Gennaro F. Vito
J. Eagle Shutt
Richard Tewksbury
Department of Justice Administration, University of Louisville
History of Felon Disenfranchisement
The Present Study
The Survey
The Sample
Data Collection Procedures
Data Analysis
Descriptive Statistics of the Sample
Estimating Disenfranchised Population
Comparing Disenfranchised Voter Preferences to Official Voter Preferences
Conclusion
FELON DISENFRANCHISEMENT, or the restriction of voting rights for convicted felons, is a staple of American criminal justice policy, practiced in one form or another in 48 of 50 American states. Nevertheless, the practice itself has become increasingly controversial in light of research suggesting disproportionate impacts on minorities and political parties. Disenfranchisement policy currently excludes one in six African- American males. For example, in the 1998 elections, at least 10 states formally disenfranchised 20 percent of African-American voters due to felony convictions (Journal of Blacks in Higher Education, 1999). Excluding felons provided “a small but clear advantage to Republican candidates in every presidential and senatorial election from 1972 to 2000” (Manza & Uggen, 2006, p. 191). In addition, felon disenfranchisement may have changed the course of history by costing Al Gore the 2000 presidential election (Uggen & Manza, 2002). Similarly, if not for felon disenfranchisement, Democratic senatorial candidates would likely have prevailed in Texas (1978), Kentucky (1984 and 1992), Florida (1988 and 2004), and Georgia (1992) (Manza & Uggen, 2006, p.194).
Since felon disenfranchisement affects the civil rights of nearly five million voters (over 2 percent of the eligible voters), critically evaluating its rationales remains a significant criminal justice policy issue (Manza & Uggen, 2004). Felon disenfranchisement proponents argue that disenfranchisement deters future crime, punishes offenders by excluding them from community political participation on moral grounds, and is necessary to ensure the proper functioning of democratic institutions. Critics, however, contend that disenfranchisement is actually criminogenic (Cholbi, 2002), derives from a historically based racial animus (Behrens, Uggen, & Manza, 2003), and harms democratic processes (Harvard Law Review, 1989; Uggen & Manza, 2002).
Nearly all American states (48) continue to exclude prisoners from voting. While disenfranchisement of the incarcerated has remained nearly ubiquitous, other forms of felon disenfranchisement have consistently decreased over time. The majority of states also prohibit active probationers and parolees from voting. Yet, state exclusion rates for these groups decreased substantially from 1950 to 2002 (from 84 percent to 58 percent and 84 percent to 70 percent for probationers and parolees, respectively) (Manza & Uggen, 2004, p. 493).
Direct estimations of probationer/parolee voting preferences have not been examined. Attempts to estimate such preferences have typically attempted to match disenfranchisees to voter preferences of similarly situated voters (e.g., Uggen & Manza, 2002) or interview/ survey prisoners within a particular location (Manza & Uggen, 2006). Such estimates introduced two serious measurement issues: First, prisoner preferences are not necessarily representative of probationer/parolee preferences. Second, prisoner preferences may not provide a representative sample for a particular state, since prisoners are concentrated in particular locales that are often outside their domiciles. No prior research has attempted to provide a representative sample of probationer/parolee preferences.
This study sought to address the deficiencies of prior voting preference estimations via a survey of voting preferences of a representative sample of probationers/parolees. Such a methodology would allow for more accurate estimates of the effects of probationer/parolee disenfranchisement.
History of Felon Disenfranchisement
As a practice, felon disenfranchisement has ancient roots. Both ancient Greece and Rome excluded offenders from voting and owning property (Harvard Law Review, 1989; Johnson-Parris, 2003; Manza & Uggen, 2004). These practices were carried over into Great Britain and then to the American colonies. Felon disenfranchisement expanded in the period following the Civil War (Behrens et al., 2003). Most states now allow felon reenfranchisement upon completion of probation/ parole supervision. Felon disenfranchisement policies reflect punitive American attitudes and court deference to legislatures to establish punishments. The continued existence of such policies is entirely consistent with America’s “get tough on crime” ethos. Felon disenfranchisement remains in place despite clear evidence that it dilutes the voting impact of minority voting populations.
