Social Exclusion as a State-Level Predictor of Changes in Collateral Sanctions

Abstract

Collateral sanctions are civil penalties or disabilities imposed upon people who are arrested, charged, or convicted of a crime. Little research is available concerning state-level predictors of these policies in the United States. Current research suggests that racial threat and political conservatism are associated with harsher sanctions or more restrictions in the realms of employment, housing, social benefits, and other categories. Using state report cards from the Legal Action Center, this study builds on existing knowledge by testing the relationship between state-level variables consistent with a social exclusion framework and collateral sanctions policies while also testing the relationship between social exclusion and changes in these policies over time. Results indicate that higher levels of social exclusion, measured by affordable housing scarcity, public benefit usage, and state fiscal health, may play a role in the adoption of state collateral sanction policies over time. In contrast to previous research, results offer mixed evidence regarding the relationship between the racial makeup of the state and the adoption of collateral sanctions policies. Policy implications are discussed.

Keywords

smart decarceration collateral sanctions social exclusion prisoner reentry criminal justice policy

Collateral sanctions are civil penalties or disabilities imposed upon people who are arrested, charged, or convicted of a crime (American Bar Association, 2017; Boire, 2007; Legal Action Center [LAC], 2009, 2004; Radice, 2012). Articulated in civil rather than criminal code, and administered by federal, state, and local agencies outside of the criminal justice system, collateral sanctions are underaccounted for in discussions about criminal justice policy-making (Chin, 2002; Pinard, 2006). However, these civil sanctions impact some of the most crucial arenas for living a healthy and independent life. There is evidence that collateral sanctions including restrictions on accessing housing, employment, education, and driver’s licenses are correlated with increased recidivism among formerly incarcerated individuals and others with arrests, charges, and convictions on their records (Boire, 2007; Bushway & Sweeten, 2007; Evans & Porter, 2014; Henry & Jacobs, 2007; LAC, 2009, 2004; Radice, 2012; Roman & Travis, 2004; Whittle, 2016). Furthermore, the literature indicates that many of the resources restricted by collateral sanctions are in fact protective factors predictive of reduced recidivism (Bruce, Crowley, Jeffcote, & Coulston, 2014; Clifasefi, Malone, & Collins, 2013; Ellison, Fox, Gains, & Pollock, 2013; Nally, Lockwood, Ho, & Knutson, 2014; Somers, Rezansoff, Moniruzzaman, Palepu, & Patterson, 2013; Uggen, 2000).

It is difficult to estimate the number of people impacted by collateral sanctions, yet it is known that around 700,000 U.S. citizens return from prison each year and nearly 24 million others cycle through local jails (Carson, 2015; Pettus-Davis, 2012). This indicates that a large portion of the U.S. population is impacted by a variety of collateral sanctions each year and that the number of people facing these sanctions continues to grow. In the meantime, recidivism among the formerly incarcerated is unprecedented. A Bureau of Justice Statistics (BJS) study found that 77% of prisoners exiting state prisons in 30 states during 2005 had been reincarcerated within 5 years of their release. Nearly 43% had recidivated in the first year after release, 60% by the end of the second year, and 68% by the end of the third year (Durose, Cooper, & Snyder, 2014). While the literature suggests that collateral sanctions contribute directly and indirectly—via reduced access to protective factors—to rearrest, it is not clear to what extent such sanctions are driving high national recidivism rates.

Likewise, we know little about what motivates the adoption and maintenance of collateral sanction policies in the face of research suggesting that they are counteractive as a form of deterrence. Using report cards from the LAC, Whittle and Parker (2014) found that indicators of political conservatism, religiosity, and racial threat are predictive of more restrictive collateral sanctions at the state level; however, little other research contributes meaningfully to understanding the determinants of state-level decisions about collateral sanctions. This study advances prior research by exploring new variables related to a social exclusion framework and by attending to changes in LAC collateral sanctions scores over time.

Background and Prior Research

The application of collateral sanctions following conviction of a criminal offense dates to ancient Greece and Rome, medieval Europe, and subsequently, the American colonies (Grady, 2012). While colonial sanctions included forfeiture of property and the right to vote, the advent of new public rights and services such as drivers’ licenses, public assistance, and criminal record keeping created new applications for collateral sanctions. Olivares, Burton, and Cullen (1996) provide evidence of increased use of collateral sanctions by the states between 1986 and 1996, pointing to the influence of tough-on-crime policies emphasizing deterrence and incapacitation during this time. In addition, the federal adoption of Megan’s Law in 1994 prompted a wave of sex offender registration laws, another type of collateral sanction (Olivares, Burton, & Cullen, 1996).

