Jump to content

7 In 10 Felons Register As Democrats


Recommended Posts

Jail survey: 7 in 10 felons register as Democrats

 

http://www.washingtonexaminer.com/jail-survey-nearly-34-felons-register-as-democrats/article/2541412

 

A new study of how criminals vote found that most convicts register Democratic, a key reason in why liberal lawmakers and governors are eager for them to get back into the voting booth after their release.

 

“Democrats would benefit from additional ex-felon participation,” said the authoritative study in The Annals of the American Academy of Political and Social Science.

 

The authors, professors from the University of Pennsylvania and Stanford University, found that in some states, felons register Democratic by more than six-to-one. In New York, for example, 61.5 percent of convicts are Democrats, just 9 percent Republican. They also cited a study that found 73 percent of convicts who turn out for presidential elections would vote Democrat.

 

But despite recent moves in states to notify convicts that they can vote again, the study finds little evidence that they do, undercutting Democratic efforts to get them to the polls.

 

 

 

 

The study looked at three states which are reminding convicts that they can vote after leaving jail: New York, New Mexico and North Carolina.

 

They provided the following Democrat-to-Republican breakdown in felon party registration patterns:

 

– New York: 61.5 percent register Democratic, 9 percent register Republican

 

– New Mexico: 51.9 percent Democratic, 10.2 percent Republican

 

– North Carolina: 54.6 percent Democratic, 10.2 percent Republican

 

Overall, said the study, there are 5.8 million eligible voters in jail, about 2.5 percent of the voting age population. Most are young black males, it added.

Edited by grassmatch
Link to comment
Share on other sites

the articles lists its source within the print.

http://ann.sagepub.com/content/651/1/220.full.pdf+html

 

 

 

 

 

Previous research documents widespread confusion about who can and cannot vote among people who have come into contact with the criminal justice system. This research, and considerable activism drawing attention to the issue, has spurred a number of state legislatures to pass laws requiring the states to notify ex-felons about their voting rights. The purpose of this article is to better understand the policy processes that produce these notification laws and to assess whether the laws affect ex-felons’ registration and turnout rates. Data on discharges from the correctional system and voter files are merged from three states that have recently passed notification laws: New Mexico, New York, and North Carolina. Our findings show little evidence of an increase in ex-felon registration or turnout after notification laws are implemented.

felon disenfranchisement

voting rights

notification

 

Criminal disenfranchisement is increasingly a contested criminal justice and electoral administration issue as wide swaths of the electorate are denied the right to vote. States have substantial autonomy to restrict the voting rights of those with criminal convictions, and they have enacted policies that range from no voting restrictions at all to lifetime disenfranchisement. An estimated 5.8 million citizens—or 2.5 percent of the voting age population—nearly half of whom had already completed the terms of their sentence, were legally disenfranchised in 2010 (Uggen, Shannon, and Manza 2012).

 

Estimates of the legal disenfranchisement rate may understate the effects of criminal disenfranchisement if disenfranchisement laws cause citizens with voting rights to not vote. One form of such de facto disenfranchisement occurs when nondisenfranchised individuals believe that they are ineligible to vote (Wood and Bloom 2008). An ex-felon may find it challenging to learn whether he or she is eligible to vote because both the fragmented nature of criminal disenfranchisement policy and the complexity of specific states’ disenfranchisement laws. How and when voting rights are restored can vary across neighboring states and depend on relatively nuanced details about the specific crime of conviction or date of discharge. Thus, it is not surprising that individuals who come into contact with the criminal justice system are frequently misinformed about their voting rights (Drucker and Barreras 2005). Some of this misinformation likely results from correctional officers and election officials also holding incorrect or inconsistent beliefs about the criminal disenfranchisement policy in their state (Ewald 2005; Allen 2011).

 

A number of state legislatures recently considered policies designed to combat this misinformation. According to data collected by the Brennan Center for Justice, almost half of all states have a statute that requires the state to inform the criminally disenfranchised about either the loss or the reinstatement of their voting rights.1 Notification procedures vary across states, with differences in who communicates the information (e.g., a judge, a correctional officer), when the information is communicated (e.g., upon sentencing, at discharge), and how the information is communicated (e.g., verbally or in written form, in person or in a mailing).

 

This article investigates whether these notification requirements increase voter registration and turnout among those discharged from a felony sentence. A number of the states with notification requirements, including New Mexico, New York, and North Carolina, passed these requirements in the last 10 years. By merging criminal justice discharge records with statewide voter records, we estimate registration and turnout rates for those individuals completing their sentence before and after these states implemented notification. If these notification laws correct misperceptions about voting eligibility among individuals who would otherwise vote, the registration and turnout rates of the cohorts discharged just after notification begins should be higher than the registration and turnout rates of the cohorts discharged just before notification begins.

 

However, we find little evidence that notification increases registration and turnout rates in New Mexico, New York, and North Carolina. The presidential election turnout rate among recent ex-felons is about 10 percent both before and after notification in all three states. While we cannot rule out that notification slightly increased registration and turnout, we can rule out the large mobilizing effects of notification that Meredith and Morse (2013) find in a similar population in Iowa. We conclude by discussing a number of potential explanations for why notification appears to have increased registration and turnout in Iowa but not in these other states.

 

 

 

Previous Section

 

Next Section

 

Relevant Literature

 

A growing body of research examines the political consequences of criminal disenfranchisement. The disenfranchised population is disproportionately young, male, African American, and less educated than the general population of voters (Western, Pattillo, and Weiman 2004). African Americans, for example, are about three times more likely to be disenfranchised because of a criminal conviction (Uggen, Shannon, and Manza 2012). This distinct profile suggests that the political preferences of disenfranchised citizens differ from those of the general voting population. Seminal works by Uggen and Manza (2002) and Manza and Uggen (2004, 2006) find that Republican candidates benefit from criminal disenfranchisement. Uggen and Manza reach this conclusion by using data on nonfelons to fit turnout and vote choice models, and then extrapolating the probability that disenfranchised voters would support Democratic candidates if they were eligible to vote. Their models predict that about 35 percent of the disenfranchised population would turn out in a presidential election, with about 73 percent of those who vote supporting the Democratic presidential candidate.