The focus of our research is Kentucky, one of only three states that permanently disenfranchise all persons with felony convictions even after termination of criminal justice supervision. Disenfranchised individuals must apply to the Governor for an executive pardon for the restoration of voting rights. Kentucky has the sixth highest rate of disenfranchisement in the country. One out of every 17 Kentucky residents is disenfranchised, a rate more than twice the national average. As a result, nearly one of every four (23.7 percent) African Americans in Kentucky is ineligible to vote, a rate nearly triple the national African-American disenfranchisement rate of 8.25 percent (League of Women Voters of Kentucky, 2006).
Initially, the enfranchisement restoration process in Kentucky was simplified in 2001 and the number of restored applicants increased. The policy required the Department of Corrections to inform all eligible individuals of their right to apply and assist them with the application process. In 2004, Governor Ernie Fletcher required all applicants to submit a written essay on why the person wants and believes he/she should receive restoration of his or her voting rights and provide three character references. The governor also required the application to be presented to prosecutors in the jurisdiction where the person lives for a recommendation on whether voting rights should be restored. As a result, the restoration rate plummeted from 97.2 percent of the applicants in fiscal year 2002-03 to 28.1 percent in the time period of July 2005 to March 2006. Fletcher’s successor, Governor Steven Beshear, has eliminated the essay requirement altogether (League of Women Voters of Kentucky, 2006).
Uggen and Manza (2002, pp. 789-790) determined that Kentucky’s felon disenfranchisement policies helped elect Republican Senator Mitch McConnell in 1984. McConnell won by 5,269 votes, while the total number of disenfranchised Kentucky voters numbered about 75,000. They estimated that about 70 percent of these voters would have cast their ballot for the Democratic senatorial candidate. They also estimated that, without felon disenfranchisement, Democrats might have controlled the U.S. Senate through the 1990s. In 1998, Republican Jim Bunning defeated Democrat Scott Baesler by 7,000 votes. Kentucky’s felon disenfranchisement laws at that time permanently excluded felons from voting, eliminating 6,000 African-American prisoners and 7,600 African-American probationers and parolees, as well as thousands more permanently disenfranchised African-American citizens. In the election itself, the vast majority of African-American voters voted Democratic (Journal of Blacks in Higher Education, 1999). On this basis, both of Kentucky’s Republican senators may owe their previous elections to felon disenfranchisement.
The Present Study
This study sought to examine the effects of felon disenfranchisement on election results. Specifically, this study estimated voting preferences of disenfranchised voters via a survey administered to a stratified random sample of Kentucky’s probationers and parolees. Respondents were asked to provide their preferences for United States president and senator, and for political parties. The projected voting patterns of disenfranchised voter preferences were then combined with official voting tallies to estimate the effect of Kentucky’s felon disenfranchisement policy on national and state elections. In effect, this study responds to Manza and Uggen’s (2006, p. 189) call for a “real world exercise in which some or all of the disenfranchised population” votes to test their assumptions.
The Survey
The survey instrument consisted of a mock ballot (patterned on Kentucky’s official version) and brief demographic questions. The instrument itself was administered only after approval by Kentucky’s Department of Corrections and the University of Louisville’s Institutional Review Board.
The three dependent variables were: 1) presidential preference, nominally categorized as a) Barack Obama (Democrat), b) John McCain (Republican), c) Other (write in your choice), and d) Undecided; 2) senatorial preference, categorized as a) Bruce Lunsford (Democrat), b) Mitch McConnell (Republican), c) Sonny Landham (Libertarian), d) Other (write in your choice), and e) Undecided; and 3) party preference, categorized as a) Democrat, b) Republican, c) Libertarian, d) Independent, and e) Other (please specify).
Predictor variables included: 1) age (in years); 2) gender; 3) educational attainment, categorized as a) did not complete high school, b) completed high school, and c) completed college; and 4) race, in nominal categories of a) African-American, b) Caucasian, and c) Other.