By 2003, collateral consequences were more broad and varied than at any prior point in U.S. history, and due to rising incarceration rates, these consequences affected a larger number of people than ever before (Petersilia, 2003). Report cards prepared by the LAC (2004, 2009), a nonprofit law firm dedicated to fighting discrimination against persons with criminal histories, provide scores for each state indicating the prevalence of collateral sanctions across multiple arenas, including adoption rights, voting rights, access to criminal records, student loans, drivers’ licensing, employment, public housing, and public assistance. LAC report cards from 2004 and 2009 show that while individual states have enacted changes to the number and type of collateral sanctions affecting people with criminal histories, on average, state-level collateral sanctions have not significantly changed since 2004. Since 2009, consistently high incarceration rates and a growing number of releases from prisons and jails have heightened the importance of explorations into the determinants and impacts of collateral sanctions for the previously incarcerated.

Sohoni (2013) and Luca (2011) explored the relationship between state collateral sanction laws and state recidivism rates. Using the LAC 2009 state-level report cards, Sohoni (2013) found that while housing and firearm restrictions were associated with lower recidivism, sanctions including access to criminal records, restrictions on employment, and restrictions to public assistance were associated with higher recidivism rates. Sohoni (2013) highlights the importance of controlling for release rates, which were linked in her study to both recidivism and policy severity. Also controlling for release rates, Luca (2011) found that when access to criminal records became publicly available online in a given state, recidivism went up, while first time arrest rates went down. While inconclusive, this research suggests that collateral sanctions may have some of their intended effect of deterring crime while also posing harm to formerly incarcerated individuals attempting to successfully reintegrate into society (Luca, 2011; Sohoni, 2013; Whittle, 2016).

Whittle (2016) highlights the challenges of measuring the effects of collateral sanctions on recidivism rates due to the lack of data on how many persons are actually affected by many collateral sanctions. For instance, it is impossible to know how many individuals have not sought or been denied access to public housing as a result of prior convictions. For this reason, research linking access to the rights and services blocked by collateral sanctions to recidivism rates provides an illuminative window into the possible consequences of large-scale sanctions. Holtfreter, Reisig, and Morash (2004) found that women with felony convictions had 83% lower odds of recidivism when they received public financial assistance. Women who utilized housing assistance reoffended at less than one third the rate of those who did not receive such assistance. In a U.K.-based study, Bruce, Crowley, Jeffcote, and Coulston (2014) found that exoffenders who received public housing assistance recidivated at a rate of 5% while those who did not recidivated at 51%.

In light of growing evidence that collateral sanctions are likely to have unintended consequences, we ask why the use of collateral sanctions by states remains relatively stable. Research on the state-level predictors of collateral sanctions is sparse. Burton, Cullen, and Travis (1987), and later Olivares et al. (1996), conducted national studies of state legal codes related to collateral consequences, showing that sanctions increased between 1986 and 1996 and that they were more restrictive in southern states across points in time. Stopping short of empirical analysis, Olivares et al. (1996) suggested the centrality of tough-on-crime ideology in influencing 10-year trends in collateral sanctions and proposed areas for future exploration, including the influence of crime rates, prison construction, population growth, and state budget reductions.

Whittle and Parker (2014) provides the most comprehensive empirical analysis in their study of state-level predictors of collateral sanctions using LAC scores from 2009. They hypothesized that racial threat and conservatism would correlate with higher LAC collateral sanction scores (harsher sanctions). Racial threat was measured by the percentage of residents identifying as Black in a state; conservatism, by whether a state was won by the Republican or Democratic candidate for president in 2008. Controlling for the significant effects of violent crime rates and the percentage of the state population identifying as male, the authors’ findings confirmed their hypotheses. However, they did find that there was a curvilinear relationship between the percentage of Black residents in a state and overall collateral sanctions scores. As such this study incorporated the squared percentage of residents who identified as Black, whether a state voted for the Republican or Democratic presidential candidate, the percentage of male residents, and violent crime rates. Higher violent crime rates, greater percentages of residents identifying as Black, higher percentages of males, and a state voting for a Republican candidate in presidential elections are thus expected to remain associated with harsher collateral sanctions laws. This study will account for Whittle and Parker’s (2014) findings and extend the model of civil penalty policy-making to account for variables related to a social exclusion framework.

The Role of Social Exclusion in the Adoption of Collateral Sanction Policies

Theories of social exclusion posit that some groups of people are excluded or marginalized from full participation in society, particularly economic participation, by systematic blocking of access to the rights, opportunities, and resources made available to other groups of people (Morazes & Pintak, 2007). These rights, opportunities, and resources are deemed essential for those experiencing exclusion to be full participants in society (Davis & Sanchez-Martinez, 2015; Morazes & Pintak, 2007). From this perspective, collateral sanctions can be seen as part of the myriad of barriers preventing marginalized groups of people from full social integration. These barriers may be exacerbated by scarcity of resources and a desire to allocate them to the most “deserving” citizens. We retrieved state-level information from 2008 for variables that are indicative of social exclusion—including affordable housing scarcity, unemployment, and strain on government benefits such as social security, cash assistance, and food stamps—to test the association between state levels of social exclusion and LAC scores in 2009 while accounting for the previously supported role of minority threat and political ideology (Whittle & Parker, 2014). We follow-up this analysis to see if state-level social exclusion variables from 2005 account for changes in LAC collateral sanction scores from 2004 to 2009.

