Principal investigator: Pedro Jorge Holanda Alves
Article title: CAN INCARCERATED FELONS BE (RE)INTEGRATED INTO THE POLITICAL SYSTEM? RESULTS FROM A FIELD EXPERIMENT
Article authors: Alan S. Gerber, Gregory A. Huber, Marc Meredith, Daniel R. Biggers, and David J. Hendry Aarhus
Location of the InterventionConnecticut, United States
Sample Size: 6.280 observations
SectorEconomic Policy and Governance
Type of InterventionThe effect of incarceration rates on political participation.
Main Variable of InterestVotes of newly freed prisoners
Evaluation method: Experimental Evaluation (RCT)
Policy Problem
In the United States, there has been a growing participation of the adult population in crime over the last few decades. With the exception of Maine and Vermont, all states mandate that any infraction of the law resulting in imprisonment is punishable by disqualification from voting. Some states are even stricter, even after the offender has been released from the judicial system. With a large number of individuals serving or having served prison sentences, the deprivation of voting rights has significant consequences for election results, and the participation of these individuals can be decisive for the democratic outcome. However, few of these individuals know that they retain the right to vote even after breaking the law, and in other cases, individuals fear how the population will react in certain situations.
Is there any type of policy that can help former prisoners integrate into political life? Contact with the justice system can cause disturbances among convicts and establish negative associations with the government, so that imprisonment leads to reduced political participation. In recent years, several states have adopted policies and reforms to increase the participation of criminals at the polls. Are these efforts effective? Is it possible to convince these criminals to register and vote?
To answer these questions, Gerber et al. (2014) conducted a field experiment in Connecticut to evaluate the effectiveness of efforts to encourage released criminals to register to vote and vote. Connecticut is a neighboring state of Massachusetts in the United States, where legally, people who have served their sentences have civil rights just like any other person, and newly released or paroleed criminals have the freedom to register and vote.
Assessment Context
The state of Connecticut is a region that establishes several steps that disqualify convicted criminals, but at the same time, restore their voting rights once a criminal has completed certain steps. If a criminal is convicted and has a voter registration, this registration is expunged from the active voter list. Therefore, after serving their sentence, former criminals should know that they can register again and must do so to become eligible.
Working in conjunction with the Connecticut state secretary, the authors randomly selected unregistered free criminals who were eligible to register and vote. The research separated these individuals into two groups, one of which received an outreach message informing free criminals of their eligibility to participate, while the other group received no contact.
Policy Details
For the experimental design, the Connecticut Secretary of State provided all monthly records of all individuals convicted of a crime between July 1, 2009, and June 30, 2012, and the list of all individuals released from prison and waived from parole during that same period. In total, the authors identified 6.441 records of former felons who had been convicted of less serious crimes and were still out of prison. To identify active records, the authors compared the list of former felons with a voter registration file issued on August 22, 2012. Date of birth, name, and address information were used to match the data.
Next, these records were randomly distributed into three groups: 3.218 cases were assigned to the group of individuals who did not receive voting information (control group), and two different treatment groups, each with 1.611 and 1.612 cases, respectively. In addition to randomization, comparisons between the control and treatment groups show no imbalances in observations by age, time since release from prison, time served in prison, and attendance behavior in 2008. After excluding some inconsistencies, the authors retained 6.280 records, of which 3.134 were the control group and 3.146 were for the two treatment groups (1.574 and 1.572).
Methodology Details
With a correctly designed experiment, it is possible to apply a simple linear regression method, which was the first method adopted by the authors. In a mean difference test between the two treatment groups, the results already show that the treatments had large differences and that there seems to be little difference between the two treatments. For this reason, all subsequent analyses were performed using both treatments together and considering only comparisons between the grouped treatments and the control groups.
However, since not all correspondences were delivered, this result is limited by the estimates representing the Intention to Treat (ITT), which are likely substantially lower than if the authors had been able to contact the entire treatment group (ATE). Therefore, the authors also use a simple two-stage linear regression method in their estimations to find the local mean treatment effect (LATE).
Approximately one week before the deadline to register by mail (and two weeks before the deadline for in-person registration), each former felon in the two treaty groups received an envelope bearing the State Secretariat's stamp providing general information related to the upcoming elections, civic duties, and voting responsibility, and informing them that they were eligible to vote. In one treaty group, the envelope contained "guarantee" and only this information. In the other treaty group, the envelope contained "expanded guarantee" and an additional paragraph alleviating any concerns related to how their criminal status might lead to them being rejected at the polls or any public embarrassment.
For analysis of the results, two periods are separated: first, the authors analyze the registration number for the November 2012 election and the notation in the elections of that same year. Then, the same experiment was analyzed using post-election data, provided in January 2013. Individuals who were not found in the post-election file were coded as individuals who did not register or vote.
Results
Since the experiment can be randomized with minimal bias, the authors analyze their experiment using simple linear regression. As a consistency test, the authors analyze the regression with only the simple specification of the treatment and also with the inclusion of other variables. Their results indicate that regardless of the specification, the estimates remain unchanged, indicating that the postal policy increases the probability of registration by 1,8 percentage points. The results also show that for those individuals who did not vote in 2008, this result was also significant, showing an increase of approximately 0,9 percentage points in the probability of these individuals voting. This means that receiving a single letter from the Secretary of State predicts an increase in registration to 7,8% (6,0 + 1,8) and in voter turnout to 4,2% (3,3 + 0,9).
Who are these mobilized criminals? In short, these results consistently show that only a portion of the criminal population is interested in politics and is easily mobilized to engage in electoral politics. Those individuals who voted in 2008 are the ones who respond best to outreach efforts. That is, simply reminding criminals of their votes makes former criminals much more likely to vote again.
The results also show that released criminals who received a message informing them were 1,8 percentage points more likely to register and 0,9 percentage points more likely to vote in the November 2012 election. These estimates under treatment represent a proportional increase in registration and voting rates of approximately 30%.
Lessons in Public Policy
The authors' conclusions indicate that simple policy interventions aimed at encouraging released criminals appear to be worthwhile. Given that a single contact generated these effects, this research suggests that more sustained outreach could be even more effective. These results show the clear value of experimental testing in reaching a population that is restricted, but also important and growing.
Reference:
GERBER, Alan S. et al. Can incarcerated felons be (Re)integrated into the political system? Results from a field experiment. American Journal of Political Science, vol. 59, no. 4, p. 912-926, 2015.