 

Subsequent work questions whether these models overstate the turnout propensities and Democratic preferences of the disenfranchised population. Uggen and Manza’s (2002) approach assumes that the demographic characteristics included in their models—gender, race, age, income, labor force status, marital status, and education—capture all the differences between the voting tendencies of felons and nonfelons. But if the unobserved variables that increase political participation also negatively associate with the propensity to commit crimes, these models overstate the probability that the criminally disenfranchised would otherwise vote. Belief in prosocial norms, for example, causes people to both obey the law and participate in politics (Manza and Uggen 2006). Consistent with this intuition, Hjalmarsson and Lopez (2010) find that observable variables only explain about half of the difference in the turnout rates of those who have been incarcerated and those who have never had contact with the criminal justice system. Studies that use public records to estimate the turnout rates of the nondisenfranchised ex-felon population also usually find smaller turnout rates than predicted by Uggen and Manza’s models (Burch 2007, 2011, 2012; Haselswerdt 2009; Meredith and Morse 2013).

 

Misinformation may be one reason why actual turnout is less than what Uggen and Manza (2002) predict. Previous research shows that many individuals who have contact with the criminal justice system hold incorrect beliefs about their right to vote. Drucker and Barreras (2005) survey 334 individuals under various forms of supervision in Connecticut, New York, and Ohio about their voting rights and find that nearly half of the population answered incorrectly or responded that they did not know when asked whether a felony conviction is permanently disenfranchising. Likewise, few of the incarcerated felons in Minnesota interviewed by Manza and Uggen (2006) understood their future voting rights. Thus, there may be a sizable number of voting-eligible ex-felons who would vote if they did not believe (incorrectly) that they are ineligible to do so.

 

These informational barriers to turnout are likely compounded by the fact that criminal justice officials and election administrators are also frequently misinformed about criminal disenfranchisement policy. Ewald (2005) finds that 37 percent of election administrators either misreport or report being unaware of a key aspect of their state’s criminal disenfranchisement policy. These misinformed elites likely contribute to the misinformation within the population who comes into contact with the criminal justice system. Misinformed elites may also impose additional barriers to political participation for voting-eligible ex-felons. For example, Allen (2011) reports that nearly half of the election boards in New York asked ex-felons who attempted to register to vote to provide documentation that is not required by state law.

 

Many states recently have considered legislative reforms to reduce misinformation about criminal disenfranchisement. Voting rights lobbies such as the Brennan Center and the Sentencing Project have made the passage of voting rights notification bills one of their top priorities. Yet little is known about whether these reforms increase political participation. The only study that we are aware of examines a quasi-experiment in Iowa in which only ex-felons discharged after a certain date were mailed a letter that notified them about the restoration of their voting rights (Meredith and Morse 2013). Meredith and Morse (2013) estimate that being mailed this letter increased ex-felon turnout by nearly 10 percentage points. However, there are a number of reasons why the effect estimated in Iowa may not generalize to other states. Although the governor’s office sent these mailings in Iowa, most states delegate responsibility for administering notification to the criminal justice system. This may affect the probability that individuals receive the notification treatment, the informational content of the treatment itself, and the response to such information. Also, recipients in Iowa were informed of a recent expansion in voting rights, while most notification programs inform recipients about a long-standing policy.

 

 

 

Previous Section

 

Next Section

 

Recent Voting Rights Notification Bills

 

Criminal disenfranchisement policy has been particularly dynamic over the past 15 years, with nearly half of states making at least one policy change (Porter 2010). One reform that a number of states have considered is a requirement that the state notify ex-felons when their voting rights are restored. Table 1 summarizes all such notification bills (to our knowledge) that were introduced in state legislatures between 2003 and 2012.2

 

 

 

View this table: In this window

In a new window

Download to PowerPoint Slide

 

Table 1

Recently Introduced Legislation on Voting Rights Notification

 

 

Because African Americans are disproportionately affected by criminal disenfranchisement, our expectation is that African American legislators will act as policy entrepreneurs on this issue (McMiller 2008; Ewald 2009). Consistent with this expectation, Table 1 indicates that at least one of the primary authors of these notification bills was an African American in six of the seven cases. This pattern is consistent with Haynie’s (2001) argument that descriptive representation is necessary for African Americans to be substantively represented in state legislatures.

 

Another expectation is that Democratic legislators will be more supportive of notification laws. The Democratic and Republican parties’ positions on how election administration should handle tradeoffs between access and integrity in political participation became increasingly differentiated in the early 2000s (Hasen 2012). Moreover, previous research shows that Democrats would benefit from additional ex-felon participation. Consistent with this idea, Table 1 shows that Democratic legislators almost always support notification, while Republican legislators’ support is more mixed. However, some other patterns in Table 1 reveal that partisanship is not the only factor determining whether notification reform is enacted. Notification bills died in committee in both North Carolina in 2005 and Mississippi in 2006 despite Democrats having unified control of both state legislatures. Moreover, notification language was added to an omnibus bill in both North Carolina in 2007 and New York in 2010 after attempts to pass a standalone notification bill in unified Democratic state legislatures stalled. Conversely, a notification law passed both the Texas House and Senate despite unified Republican control.

 

Partisanship does a better job predicting gubernatorial behavior. While Democratic governors signed all three notification bills that they received, Republican governors vetoed two of the three that they received. In a statement accompanying his veto of Minnesota’s 2009 notification bill, Tim Pawlenty wrote that the bill was redundant and, further, that ex-felons needed to take personal responsibility for learning their own rights. In 2007, Texas Governor Rick Perry vetoed a notification bill and made a similar reference to personal responsibility, arguing that it was unfair to provide information about voting rights to ex-felons when the same service was not also provided to law-abiding citizens. Given that both Pawlenty and Perry ran for the 2012 Republican presidential nomination, these vetoes may have been motivated by beliefs about the Republican presidential primary electorates’ preferences regarding ex-felon voting rights.3

 

The remainder of this article evaluates whether the notification laws passed by New Mexico, New York, and North Carolina affect the rates at which ex-felons register and vote.4 Since 2001, voting rights in New Mexico have been restored automatically upon discharge from prison, probation, and parole. An ex-felon who wants to register to vote needs to present a discharge certificate to his or her county election clerk. This certificate, which is presented in Figure A1 in the appendix, informs both the ex-felon and the county election clerk that the ex-felon’s voting rights have been restored. Prior to the 2005 notification law, ex-felons needed to request that the Department of Corrections provide them with this certificate. Starting on July 1, 2005, all ex-felons began automatically receiving this certificate upon discharge.