Affordable Housing Scarcity

We designed a proxy to capture the availability of affordable housing for the poorest residents of the state. This measure accounts for the percentage of households earning under US$20,000 annually who are experiencing “housing cost burden,” otherwise known as using more than 30% of their household income on housing costs such as rent and utilities (Department of Housing and Urban Development, 2016). We refer to these households as housing cost-burdened low-income households. States with higher rates of cost-burdened low-income households may be more likely to exclude the least deserving from access to scarce housing resources which limits the ability of residents to participate fully in society. We hypothesize that, due to scarcity, (1) higher rates of cost-burdened low-income households in 2008 are associated with higher collateral sanctions scores in 2009 and (2) higher rates of cost-burdened, low-income households in 2005 are associated with increased (more restrictive) LAC scores from 2004 to 2009.

Unemployment Rates

Productive work is an essential factor for full participation in society (Davis & Sanchez-Martinez, 2015). State unemployment rates are one indicator of the availability of jobs for those who wish to work. High unemployment may motivate states to limit workforce participation by those viewed as least deserving; therefore, we hypothesize that (1) higher rates of unemployment in 2008 are predictive of higher LAC scores in 2009 and (2) higher unemployment rates in 2005 are predictive of increased collateral sanctions scores from 2004 to 2009.

Strain on Benefits

The percentage of people accessing financial benefits such as social security, cash assistance, or food stamps in each state is an indicator of the prevalence of residents having trouble meeting basic needs. We view high service utilization as another indicator of scarcity in the state environment, likely to lead to exclusionary policy-making. Accordingly, we hypothesize that (1) increased utilization of benefits in 2008 will be associated with higher LAC scores in 2009 and (2) higher rates of benefit usage in 2005 will be associated with increases in collateral sanctions from 2004 to 2009.

State Fiscal Health

States that have limited resources of their own may be unable or unwilling to meet the needs of their most impoverished residents. We use the ratio of federal taxes paid to federal spending received as an indicator of state fiscal health. A ratio above 1 indicates that states give more money in taxes to the federal government than they receive in federal spending, while ratio below 1 indicates the opposite. This measure approximates whether a state brings in enough revenue to provide its citizens with financial benefits and other essential services as opposed to relying on the federal government to make these services possible. States that are less reliant on the federal government for the money to provide benefits and services are in better fiscal health and are likely to be able to provide benefits and services to more people. A dearth of research suggests that poor fiscal health promotes policy-making based on resource withholding or regeneration (Nice, 1994). From a social exclusion framework, we view collateral sanctions as policies that restrict access to scarce resources for the least deserving. Therefore, we hypothesize that (1) states with ratios above 1 in 2005 will have lower LAC scores in 2009 and (2) states with ratios above 1 in 2005 will be associated with decreased (more lenient laws) LAC collateral sanctions scores from 2004 to 2009.

Again, as previous research has already indicated that conservatism, minority presence, percentage of male population, violent crime rates, and prison releases are significantly associated with more severe collateral sanctions, these variables were also incorporated into the analysis and higher rates of each of these are hypothesized to be associated with higher LAC scores in 2009 and increases in LAC collateral sanctions scores from 2004 to 2009.

Research Question and Hypotheses

Prior research on the adoption of collateral sanction policies is limited by the narrow range of variables explored as well as a “point-in-time” focus which fails to capture the temporal relationship between state characteristics and policy outcomes. This study explores the relationship between state-level factors indicative of social exclusion (along with variables previously found to have a significant relationship with collateral sanctions policies) and LAC scores as well as the relationship between those factors and changes in LAC collateral sanctions scores over the period of 2004–2009. We hypothesize that when taking into account the percentage of residents identifying as Black, percentage of the state population that is male, violent crime rates, rates of release, and political conservatism, the following state-level variables will be associated with higher LAC scores in 2009 and increases in LAC collateral sanction scores in from 2004 to 2009:

higher unemployment rates;

higher rates of housing cost-burdened low-income households;

a higher percentage of the populace accessing public financial benefits; and

a negative state financial status.