The Sample
This study utilized a sampling methodology because, for several reasons, compiling a complete record of voter preferences of Kentucky’s probation and parole population was impractical. First, the probationer/parolee population was too large and geographically scattered to be fully canvassed within a short time period. According to directly obtained in-house figures, Kentucky’s Division of Probation and Parole managed 28,342 probationers and 10,252 parolees, as of January 3, 2009, for a total of 38,594 supervisees. Time constraints posed a practical hurdle as well: Since political opinion may exhibit considerable variation over time, preferences were recorded during the 30 days immediately preceding the November 4, 2008 election.
To account for potential regional differences, this study utilized a random stratified sampling design. Kentucky’s Division of Probation and Parole has partitioned the state’s 120 counties into 19 geographical districts, which are further subdivided into 57 offices. Kentucky’s population can be divided into thirds by county size: 1) counties with populations 13,885 or less; 2) 13,886 to 27,129; and 3) 27,130 and over. Each district was classified into 1) large urban, 2) midsize urban and 3) rural areas. We then chose a total of 11 county sites at random, ensuring proportional representation of large urban, mid-sized urban and rural communities. The final sample comes from three urban areas, three mid-size areas, and five rural areas.
Data Collection Procedures
The study’s research design attempted to account for respondent biases, which create threats to internal validity (Thye, 2000). Social desirability (Reynolds, 1982; Hays, Hayashi, and Stewart 1989) and ordering presentation (Dean 1980; Presser & Stinson, 1998; in voting contexts, see Rubinstein & Salant, 2006; cf. Alvarez, Sinclair, & Hasen, 2006) bias research suggest that: 1) respondents may vary their responses due to their desire to please; 2) certain topics, such as status, race, or sexuality, may be more likely to trigger biases; and 3) temporal, spatial, and verbal presentation of survey items may bias results. To avoid such potential biases, the voting process itself was anonymous and confidential. Other than signed informed consent on separate documents, no identifiers were kept. To simulate actual voting conditions, respondents were asked to complete their ballots in secret, fold them, and deposit them in a locked ballot box. Due to voting secrecy, researchers were unable to monitor respondent answers; consequently, some respondents failed to answer all questions, leading to missing data.
To control for presentation ordering biases, three separate ballots were administered, which varied in candidate presentation. For example, Democrat, Republican, and Libertarian senatorial candidates were the first options in Ballots A, B, and C, respectively.
Researchers arrived at respective sites on previously arranged reporting days, after receiving permission from office supervisors. In the waiting room, researchers approached individuals, and after confirming that potential respondents were on probation or parole for felonies, asked whether they would be willing to participate in an anonymous and confidential survey of voting preferences, wherein they would be asked to give their choices for United States President and Senate. Respondent participation exceeded 90 percent at every site, though subjects received no compensation.
Data Analysis
After data collection, the completed surveys (n = 425) were recorded in SPSS, and descriptive statistics were generated. Probationer/ parolee felon disenfranchisement effects were first estimated by projecting voter survey results to Kentucky’s statewide and countylevel probationer/parolee populations. The voting preferences of the felon disenfranchisees were compared to those of official voters. Using estimates of the total population of felon disenfranchisees, we examined whether similar disenfranchised voter preferences would have changed election outcomes.
Descriptive Statistics of the Sample
The typical disenfranchisee was a White male in his thirties with a high-school education or its equivalent. While the majority (66.4 percent) had completed high school or an equivalent degree, one-third had not completed high school.
Despite Kentucky’s Republican tendencies, disenfranchised voters clearly skewed Democratic. Respondents were more than twice as likely to favor the Democratic Party over the Republican Party (57 percent to 24.2 percent).
Estimating Disenfranchised Population
Kentucky’s actual disenfranchised population remains a dark figure. Clearly, felons in prison, probation, or parole cannot vote, and such numbers can be determined with considerable specificity. However, it is unclear precisely how many felon ex-supervisees are in the state, have had their voting rights restored, have died, and/or left the state. Kentucky maintains no official disenfranchisement statistics.