 

Notification in New York and North Carolina takes a slightly different form than in New Mexico. Both states include a voter registration form and relevant literature in the packets given to individuals upon discharge. Figure A2 and Figure A3 in the appendix show the voting literature that New York began including in discharge packets on October 4, 2010. It is noteworthy that neither form explicitly states that the individual’s voting rights have been restored and that the only reference to criminal disenfranchisement in this literature is a single line on the voter registration form that reads “to register you must . . . not be in prison or parole for a felony conviction.”

 

North Carolina instituted a similar notification program in two separate waves. The North Carolina Department of Public Safety distributed a memo on February 6, 2007, ordering that a brochure produced by the American Civil Liberties Union (ACLU) be included in discharge packets. Over the summer, the state legislature passed and the governor signed a notification law. Starting on October 1, 2007, dischargees received both the brochure displayed in Figures A4 and A5 in the appendix and a registration form in their discharge packets. The brochure distributed after October 1, 2007, is quite similar to the brochure that the ACLU produced. Unlike the literature provided by New York, this brochure explicitly states that voting rights are restored upon discharge, although this information is not centrally located in the brochure.

 

 

 

Previous Section

 

Next Section

 

Constructing Turnout Rates from Public Records

 

The two most common ways to measure registration and turnout in the political science literature are either via actual registration and turnout data aggregated by precinct, county, or state, or self-reported registration and turnout data in individual-level survey data. Unfortunately, neither of these approaches is useful for studying the registration and turnout behavior of ex-felons. Because ex-felons make up a small proportion of the electorate, it is difficult to detect the effects of felon disenfranchisement laws in aggregated turnout data (Miles 2004). Moreover, there is no survey of sufficient size that asks about both criminal history and voter turnout (Burch 2011).

 

Because these traditional ways of measuring registration and turnout are insufficient, we follow the example of recent research, combining information in criminal justice discharge records and state voter files to create our own measures (Burch 2007, 2011, 2012; Haselswerdt 2009; Meredith and Morse 2013). Voter files are databases that contain the name, address, demographic information, and turnout history of all registered voters in a state. These files are public record, although the cost of obtaining the data and the specific demographic information available in the file vary across states (Cooper, Haspel, and Knotts 2009). We search for whether there are records in the state voter file that match the information provided for each record in the discharge data. We then estimate registration and turnout rates by calculating statistics such as the percentage of criminal justice records that match at least one voter file record and the percentage that match at least one record of voting in a given election.

 

Implementing this matching process requires that we define a criterion for what constitutes a match between a record in the corrections data and a record in the voter file.5 Perhaps obviously, we can match only on information that is contained in both the corrections data and voter file. Because there is substantial variation in what information is contained in a state’s corrections data and voter file, it is difficult to construct a common match criterion that can be used across states. In the next section, we detail the various criteria used in each state.

 

Multiple forms of measurement error could cause the percentage of discharge records that match a voter file record to differ from the actual registration rate. One form of measurement error occurs when a corrections record matches another individual’s voter file record. Such measurement error is likely to be more prevalent when individuals have a common name and when there are fewer other common variables between the corrections and voter file data. A second form of measurement error occurs when a corrections record fails to match the individual’s corresponding voter file record. Such measurement is more likely to occur when names are presented differently in the two sources or when the files contain typographical errors. Finally, some ex-felons will have passed away or have moved to other states. Because these different forms of measurement error operate in different directions, it is unclear whether the number of matched discharge records will underestimate or overestimate the actual registration rate.

 

Following Meredith and Morse (2013), we use placebo matching to assess the number of discharge records that match another individual’s voter file record. We permute the felon discharge records so that the date of birth or age is slightly modified. The number of voter file records that match these fake discharge records provides an estimate of the number of erroneous matches in our actual data.

 

Assessing the number of missed true matches is more difficult. To reduce the number of missed matches, we convert variants of a common name back to a root name (e.g., “Mike” and “Mikey” are converted back to “Michael”). When computational power permits, we also expand our matching criterion to include near matches. For example, records are matched if they are the same except that the first names differ by a single character (e.g., “Michele” and “Michelle”) or the last name in one source is a subset of the last name in the other (e.g., “Clinton” and “Rodham Clinton”). However, there are no modifications that can be made to prevent certain errors. For example, Meredith and Morse (2013) show patterns consistent with more female matches being missed because females are more likely to change their last name.

 

Measurement error complicates any attempt to use matched corrections and voter file data to make comparisons of ex-felons’ political participation across states, particularly because typographical errors, one of the primary causes of missing matches, likely vary based on the quality of record keeping in the state. These data are better suited for making the types of intrastate comparisons of political participation that we make in this article. Our primary analysis consists of comparisons of registration and turnout rates for cohorts of individuals discharged before and after the implementation of notification laws. Many of the sources of measurement error in our estimated registration and turnout rates are likely to be constant within a state across time. Those factors that vary across time, such as moving out of state or death, are likely to have similar effects on the cohorts discharged just prior to and after these laws are implemented. Thus, looking at how registration and turnout rates vary among those discharged in close proximity to the implementation of these laws can help us to learn about whether these notification policies cause a change in registration and turnout behavior.

 

 

 

Previous Section

 

Next Section

 

Data

 

We were able to collect discharge records and voter file data from New Mexico, New York, and North Carolina. This section presents the information that is available in the discharge records and voter files in each state. We then detail the procedure used in each state to match these records. Finally, descriptive statistics are presented to illustrate the results of these matches.

 

New York

The New York State Department of Corrections and Community Supervision provided us with ex-felons’ full name, date of birth, current parole status, and parole status date. In New York, individuals convicted of a felony regain their right to vote when they have completed both prison and parole supervision. Thus, our analysis focuses on those individuals whose parole status is listed as “discharged,” with the parole status date indicating the day on which these individuals regained their voting rights. As only an individual’s most recent change in supervision status is observed in these data, the registration and turnout rates estimated for New York should be interpreted as registration and turnout rates for individuals who are discharged from parole and do not subsequently end up back under supervision in New York.6

 

Our copy of the New York voter file contains the full name, date of birth, registration date, party affiliation, and voting history of all individuals registered to vote in New York as of April 2013. Registration date refers to the date that registration was last modified. This differs from the date that registration was initiated in cases where a registrant updated his or her address or party affiliation subsequent to initial registration.