Method

This study takes an exploratory approach using observational state-level data to assess the relationship between (1) state-level variables consistent with a social exclusion framework from 2008 and LAC collateral sanctions scores from 2009 and (2) the same state-level variables from 2005 with changes in LAC collateral sanctions scores from 2004 to 2009. To answer the first research question, we emulate Whittle and Parker’s 2014 study—with the addition of social exclusion variables—using two linear regression models to assess the relationship between 2008 state-level characteristics and 2009 LAC scores. To answer the second research question, two similar linear regression models are used to assess the relationship between social exclusion and changes in LAC collateral sanctions scores using state-level data from 2005, except in the case of the results of the 2004 United States presidential election, which was used to determine whether a state leaned conservative or liberal at the time. State-level demographic and financial data from 2005 are incorporated as this was the first year of data collection for the 2009 collateral sanctions scores and thus, the first opportunity for these variables to theoretically impact any changes in those scores.

Data and Measures

Dependent variables

The collateral sanctions scores were made available by the LAC who compiled state grades (scores) in seven categories of sanctions in both 2004 and 2009. These categories are as follows: employment, public assistance, access to records, voting, public housing, parenting, and driver’s licenses (LAC, 2009, 2004). In each category, 0 is the lowest possible score and 10 is the highest, making the highest possible state total 70 points. The scores for each category are summed to produce a cumulative score for each state. Higher scores are indicative of more restrictive collateral sanctions in a state. LAC (2009) provides an in-depth account of the methodology employed in developing state-level scores, and Whittle and Parker (2014) offer a concise summary.

The LAC collateral sanction score for 2009 is the dependent variable for the first part of this study. Scores ranged from 6.5 to 46 (M = 28.47, SD = 9.54). Illinois has the lowest score indicating the least restrictive policies, and Alaska has the highest indicating that their policies are the most restrictive to those with felonies.

The numeric value of the actual difference in LAC scores from 2004 to 2009 served as the dependent variable in the second part of this study. In this case, we subtracted 2009 scores from 2004 scores in order for positive scores to be associated with improved (less harsh) collateral sanctions policy. Conversely, negative differences indicate that a state’s collateral sanction grade became worse or indicative of harsher collateral sanctions. The difference in total state LAC scores from 2004 to 2009 ranged from −11 to 16 (M = .96, SD = 5.65), with New Mexico adopting the most restrictive changes and Illinois adopting the most liberating changes. Table 1 shows the five states that improved their scores the most and the five states that had the most negative changes in scores from 2004 to 2009.

Table 1.

Five Most Positive and Negative Changes in Legal Action Center Scores 2004–2008.

State	Five Most Positive Changes	Five Most Negative Changes
Illinois	16
Maryland	13
Colorado	11
New Jersey	10
Connecticut	9.5
New Mexico		−11
Alaska		−10
Rhode Island		−9
Tennessee		−9
Maine		−7

Explanatory variables

Data for the unemployment rates, percentage of cost-burdened, low-income households, percentage of the population that identified as Black, percentage of the population identifying as male, and percentage of households accessing public financial benefits are drawn from the American Community Survey (ACS), which is publicly available through U.S. Census Bureau’s American Fact Finder Database (2016). These are continuous variables measured in percentages. However, due to previous research finding, a curvilinear relationship between the percentage of the population that identifies as Black and LAC collateral sanctions scores, we added the squared value of the percentage of Black residents in a second model in both analyses. For the first analysis, assessing the relationship between social exclusion and the 2009 LAC collateral sanctions scores, we used 2008 ACS 3-year estimates for each variable. For the second analysis, assessing the relationship between social exclusion and changes in LAC collateral sanctions scores from 2004 to 2009, we used ACS data from 2005, as 2005 was the year immediately following the release of the first LAC scores and thus becomes the first year that changes in the state environment were theoretically likely to impact the state’s next (2009) LAC score.

The ACS variable, “household gross rent as a percentage of household income in the last 12 months,” was used to calculate the percentage of housing cost-burdened low-income households. We combined the total number of households earning less than US$10,000 and between US$10,000 and US$19,999 annually that paid more than 30% of their income in gross rent/utilities and divided that number by the total amount of the population earning less than US$20,000 annually to derive this figure. The variable from the ACS titled, “public assistance income or food stamps in the past 12 months for households” was used to calculate the percentage of households accessing public financial benefits by dividing the households that utilized these benefits by the total population of the state. The ACS variable, “age and sex,” was used to calculate the percentage of male residents by dividing the total number of males by the total population of each state. The percentage of the population identifying as Black was calculated using the ACS variable, “Race.” We only used estimates for those who identified as Black or African American Alone and divided the number of these residents by the total population of the state. As previously mentioned, this percentage was then squared to develop another variable to represent the possibility of a curvilinear relationship between the percentage of Black residents and LAC collateral sanctions scores in the second model of both analyses (Whittle & Parker, 2014). Unemployment rates were taken directly from data available in the ACS variable, “employment status.” These variables are all available as both 2005 estimates and the 2008 3-year estimates that were used in this study.

