mailing. The experiment provides an ideal population for our study, given that among those who eventually engage in criminal behavior, criminality largely begins prior to turning 20. Its successfulness (turnout in the treatment group was higher than in the control group) allows us to overcome the issue of nonrandom treatment assignment; that is, we do not need to rely on naturalistic variation in participation, as the assignment to receive the letter was unrelated to any differences in expected participation and criminality.
After merging administrative records on incarceration to our field experiment sample, we confirm the results of prior work that those who vote are less likely that those who do not vote to partake in subsequent criminal activity. When we (through an instrumental variables approach) use the randomly induced variation from the experiment to assess the effect of participation on the risk of being incarcerated, however, we find that voting does not, in fact, reduce criminality. This is true for the overall sample, as well as for those estimated (based on pretreatment covariates) to be at the lowest and highest risks of incarceration. Furthermore, the results are robust to limiting the sample to states where we know that incarceration postdates the 2010 election or for which we have information about a broader set of state supervision statuses (e.g., parole and/or probation).
Our findings have several important implications. For one, it does not appear that the direct effect of voting produces a measurable reduction in criminal activity. It also does not appear that voting immediately triggers the development of a democratic character, at least sufficiently to reduce the propensity of incarceration. This does not mean that proponents of participatory democracy are necessarily wrong about the transformative effects of participation. That transformation, for example, may require multiple participatory acts. Or, voting may be too weak a form of participation, with only more involved and engaging types of participation in the process of debating, governing, and compromising encouraging the development of that character. Either way, while voting may be good for people, it does not appear to stop them from