 

Parole discharge records are matched to the voter file using a process nearly identical to that described in Meredith and Morse (2013). First, discharge records are merged to every record in the voter file that reports the same birthdate. Then, the first and last names in the two sources are compared, and those merges with sufficiently similar first and last name are kept. Finally, the middle names in the two sources are compared, and the remaining merges with consistent middle names are kept. Because the middle name is often not reported in one or both sources, middle names are considered to be consistent if either source is missing a middle name.

 

Table 2 presents descriptive statistics of the matches obtained using this matching procedure. The first column of the first row shows that 32.4 percent of the 48,628 discharge records from between November 5, 2008, and September 30, 2012, match a record in the New York voter file. However, a large number of these registration records are inactive, as only 19.3 percent match to an active registration record. If these registrations are the result of notification, we might expect that the registration would occur in close proximity to when the notification takes place. But we observe that only 1.1 percent of discharge records match to a registration record with a registration date within 90 days of the discharge date.

 

 

 

View this table: In this window

In a new window

Download to PowerPoint Slide

 

Table 2

New York Discharge Records Matched to 2013 Voter File

 

 

Those ex-felons who are registered overwhelmingly register as Democrats. Of those discharge records that match to at least one voter file record, 61.5 percent match only to Democratic voter records. In contrast, 25.5 percent match only to voter records with no affiliation or an affiliation with a minor party, while 9 percent match only to Republican voter records. The remaining 4 percent of matched discharged records are cases where the discharge record matches with multiple voter file records that have inconsistent party identification. This could happen because someone changed their partisan identification across time or because a discharge record matches the voter record of two different individuals.7

 

The final three columns of Table 2 show that recently discharged ex-felons in New York vote at extremely low rates. While low turnout in 2008 and 2010 can partially be explained by some of the population being disenfranchised, only 8.3 percent voted once eligible in the 2012 presidential election. As a point of comparison, just over 15 percent of males between the ages of 18 and 44 without a high school degree reported voting in the November 2012 Current Population Survey. The bottom two rows of Table 2 show that the cohort that was discharged before New York began distributing informational pamphlets and voter registration forms turned out at a slightly higher rate in 2012 than the cohort discharged afterward.

 

One concern discussed above is that a discharge record may be matching another individual’s voter registration record. To assess the prevalence of this type of matching, we create a dataset of placebo discharge records that are nearly identical to the actual discharge records and match these placebo records to the voter file. These fake discharge records are constructed by permuting the birthdates in the discharge records by 35 days. Rows two and three of Table 2 show that only a small number of fake discharge records match a voter file record, suggesting that we are unlikely to have many false matches in our actual data.

 

New Mexico

The New Mexico Corrections Department provided us with information on the full name, date of birth, and discharge date for former inmates. Discharge in New Mexico refers to the completion of all prison, parole, and probation supervision, which is when individuals convicted of a felony regain the right to vote. Only an individual’s most recent discharge date is observable in these data. Thus, the registration and turnout rates estimated for New Mexico by discharge date should be interpreted as registration and turnout rates for individuals discharged on a given day who do not subsequently end up back in supervision.

 

Our copy of the New Mexico voter file contains the full name, year of birth, registration date, party affiliation, and voting history of all individuals registered to vote in New Mexico in February 2009. Registration date refers to the date that registration was last modified rather than the date that registration originated.

 

Matching the New Mexico discharge records to the voter file is made more complicated because we observe year of birth, rather than date of birth, in the New Mexico voter file. As a result, we cannot use date of birth as the initial variable on which we screen potential matches. Moreover, initially screening only on year of birth results in too many potential matches to feasibly search given computational constraints. Instead, we exactly match on year of birth, root first name, and root last name, and keep those matches with a consistent middle name in both sources.8

 

Table 3 shows that the characteristics of the matches in New Mexico are broadly similar to the matches in New York. A slightly higher percentage of New Mexico discharge records match to an active record in the New Mexico voter file than in New York. Once again, registered ex-felons in New Mexico tend to be overwhelmingly Democrat: 51.9 percent match to only registered Democrats, 18.9 percent match to only registered Republicans, 21.7 percent match to only individuals registered neither as Democrats nor Republicans, and 7.5 percent match to multiple individuals who affiliate with different parties.9 Only 12.9 percent of discharge records match to a voter file record that shows turnout in the 2008 presidential election, and this rate is nearly identical among the cohorts discharged before and after the state began automatically issuing discharge certificates.

 

 

 

View this table: In this window

In a new window

Download to PowerPoint Slide

 

Table 3

New Mexico Discharge Records Matched to 2009 Voter File

 

 

One difference between New Mexico and New York highlighted by Table 3 is the number of matches that we find between our placebo discharge records and the voter file. When the age in the discharge data is permuted to be two years below and above the actual age, we find that 6.2 percent and 6.5 percent of discharge records match an active registration in the voter file, respectively. Because year of birth is less discriminating than date of birth, this increase in placebo matches is expected. The number of placebo matches indicates that even though the number of matches is somewhat higher in New Mexico, the active registration and turnout rates are likely quite similar in New York and New Mexico.

 

North Carolina

The North Carolina Department of Public Safety provided us with information on the full name, age at discharge, and discharge date for all inmates discharged from the correctional system. Just as in New Mexico, discharge in North Carolina refers to the completion off all prison, parole, and probation supervision, which is when an individual convicted of a felony regains the right to vote. We observe multiple discharge dates if an individual has been discharged from the system multiple times. Thus, unlike in New York and New Mexico, the registration and turnout rates estimated for North Carolina by discharge date should be interpreted as registration and turnout rates unconditional of future supervision status.10

 

Our copy of the North Carolina state voter file contains the full name, age, registration date, party affiliation, and voting history of all individuals registered to vote in North Carolina as of March 2013.11 Registration date refers to the date that registration was last updated rather than the date that registration originated.

 

Matching the discharge records to the voter file is even more complicated in North Carolina than in New Mexico because both sources provide age rather than year of birth or date of birth. As a result, most discharge records could be listed under two potential ages in the voter file.12 Thus, we run the same matching procedure that we used on the New Mexico data but search for records in the voter file that match either of these two potential ages.13 Because we cannot use age to reduce the number of potential matches in the voter file as effectively in North Carolina as in New Mexico, we expect to observe more false positive matches in our placebo data.