Violent crime rates were calculated using 2008 Unified Crime Report (UCR) data from the Federal Bureau of Investigation (FBI) for the first part of the analysis assessing the relationship of social exclusion with 2009 LAC collateral sanctions scores. UCR data from 2005 were used for the second analysis that assesses the relationship between social exclusion and changes in LAC scores from 2004 to 2009. These variables were measured continuously and made up of the number of violent crimes per 100,000 people committed in each state in the years 2008 and 2005. The FBI (2016) makes these data publicly available each year.

Data for released prisoners came from the BJS’ report on prisoners from 2009, which provided information on releases over the course of 2000–2009 (West, Sabol, & Greenman, 2010). In the first analysis assessing the relationship between social exclusion and 2009 LAC collateral sanctions scores, average prison release percentages were categorized as having gone up, down, or having no change from 2000 to 2008. Increases in annual average release percentage were coded as 1, while no change and decreases in annual average releases were coded as 0 for this dichotomous variable. In the second analysis assessing the relationship between social exclusion and changes in LAC collateral sanctions scores from 2004 to 2009, prison releases were categorized as having gone up, down, or having no change from 2004 to 2005. Increases in releases were coded as 1, while no change and decreased releases were coded as 0 for this dichotomous variable.

Data regarding the voting record for each state in the 2008 and 2004 presidential elections were used to assess the political leaning of a state in both 2008 and 2005. These data came from the U.S. House of Representatives (2016): history, art and achieves website. The political leaning of a state was based on whether a state voted Republican or Democrat in both the 2008 and 2004 presidential elections. Democratic (liberal) states were coded as 1 and Republican (conservative) states were coded as 0 in both analyses for this dichotomous variable.

Data regarding the financial health of a state came from the Tax Foundation (TF), an independent tax policy research organization in operation since 1937 (TF, 2016). They provide detailed information regarding each state’s federal taxes paid versus the federal spending received by that state for the years 1981–2005 (TF, 2007). The 2005 data were accessed for both analyses in this study. The status for each state was determined by the ratio of federal tax paid to the amount of federal spending a state received during the year. Ratios above 1 indicated that a state was giving more money than they were receiving to the federal government, while ratios below 1 indicate that a state received more money than they gave. States with ratios above 1 were considered financially healthy and coded as 1 (giver to fed), while states with ratios below 1 were considered financially unhealthy and coded as 0 (taker from fed). See Table 2 for a concise description of all variables.

Table 2.

Variables: Levels of Measurement and Sources.

Variable	Level of Measurement	Type	Source
2009 LAC scores	Continuous	Dependent (Analysis 1)	LAC (2009)
Change in LAC score	Continuous	Dependent (Analysis 2)	LAC (2004, 2009)
% Male	Continuous	Explanatory	ACS (USCB, 2016)
Unemployment rate	Continuous	Explanatory (social exclusion)	ACS (USCB, 2016)
Percentage of cost-burdened low-income households	Continuous	Explanatory (social exclusion)	ACS (USCB, 2016)
% Black	Continuous	Explanatory	ACS (USCB, 2016)
Percentage of Black²	Continuous	Explanatory	ACS (USCB, 2016)
Percentage of accessing public financial benefits	Continuous	Explanatory (social exclusion)	ACS (USCB, 2016)
Violent crime rate	Continuous	Explanatory	FBI (2016)
Releases up or down	Dichotomous	Explanatory	BJS (West, Sabol, & Greenman, 2010)
Republican/Democrat	Dichotomous	Explanatory	USHOR (2016)
Gives more to fed	Dichotomous	Explanatory (social exclusion)	TF (2007)

Note. ACS = American Community Survey; BJS = Bureau of Justice Statistics; LAC = Legal Action Center.

Participants

The participants in this study were all 50 of the United States.

Analyses

Both analyses were run using STATA 13. For the first part of the study which tests the relationship between social exclusion and LAC collateral sanctions scores from 2009, two multivariate regression models are used. The first model incorporates all the explanatory variables discussed earlier, except for the squared percentage of the population identifying as Black, regressed on LAC collateral sanction scores for 2009. The second model adds the squared percentage of the population identifying as Black to account for the possibility of a nonlinear relationship between a state’s Black population and punitive policies found in some previous research (Whittle & Parker, 2014).

The assumptions for regression were tested and met for the first model. All the explanatory variables showed a linear relationship with the outcome variable according to scatterplots. The observed pattern of the data made possible with the lowess command in STATA suggested a linear relationship between the explanatory variables and the outcome variable. Due to evidence of heteroskedasticity, we used robust standard errors in our models (Stock & Watson, 2003). Multicollinearity is often an issue in regression models, so we tested for its presence using variance inflation factors (VIF). Typically, VIF above 10 indicates the presence of multicollinearity (Kutner, Nachtsheim, & Neter, 2004); however, Whittle and Parker (2014) point out that in most empirical studies anything above 4 is considered problematic. In this model, VIF ranged from 1.29 to 3.36 with a mean of 2.37 indicating a lack of multicollinearity. The Shapiro–Wilk test of normality indicated that the residuals were normally distributed, as we failed to reject the null hypothesis that they are distributed normally (p = .19). No evidence of omitted variables was found, as we failed to reject the null hypothesis that the model does not have omitted variable bias (p = .96). There was evidence of multicollinearity between the percentage of the population identifying as Black and its square in the second model but this is to be expected.