Manza and Uggen (2006) calculated a 5.9 percent disenfranchisement rate among Kentucky voting-age residents, as of December 31, 2004. Their estimate relied on conservative assumptions, designed to underestimate rather than inflate the actual population of disenfranchisees. Specifically, Manza and Uggen : 1) used highly reliable official statistics of criminal justice supervisees currently disqualified in Kentucky; 2) calculated empirically informed, adjusted annual prisoner/parolee reoffense and separate probationer/jail inmate reoffense rates that deliberately exceeded those found in most long-term recidivism studies; and 3) calculated mortality based on the median age of incarceration release of the most vulnerable demographic group, African- American males, adjusted by 1.46 to match high death rates found by a Justice Department recidivism study (Bureau of Justice
A key methodological issue is whether Manza and Uggen’s (2006) December 31, 2004 disenfranchisement rate has remained a useful, conservative approximation as of November, 2008. To assess the rate’s continued utility, the table was updated with official statistics.1
Comparing Disenfranchised Voter Preferences to Official Voter Preferences
Kentucky’s official voting results (Kentucky State Board of Elections, 2008) contrast sharply with disenfranchisee preferences. Though disenfranchised voters favored Democratic candidates by sizeable majorities, actual voters favored Republicans.2 Actual voters favored McCain over Obama 57.4 percent to 41.2 percent. The felon disenfranchisees in the sample voted for Obama (54.7 percent) over McCain (32.3 percent). Extended to the entire estimated sample of Kentucky felon disenfranchisees, this margin was even greater—Obama (61.5 percent) and McCain (36.3 percent). However, when these results are combined with those of registered voters, the final Kentucky outcome would remain the same, with McCain (56.3 percent) carrying the state over Obama (42.3 percent). If Kentucky felon disenfranchisees had been permitted to vote, they would not have altered the final result and John McCain would still have won the state.
In the official election, incumbent Republican U.S. Senator Mitch McConnell (53.0 percent) defeated his Democratic challenger Lunsford (47.0 percent). Again, the Republican candidate carried the statewide election. Here again, the sample of felon disenfranchisees voted Democratic, favoring Lunsford over McConnell, but by a vote of 41.8 percent to 28.6 percent. Projecting these results to the entire estimated Kentucky felon disenfranchisee population, we found Lunsford favored by a margin of 59.4 percent to 40.6 percent. Yet, these results would not have altered the official outcome of the 2008 Kentucky U.S. Senate election. In the final analysis, Senator McConnell would still be victorious over candidate Lunsford (52.3 percent to 47.7 percent).
Conclusion
The results of this Kentucky study clearly demonstrate that full participation by felon disenfranchisees would not have altered the outcome of the 2008 senatorial and presidential election in the state. Our analysis was based upon a survey of voting preferences of a sample of Kentucky probationers and parolees. Results show that there are small but real differences that disenfranchisement can make in elections, and as others have shown, some previous elections may have had different outcomes had disenfranchised individuals been able to vote. However, while real, the effects of currently disenfranchised persons voting results in only very minor changes to distribution of votes. Although felon disenfranchisees would likely favor Democratic candidates more than the Kentucky official voting results indicate, this difference had no effect upon the winners of the final voting tally. Republicans McCain and McConnell would still have carried the state in 2008. Clearly, political concerns about disenfranchisees having significant, sizable effects on election outcomes are unfounded. While very close elections could be swayed by the votes of currently disenfranchised persons, this is likely to occur in only very rare instances. Therefore, there is no legitimate reason to continue to deny the vote to this population.
Continuation of felon disenfranchisement sends a negative and deleterious message to citizens. It fails to recognize felons’ performance in completing their sentence, including a period of supervision in the community on probation or parole. Although sentences may be completed and individuals are free from correctional supervision, they remain disadvantaged and carry less than complete rights of citizenship. Disenfranchisement separates them from community life and continues the stigma of conviction into the future. Felon disenfranchisement is an “invisible punishment” and a barrier to rehabilitation (Manza & Uggen, 2006, p. 37). In fact, research results indicate that former felons who do have the right to vote have lower recidivism rates (Manza & Uggen, 2006 p. 133). What is there to fear from granting this right to felons who have completed supervision and their term of punishment? As this study shows, political fears are unfounded, the results of nearly all elections are unlikely to have been changed, and restoration of voting rights carries with it important, positive correlates for former felons.