 

Table 4 shows that more discharge records match a voter file record in North Carolina than in New Mexico or New York, with 33.2 percent and 21.0 percent of discharge records matching an active voter registration and 2008 turnout record, respectively. As expected, there are also more placebo observations that match a voter file record in North Carolina than in New Mexico. After accounting for these placebo matches, the 2008 turnout rate appears to be just above 10 percent, which is a couple percentage points higher than what we observed in New Mexico in 2008 and New York in 2012.14 Once again ex-felons are more Democratic than the general population of active registrants.15

 

 

 

View this table: In this window

In a new window

Download to PowerPoint Slide

 

Table 4

North Carolina Discharge Records Matched to 2013 Voter File

 

 

Descriptive statistics

Table A1 in the appendix shows the observable characteristics in each state’s discharge data that we relate in the next section to the registration and turnout behavior of ex-felons. We observe an individual’s age, gender, and time since discharge in all three states. The discharged population is both younger and more male than the general population. We also observe an individual’s race in New York and North Carolina. In New York, we observe four categories of race—white, black, Hispanic, and all others—while in North Carolina we only observe three categories—white, black, and all others. About half of the discharged population is black in both New York and North Carolina, with Hispanics making up an additional quarter of the discharged population in New York. Finally, we observe whether an individual was discharged from prison, or from parole or probation and the amount of total time he or she served in North Carolina.

 

 

 

Previous Section

 

Next Section

 

Results

 

Our empirical approach compares the registration and turnout rates of individuals discharged from the criminal justice system before and after New York, New Mexico, and North Carolina implemented notification. Because only those people discharged after the laws were implemented receive the notification treatment, we wish to interpret any difference we observe between the turnout rates of individuals discharged from the criminal justice system before and after implementation as the effect of the law changes. The biggest threat to such an interpretation is that these cohorts differ in other ways aside from their exposure to notification. We adopt an empirical specification that draws on the regression discontinuity design (RDD) literature to account for these other potential differences between those who are discharged before and after notification.

 

Graphical analysis

Our analysis begins by plotting registration and turnout rates as a function of the date of discharge as is standard in RDD analysis. If notification laws are causing ex-felons to register and to vote at greater rates, we should observe an increase in registration and turnout that corresponds to the start of notification. Figure 1 shows little evidence of such an increase in New York. The top panel looks at the percentage of discharge records that match an active voter registration record. This panel shows a downward slopping trend, indicating that individuals are more likely to be registered the more time has passed since discharge. The declines in registration are particularly pronounced for cohorts discharged just after the 2008 presidential election and the 2009 New York City mayoral election.16 Most importantly, we do not observe the jump in registration rates following the implementation of notification that we would expect to observe if notification were causing a large increase in participation.

 

 

Figure 1

View larger version: In this page

In a new window

Download to PowerPoint Slide

 

Figure 1

Registration and 2012 Turnout by Quarter of Discharge in New York

 

 

If individuals were being mobilized to register by the information and registration form provided in their discharge packet, we expect some of the registration to occur in relatively close proximity to their discharge date. However, the middle panel of Figure 1 shows no increase in registration within 90 days of parole discharge after notification begins. Only about 1 percent of individuals register within 90 days of discharge in most quarters. The exceptions are quarters immediately before presidential elections, when a substantially higher percentage of ex-felons immediately register to vote. Thus, while Table 2 shows a slightly higher rate of registration within 90 days in the postnotification cohort, Figure 1 suggests that this likely reflects the mobilizing effects of the 2012 presidential election rather than the effect of notification.

 

The final panel of Figure 1 shows relatively similar rates for the 2012 presidential election turnout in the cohorts discharged just before and after notification begins in New York. Again there appears to be some residual effect of eligibility in the 2008 presidential and 2009 New York City mayoral election on the turnout rates. Among the cohort discharged after 2009 New York City mayoral election, 2012 turnout rates range between 7 and 9 percent with no clear time trend or evidence of a jump after notification.

 

Figure A6 and Figure A7 in the appendix replicate Figure 1 for New Mexico and North Carolina, respectively. Both figures in the appendix are similar to Figure 1 in that they show little difference in registration and turnout rates before and after ex-felons are notified about their voting rights. Figure A6 and Figure A7 both show that registration increases in the lead up to a presidential election. Unlike in New York, there is a less clear relationship between time since discharge and registration and turnout. There also appears to be an increase in registration in both states and 2008 turnout in North Carolina among those discharged immediately before the 2008 presidential election.

 

Put together, these figures cast doubt on the hypothesis that these state’s notification programs dramatically increase registration and turnout. In all three figures, we observe similar registration and turnout rates in the cohorts discharged just before and after the notification laws take effect. As these cohorts should be similar except in their exposure to notification, we would expect an increase in registration and turnout among those discharged immediately after notification begins if these laws were having a large effect on participation. While we cannot rule out that there was a small increase in registration and turnout because of notification based on these figures, we can rule out the 5 to 10 percentage point increase in turnout that Meredith and Morse (2013) estimated to have occurred in Iowa because of notification.

 

Empirical specification

Having observed little graphical evidence of participation differences before and after notification, we next develop an empirical specification that tries to quantify how much participation changed after notification began. The biggest challenge to doing so is that individuals who receive notification differ from those who do not in terms of the amount of time that has passed since discharge. Time since discharge may relate to registration and turnout in a number of ways. Individuals who have been discharged for longer have more opportunity to be exposed to voter registration. Time since discharge also likely reduces the probability of still residing in the state and increases the probability of being disenfranchised again by a new conviction. The characteristics of those individuals discharged before and after the laws are implemented may also differ in both observable and unobservable ways as a result of populations trends or other changes in state law.

 

Separating the effect of notification and time since discharge involves a trade-off between potential bias and efficiency. The most comparable individuals are those who were discharged in close proximity to one another. Thus, comparing the participation of those who were discharged just before and after the implementation of notification provides a straightforward test of the effect of notification. However, restricting the sample in such a way restricts our statistical power to detect small differences in participation between the two cohorts. Expanding the sample to include people discharged in a wider time frame gives us more statistical power to detect differences between the two cohorts, but requires us to make more assumptions about the relationship between time since discharge and political participation. We develop a specification that includes everyone discharged after the previous presidential election to maximize our statistical power, while noting that figures in the previous section also show little difference in the participation rates of individuals discharged just before and just after notification begins.