For the second part of the study which tests the relationship between social exclusion and changes in LAC collateral sanctions scores from 2004 to 2009, two multivariate linear regression models are used. The actual numerical difference in each state’s LAC score is the outcome variable. The explanatory variables are all the variables discussed in the data and measures section with the squared percentage of the Black population added in the second model.

The critical assumptions for multivariate regression were tested and met for the first model. The explanatory variables have a linear relationship with the dependent variable according to scatterplots and the observed pattern of the data. Although there was evidence of heteroskedasticity in our graphing of the variance of the residuals, we again follow Stock and Watson (2003) and use robust standard errors to account for its presence. To assess for evidence of multicollinearity, we checked the VIF in our model. The VIF in this multivariate regression model ranged from 1.40 to 4.06 with a mean of 2.67 suggesting a lack of multicollinearity. The Shapiro–Wilk test of normality indicated the residuals were normally distributed (p = .19), and we found no evidence of omitted variables (p = .74). The square of the percentage of the population identifying as Black and the unaltered percentage showed signs of multicollinearity in the second model, but again, this is to be expected.

Because of the small sample size in both analyses, which increases the likelihood of Type II error, and because of the exploratory nature of this study, statistical significance was measured using an α level of .10 (Labovitz, 1968). There is a precedent in a variety of state-level research for using an α level of .10, enabling researchers to capture changes in an inherently small sample of 50 states (see Hoefer, Black, & Salehin, 2012; Sliva, in press; Whittle & Parker, 2014, for instances). The limitation of this approach is an increased likelihood of Type I error.

Results

Descriptive Statistics

A detailed account of the descriptive statistics for the variables used in each analysis are available in Table 3.

Table 3.

Descriptive Statistics.

Variable	Mean	SD
Analysis 1
2009 LAC Score	28.47	9.54
Percentage of Black²	190.17	306.98
Percentage of Black	10.08	9.50
Percentage of Male	49.36	0.78
Democrat or Republican	0.60	0.49
Violent crime rate	399.12	172.72
Average release rate up	0.82	0.39
Gives more to fed	0.34	0.48
Percentage of public financial benefits	9.05	2.78
Unemployment rate	5.96	1.24
Percentage of cost-burdened low-income households	74.37	4.24
Analysis 2
Difference in LAC score	0.96	5.65
Percentage of Male	49.01	0.63
Percentage of Black²	185.66	302.32
Percentage of Black	9.86	9.50
Releases up	0.86	0.35
Democrat or Republican	0.38	0.49
Violent crime rate	137.00	77.11
Gives more to fed	0.34	0.48
Percentage of public financial benefits	18.78	5.49
Unemployment rate	6.48	1.21
Percentage of Cost-burdened low-income households	73.77	5.01

Note. LAC = Legal Action Center.

Multivariate Results

Analysis 1: The association between social exclusion and 2009 LAC collateral sanction scores

The results from the first regression model, F(9, 40) = 5.97, p < .001, R ² = .39, indicate that much like in Whittle and Parker (2014) increased percentages of Black population and male population in 2008 are statistically significantly associated with higher or more severe 2009 LAC collateral sanctions scores (see Table 4). We also found some evidence supporting the theory that social exclusion plays a part in collateral sanctions policy-making. For each 1% increase in housing cost-burdened low-income households in 2008, 2009 LAC collateral sanction scores increased by 0.94 points. Also, states that give more money to the federal government in taxes than they receive in federal spending’s 2009 collateral sanction scores are 9.72 points lower. One finding goes against our hypothesized relationship between social exclusion and 2009 LAC collateral sanctions scores. For each 1% increase in the percentage of households accessing public financial benefits in 2008, 2009 LAC scores drop 1.56 points (see Table 4).

Table 4.

Results of OLS Regression for 2009 Score.