 

The equation below illustrates how we attempt to separate the effect of notification from the effect of time since discharge. We want to estimate the determinants of whether discharge i takes action Yi (e.g., registers or votes). Our primary parameter of interest, θ, is the coefficient on Notifiedi the indicator for whether a discharge i occurred after the state implemented notification. We also include a 4th order polynomial of DischargeYearsi to control for the direct effect that time since discharge has on political participation in a relatively flexible manner.17 We control for a vector of individual level characteristics, Xi , listed in Table 5 that we observe about dischargees in each state. Assuming that ϵi is distributed normal, we estimate the parameters in the equation using a standard probit analysis.

 

 

 

View this table: In this window

In a new window

Download to PowerPoint Slide

 

Table 5

Determinants of Ex-Felon Registration and Turnout

 

 

 

Formula

 

 

Regression results

The results presented in Table 5 continue to show no evidence of any increase in registration and turnout after notification is implemented. There is not a statistically significant positive coefficient on being discharged after notification begins in any of the six regressions. There is a marginally significant negative coefficient on being discharged after notification in North Carolina, although this result is not robust to alternative specification.

 

While postnotification discharge does not relate to registration and turnout, Table 5 shows that a number of additional characteristics of ex-felons do relate to the probability of registering and voting. Age, which is well known to positively relate to political participation generally in the United States (see for example Rosenstone and Hansen [1993]), also positively relates to turnout in all three states. The increase in the probability of turnout from an additional year of age ranges from 0.16 percentage points in New Mexico to 0.43 percentage points in North Carolina.18 Females also register and vote at higher rates. The difference is particularly stark in New York, where females are about 7 percentage points more likely to be registered and 5 percentage points more likely to have voted in 2012 than their male counterparts. Much smaller differences are observed in North Carolina and New Mexico. These point estimates likely understate the actual gender differences because our matching procedure is more likely to miss female matches (Meredith and Morse 2013).

 

One additional interesting relationship in Table 5 is that between race and participation. Much like Burch (2011), we find that African Americans vote and register at higher rates than whites in both New York and North Carolina. The difference in the probability of black turnout relative to white turnout is about 5 percentage points in New York and 10 percentage points in North Carolina. While some of this difference may relate to Barack Obama’s presence at the top of the ticket in 2008, unreported regressions also show that black turnout was about 2.5 percentage points higher in the 2010 midterm election when Obama was not on the ballot. Conversely, ex-felons who are neither black nor white vote at significantly lower rates in both New York and North Carolina.

 

 

 

Previous Section

 

Next Section

 

Conclusion

 

Concerns about ex-felons being misinformed about their voting rights, in concert with a wide range of activists promoting the issue of notification, has spurred a number of state legislatures to pass laws that require the state to notify ex-felons about their voting rights. The evidence presented in this article suggests that these reforms have been unsuccessful at increasing registration and turnout among the ex-felon population. We reach this conclusion because we observe little difference in the registration and turnout rates of individuals discharged before and after the implementation of notification.

 

A number of caveats should be applied to our conclusion that notification has been unsuccessful at increasing participation. Although it is unlikely that we could observe these data if there were a large participatory effect of notification, there is a possibility that either sampling or specification error is masking a smaller effect. Our empirical strategy is also built on an assumption that any effect of notification on participation will occur quickly after it is implemented. There are a number of reasons why this may not occur. We cannot be sure that street-level criminal justice officials adhered to the date of implementation specified by their superiors. The success of notification may also depend on the context. For example, notification may be more effective when it occurs during a time of heightened political interest, such as the period right before an election or a well-publicized law change.

 

We also do not interpret these results as implying that notification cannot increase participation. It is possible that notification could be ineffective because ex-felons who wish to vote already invest in learning their rights. However, the findings in Meredith and Morse (2013) suggest that a substantial number of ex-felons can be mobilized by notification. An alternative explanation for the null results presented here is that a number of features of the notification protocols adopted by these states might limit their effectiveness. Given that these notification treatments are designed to correct misinformation about voting rights, the treatments in North Carolina and New York are striking for their lack of clarity. The voting rights information is buried in densely worded pamphlets. Presenting information in this manner may be particularly problematic given that a large percentage of the prison population does not read above a grade-school level (Kozol 1985). Further, these pamphlets are distributed in an exit packet that often contains a lot of other important documents, and this may cause information about voting rights to be crowded out.

 

Contrast these treatments with the letter that Meredith and Morse (2013) found increased ex-felon turnout by 5 to 10 percentage points in Iowa. First, the letter in Iowa, two paragraphs long and written in large font, focused solely on informing ex-felons about what rights were and were not being restored upon discharge. It was also personally addressed to the dischargee, making it clear that these voting rights applied to him or her. Moreover, the language used in the letter conveys not only that the ex-felon is eligible to vote but also that the state wants him or her to vote. Finally, the letter was sent via mail, and was thus likely to be read separately from all of the other documents distributed upon discharge.19

 

Finally, these findings highlight the importance of thinking about state disenfranchisement policy in broad terms. Much of the academic, legal, and policy literature on criminal disenfranchisement focuses on the length of time that individuals remain legally ineligible to vote. However, legal disenfranchisement is only one channel through which criminal disenfranchisement affects ex-felon turnout at the polls. A number of policies, such as notification, can affect the participation rates of the formerly disenfranchised. Because the flow of individuals out of the criminal justice system at any point in time is relatively small compared to all individuals who have exited the system across time, the total number of voters disenfranchised in practice is likely more affected by de facto disenfranchisement than by small distinctions in when the voting rights are restored. This means that, ultimately, parties interested in increasing ex-felon political participation may find it more valuable to use their resources to develop more effective protocols with which to inform ex-felons about their voting rights than fighting legislative battles to extend legal voting rights.

 

 

 

Previous Section

 

Next Section

 

Appendix

 

 

 

View this table: In this window

In a new window

Download to PowerPoint Slide

 

Table A1

Explanatory Variable Means by State

 

 

 

Figure A1

View larger version: In this page

In a new window

Download to PowerPoint Slide

 

Figure A1

Sample New Mexico Discharge Certificate

 

 

 

Figure A2

View larger version: In this page

In a new window

Download to PowerPoint Slide

 

Figure A2

New York Notification Literature

 

 

 

Figure A3

View larger version: In this page

In a new window

Download to PowerPoint Slide

 

Figure A3

More New York Notification Literature

 

 

 

Figure A4

View larger version: In this page

In a new window

Download to PowerPoint Slide

 

Figure A4

North Carolina Notification Brochure (Front)

 

 

 

Figure A5

View larger version: In this page

In a new window

Download to PowerPoint Slide

 

Figure A5

North Carolina Notification Brochure (Back)

 

 

 

Figure A6

View larger version: In this page

In a new window

Download to PowerPoint Slide

 

Figure A6

Registration and 2008 Turnout by Quarter of Discharge in New Mexico

 

 

 

Figure A7

View larger version: In this page

In a new window

Download to PowerPoint Slide

 

Figure A7

Registration and 2008 Turnout by Quarter of Discharge in North Carolina

 

 

 

 

Previous Section

 

Next Section

 

Notes

 

•NOTE: We thank Maria Silfa for her excellent research assistance and Alec Ewald and Lisa Miller for helpful comments and suggestions. We also thank Lonna Atkeson and Lisa Bryant for providing us with a New Mexico voter file and David Nickerson for providing us with a North Carolina voter file.

 

 

•↵1. The Brennan Center data reported to us by Erika Wood, Deputy Director of the Democracy Program, in an email dated March, 11, 2011.

 

 

•↵2. In addition, Florida and Virginia passed laws mandating that the state provide information to individuals about how they can restore their voting rights. Bills were also introduced in Alabama and New Jersey that included notification requirements as part of a package of reforms that expanded the voting rights of the criminally disenfranchised.

 

 

•↵3. In fact, Mitt Romney, the eventual Republican nominee, did run television ads criticizing Rick Santorum for voting in support of ex-felon voting rights in the U.S. Senate (Dinan 2012).

 

 

•↵4. Louisiana is not included in our analysis because criminal justice discharge records are not public record.

 

 

•↵5. For example, Burch (2011) defines a match as a voter file record that has the same last name, first name, and date of birth as a record in the discharge data.

 

 

•↵6. Individuals who do not go onto parole upon discharge from prison will also not appear in our data. Auxiliary data show between 10 and 20 percent of prison discharges do not go on parole.

 

 

•↵7. As a point of comparison, 49, 24, and 26 percent of all active registrants affiliate as Democrats, Republicans, and with no party or a minor party, respectively.

 

 

•↵8. As a robustness check, we also run the New York matching procedure initially screening on birth year for a random sample of 10 percent of the New Mexico discharges between 1/1/2005–9/30/2008; 2.7 percent of the observations in this sample did not match to an observation in the voter file using our original matching procedure, but did match to an observation in the voter file using the more inclusive name matching procedure. Further examination revealed that a majority of these matches appeared to be a discharge record matching to another person’s voter file record, suggesting that we would not want to use this matching procedure even in the absence of computational constraints.

 

 

•↵9. As a point of comparison, 51, 32, and 18 percent of all active registrants affiliate as Democrats, Republicans, and with no party or a minor party, respectively.

 

 

•↵10. Around one-third of dischargees between November 3, 2004, and September 30, 2008, recidivate.

 

 

•↵11. One concern about using a 2013 voter file to measure 2008 turnout is that a number of 2008 voting records may have been attenuated out of the file.

 

 

•↵12. For example, we do not know whether an individual who was discharged in October 2007, at age 23 would be 28 or 29 in April 2013.

 

 

•↵13. Of the observations in this sample, 6.3 percent did not match to an observation in the voter file using our original matching procedure, but did match to an observation in the voter file using the more inclusive name matching procedure. Like in New Mexico, further analysis suggested that most of these matches were likely false matches.

 

 

•↵14. The turnout rate is about 2 percentage points larger if we condition on not recidivating as we do in New Mexico and New York.

 

 

•↵15. Forty-three, 31, and 26 percent of active registrants affiliate as Democrats, Republicans, and with no party or a minor party, respectively.

 

 

•↵16. This is consistent with the finding of Meredith (2009) that eligibility to vote in previous elections increases the probability of being registered and voting in subsequent elections.

 

 

•↵17. An alternative approach would be to use a more traditional RDD specification that estimates separate polynomials for time since discharge before and after notification (Meredith and Morse 2013). This approach attempts to calculate the discontinuous change in political participation on the exact date that notification begins. Because we received slightly conflicting information about the exact date that the notification mandate began in some states and we cannot be sure that street-level criminal justice officials began implementing notification on this date, we decided not to use this approach.

 

 

•↵18. All effect sizes reported, holding all other variables at their sample means.

 

 

•↵19. Sending information via mail is not without its own downsides, as contact information for this population may quickly become out of date. Gerber et al. (2013) find that 39 percent of letters were returned in a field experiment that targeted ex-felons at their last known address.

 

 

 

 

Marc Meredith is an assistant professor in the Political Science Department at the University of Pennsylvania. His research examines the political economy of American elections, with a particular focus on the application of causal inference methods. His research has been published in the American Political Science Review, the Journal of Politics, the Proceedings of the National Academies of Science, Political Analysis, and the Quarterly Journal of Political Science, among other outlets.

 

Michael Morse is a research fellow at Stanford Law School. His undergraduate honors thesis, titled “Discretionary Disenfranchisement,” received the Frederick W. Meier Jr. Prize for best undergraduate senior thesis by the Delta chapter of Phi Beta Kappa.