Variable	Model 1		Model 2
Variable	Coefficient (Standard Error)	90% Confidence Interval	Coefficient (Standard Error)	90% Confidence Interval
Percentage of male	4.48 (2.66)*	[0.01, 8.96]	4.30 (2.97)	[−0.71, 9.31]
Percentage of Black	0.34 (0.17)**	[0.06, 0.63]	0.21 (0.66)	[−0.87, 1.30]
Percentage of Black²			0.004 (0.02)	[−0.02, 0.03]
Democrat or Republican	−1.01 (3.05)	[−6.15, 4.12]	−1.01 (3.11)	[−6.24, 4.22]
Violent crime rate	0.01 (0.01)	[−0.001, 0.02]	0.01 (0.01)	[−0.004, 0.03]
Average releases up	−3.27 (3.66)	[−9.43, 2.88]	−3.35 (3.69)	[−9.57, 2.86]
Gives more to fed	−9.72 (3.34)***	[−15.35, −4.09]	−9.63(3.44)***	[−15.41, −3.84
Percentage of public financial benefits	−1.56 (0.69)**	[−2.73, −0.40]	−1.62 (0.78)**	[−2.93, −0.30]
Unemployment	0.80 (1.68)	[−2.03, −3.63]	0.87 (1.74)	[−2.06, 3.81]
Percentage of cost-burdened low-income households	0.94 (0.49)*	0.12, 1.76]	0.95 (0.50)*	[0.19, 1.79]
Constant	−207.78 (135.98)	[−436.74, 21.19]	−198.80 (151.15)	[−453.46, 55.86]
R ²	.39		.39
Change in R ² (p value)	.0007 (p = .82)

Note. 2009 LAC score = outcome; LAC = Legal Action Center.

*p < .10. **p < .05. ***p < .01.

The second model does not provide evidence of the nonlinear relationship between a state’s Black population and collateral sanctions scores found in some previous research, nor does it significantly increase the amount of variance explained, F(10, 39) = 5.47, p < .001, R ² = .39. However, variables based on a social exclusion framework continue their significant relationship with LAC scores. Financially, healthy states scores are 9.62 points lower, a 1% increase in housing cost-burdened low-income households and results in an increase of 0.95 points, and a 1% increase in households accessing public financial benefits induces a decrease of 1.62 points (see Table 4).

Analysis 2: The association between social exclusion and changes in LAC collateral sanctions scores from 2004–2009

The first multivariate regression model, F(9, 40) = 3.71, p = .002, R ² = .39, revealed that the percentage of residents identifying as Black in a state in 2005 and being a financially healthy state in 2005 are both statistically significantly associated with lower LAC collateral sanction scores from 2004 to 2009. Financially healthy states decreased their 2009 LAC scores by an average of 6.02 more points than states financially dependent on federal funding. Conversely, increased percentages of housing cost-burdened low-income households and increased percentages of households accessing public financial benefits were statistically significantly associated with increased LAC scores from 2004 to 2009. A 1% increase in the percentage of housing cost-burdened low-income households is associated with a 0.29-point increase in a state’s LAC score. A 1% increase in the percentage of households accessing public financial benefits was associated with an increase of 0.43 points in a state’s collateral sanction score. None of the other explanatory variables had a statistically significant relationship with the difference in LAC collateral sanction scores from 2004 to 2009. We should note that increases in the Black population are associated with changes toward more lenient LAC scores from 2004 to 2009, which is not consistent with our first analysis, our hypothesis regarding this variable, or previous research. Similarly, the percentage of households accessing public financial benefits are associated with an increased or more restrictive score from 2004 to 2009, which is consistent with our hypothesis but not consistent with the relationship found in our first analysis (see Tables 4 and 5 for full results).

Table 5.

Results of OLS Regression for Difference in Score.

	Model 1		Model 2
Variable	Coefficient (Standard Error)	90% Confidence Interval	Coefficient (Standard Error)	90% Confidence Interval
Percentage of male	−1.13 (1.91)	[−4.35, 2.08]	−1.47 (2.71)	[−6.04, 3.09]
Percentage of Black	0.24 (0.13)*	[0.03, 0.46]	0.15 (0.48)	[−0.65, 0.95]
Percentage of Black²			0.003 (0.01)	[−0.02, 0.02]
Releases up	2.07 (2.59)	[−2.29, 6.43]	2.09 (2.63)	[−2.34, 6.52]
Democrat or Republican	−1.83 (2.13)	[−5.42, 1.76]	−2.10 (2.58)	[−6.45, 2.52]
Violent crime rate	−0.01 (0.01)	[−0.03, 0.01]	−0.01 (0.01)	[−0.03, 0.01]
Gives more to fed	6.02 (2.08)***	[2.53, 9.53]	6.21 (2.11)***	[2.65, 9.76]
Percentage of public financial benefits	−0.43 (0.22)*	[−0.80, −0.06]	−0.45 (0.23)*	[−0.86, −0.05]
Unemployment	−0.32 (0.97)	[−1.95, 1.31]	−0.27 (0.96)	[−1.89, 1.35]
Percentage of cost-burdened low-income households	−0.29 (0.15)*	[−0.53, −0.04]	−0.27 (0.17)#	[−0.55, 0.00]
Constant	84.15 (92.33)	[−71.32, 239.62]	100.57 (130.42)	[−119.18, 320.32]
R ²	.39		.39
Change in R ² (p value)	.001 (p = .83)

Note. LAC = Legal Action Center ; Difference in LAC score = outcome.