 

 

 

Previous Section

 

 

 

References

↵ Allen Jessie

. 2011. Documentary disenfranchisement. Tulane Law Review 86:389–464. Google Scholar

↵ Burch Traci

. 2007. Punishment and participation: How criminal convictions threaten American democracy. PhD diss., Harvard University, Cambridge, MA. Google Scholar

↵ Burch Traci

. 2011. Turnout and party registration among criminal offenders in the 2008 general election. Law & Society Review 45 (3): 699–730. CrossRefWeb of ScienceGoogle Scholar

↵ Burch Traci

. 2012. Did disfranchisement laws help elect President Bush? New evidence on the turnout rates and candidate preferences of Florida’s ex-felons. Political Behavior 34 (1): 1–26. CrossRefWeb of ScienceGoogle Scholar

↵ Cooper Christopher A.,

Haspel Moshe,

Knotts H. Gibbs

. 2009. The value of voter files for U.S. state politics research. State Politics & Policy Quarterly 9 (1): 102–21. Web of ScienceGoogle Scholar

↵ Dinan Stephen

. 16 January 2012. Santorum defends support for restoring felons voting rights. Washington Times. Google Scholar

↵ Drucker Ernest,

Barreras Ricardo

. 2005. Studies of voting behavior and felony disenfranchisement among individuals in the criminal justice system in New York, Connecticut, and Ohio. Washington, DC: The Sentencing Project. Available from http://www.sentencingproject.org/doc/publications/fd_studiesvotingbehavior.pdf.

↵ Ewald Alec C.

2005. A “crazy quilt” of tiny pieces: State and local administration of American criminal disenfranchisement law. Washington, DC: The Sentencing Project. Available from http://www.sentencingproject.org/doc/publications/fd_crazyquilt.pdf.

↵ Ewald Alec C.

2009. Criminal disenfranchisement and the challenge of American federalism. Publius 39 (3): 527–56. Abstract/FREE Full Text

↵ Gerber Alan S.,

Huber Gregory A.,

Meredith Marc,

Biggers Daniel R.,

Hendry David

. 2013. Felony status, patterns of participation, and the possibility of political reintegration: Results from a field experiment. New Haven, CT: Yale University mimeo. Google Scholar

↵ Haselswerdt Michael V.

2009. Con job: An estimate of ex-felon voter turnout using document-based data. Social Science Quarterly 90 (2): 262–73. CrossRefGoogle Scholar

↵ Hasen Rick L.

2012. The voting wars: From Florida 2000 to the next election meltdown. New Haven, CT: Yale University Press. Google Scholar

↵ Haynie Kerry L.

2001. African American legislators in the American states. New York, NY: Columbia University Press. Google Scholar

↵ Hjalmarsson Randi,

Lopez Mark

. 2010. The voting behavior of young disenfranchised felons: Would they vote if they could? American Law and Economics Review 12 (2): 265–79. Abstract/FREE Full Text

↵ Kozol Jonathan

. 1985. Illiterate America. New York, NY: Anchor/Doubleday. Google Scholar

↵ Manza Jeff,

Uggen Christopher

. 2004. Punishment and democracy: Disenfranchisement of nonincarcerated felons in the United States. Perspectives on Politics 2 (3): 491–505. Google Scholar

↵ Manza Jeff,

Uggen Christopher

. 2006. Locked out: Felon disenfranchisement and American democracy. New York, NY: Oxford University Press. Google Scholar

↵ McMiller Darryl L.

2008. The campaign to restore the voting rights of people convicted of a felony and sentenced to probation in Connecticut. American Behavior Scientist 51 (5): 645–58. Abstract/FREE Full Text

↵ Meredith Marc

. 2009. Persistence in political participation. Quarterly Journal of Political Science 4 (3): 187–209. CrossRefWeb of ScienceGoogle Scholar

↵ Meredith Marc,

Morse Michael

. 2013. The politics of the restoration of ex-felon voting rights: The case of Iowa. Philadelphia, PA: University of Pennsylvania mimeo. Google Scholar

↵ Miles Thomas J.

2004. Felon disenfranchisement and voter turnout. Journal of Legal Studies 33 (1): 85–129. CrossRefWeb of ScienceGoogle Scholar

↵ Porter Nicole D.

2010. Expanding the vote: State felony disenfranchisement reform, 1997–2010. Washington, DC: The Sentencing Project. Available from http://www.sentencingproject.org/doc/publications/publications/vr_ExpandingtheVoteFinalAddendum.pdf.

↵ Rosenstone Steven,

Hansen John M.

1993. Mobilization, participation and democracy in America. New York, NY: MacMillan. Google Scholar

↵ Uggen Christopher,

Manza Jeff

. 2002. Democratic contraction? Political consequences of felon disenfranchisement in the United States. American Sociological Review 67 (6): 777–803. CrossRefWeb of ScienceGoogle Scholar

↵ Uggen Christopher,

Shannon Sarah,

Manza Jeff

. 2012. State-level estimates of felon disenfranchisement in the United States, 2010. Washington, DC: The Sentencing Project. Available from http://www.sentencingproject.org/doc/publications/publications/vd_State_Level_Estimates_of_Felon_Disen_2010.pdf.

 

 

Western Bruce,

Pattillo Mary,

Weiman David

. 2004. Introduction. In Imprisoning America: The social effects of mass incarceration, eds. Pattillo Mary, Weiman David, Western Bruce , 1–18. New York, NY: Russell Sage Foundation. Google Scholar

↵ Wood Erika,

Bloom Rachel

. 2008. De facto disenfranchisement. New York, NY : Brennan Center for Justice. Available from http://www.brennancenter.org/sites/default/files/legacy/publications/09.08.DeFacto.Disenfranchisement.pdf.

Link to comment
Share on other sites

7 In 10 Felons Register As Democrats

 

And why not?   The sociopath and psycopath repubs put them there,  of course they are going to be dems

IF they can vote.

 

Check it out,  felons are not even allowed to vote in 11 GOP States.  GOP builds prisons to put democrats in them, whether they deserve to be there or not,  then if you were unfortunate enough to be Felonized by the fknrepublicans in a GOP state  you cant even vote to change unjust and prejudiced laws.  Further evidence of GOP attack on Democracy.

 

The supreme court is majority repub,  the house is majority repub, the senate is majority repub, our entire state leadership is majority repub,  the popo union is repub......of course you are going to see a majority dems in jail thats the 1% repub plan!.  Once you figure out how the system is rigged the game is fixed and they stick you in the hole you have no choice but be a dem.

 

Your going to have to go to the Country Club Prisons to find the pampered GOP (guests), regular folk like us dems never see those accomodations.   If you are a GOP white collar Warwhore or Bankster  you get the Ritz  all the rest of us get Rikers

 

100% of inmates...Dont trust the System

 

http://felonvoting.procon.org/view.resource.php?resourceID=000286

Edited by beourbud
Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

Loading...
 Share

×
×
  • Create New...