#p = .10. *p < .10. **p < .05. ***p < .01.

Again, the second model did not show support for a nonlinear relationship between the percentage of residents identifying as Black and punitive policy nor did the variance explained significantly increase, F(10, 39) = 3.36, p = .003, R ² = .39. However, just as in the first analysis, social exclusion variables remain significantly associated with changes in LAC collateral sanctions scores. Financially healthy states reduced their scores by 6.21 points, while a 1% increase in housing cost-burdened low-income households and results in an increase of 0.27 points, and a 1% increase in households accessing public financial benefits results in an increase of 0.45 points (see Table 5).

Discussion

The findings of this study provide some preliminary evidence that indicators of social exclusion help predict state decisions about collateral sanctions. States demonstrating indicators of affordable housing scarcity and high social benefit usage are more likely to institute restrictive sanctions across time, while financially healthy states are more likely to have lower LAC scores and remove sanctions across time. It is important to note that restricted access to public housing and public assistance are specific categories of collateral sanctions identified by the LAC (2009, 2004). The likelihood that affordable housing and social welfare policies are in part motivated by scarcity and social exclusion prompts further research and policy analysis about the intended and unintended outcomes of state housing and social welfare policies. Further, the impact of state fiscal solvency on LAC scores has critical implications for the poorest residents of struggling states. Policy makers and analysts must meaningfully explore the extent to which policy choices represent best practices versus financially convenient practices. In addition, a critical lens is required to consider how decisions about resource allocations affect poor people and people of color and further, how they hinder the social integration of those with criminal convictions who have completed their time served.

The unexpected finding that increased percentages of residents identifying as Black are associated with improved LAC scores from 2004 to 2009 warrants some discussion. This contradicts most previous findings—including our own findings concerning the cross-sectional relationship of minority presence and static LAC scores in 2009—regarding the impact of racial minority presence on punitive law making, especially in the criminal justice arena (See Sliva, 2016). However, in a more recent study of restorative justice policy adoptions, Sliva (in press) found that states with high percentages of Black residents were more likely to expand restorative justice legislation. This suggests that there may be a tipping point in the growing racial diversification of the United States at which minority populations represent a political power rather than a political threat (Sliva, in press). It is likely that a politically viable Black population would not support harsh collateral sanctions policies due to their disparate impact on Black Americans and other people of color (Alexander, 2012). While states with a larger Black population have higher LAC scores overall in 2009, our assessment of changes in LAC scores over the 4 years prior may provide evidence of the growing influence of Black citizens and point to an eventual turning point in a decades-old trend. Still, given the preponderance of findings that support the influence of racial threat on state policy, it is too soon to draw definitive conclusions about the findings of the second analysis in this study.

The relationship between public benefit usage and LAC scores revealed similar inconsistencies in cross-sectional versus longitudinal effects. While higher usage rates in 2008 are associated with more lenient collateral sanctions scores the following year, higher usage rates in 2005 are predictive of increasingly harsh sanctions over the next 4 years. This outcome partially supports our hypothesis that public benefit usage may trigger the implementation of civil disabilities. It may be that the relationship between benefit usage and collateral sanctions is more complex than imagined or better captured longitudinally than at a single point in time. Further research, perhaps utilizing case methodologies, is needed to fully understand this relationship.

Prior research by Whittle and Parker (2014) has linked political ideology to state collateral sanctions policies as operationalized by the LAC. Although we accounted for political ideology in both of our models, it was not a significant predictor of 2009 LAC scores when accounting for variables from a social exclusion framework or a significant predictor of changes in LAC scores between 2004 and 2009. It may be that the influence of political ideology is not significant when controlling for the indicators of social exclusion introduced in this study, or it may be that political ideology is not as strong of a predictor of change over time in collateral sanctions policies. Furthermore, state-level ideologies measured by the results of presidential elections can be quite fluid, as evidenced by the differences in results of the 2004, 2008, and 2016 elections.

Limitations and Implications for Future Research

Social exclusion is an admittedly difficult concept to define. The variables chosen in this study hardly represent all possible experiences or measures of social exclusion. While statistically significant, variables measured here explain only a small portion of the variation in state policy changes. Future studies may explore the influence of other factors such as political ideology or other measures of resource allocation. Further, due to the use of observational data, it should be emphasized that this study associates characteristics of population makeup, financial health, and affordable housing availability with changes in LAC collateral sanction scores. It is too soon to assert that any of these factors are causal agents in the changes in these scores. Case studies focused on states who have made significant policy changes may be particularly enlightening for future research, as these studies have the ability to explore the legislative and administrative processes behind the changes, identify contextual variables, and explore the influence of factors which are difficult to operationalize and measure, such as culture and public opinion.

Finally, LAC scores represent the cumulative scores of seven categories of collateral sanctions. Future research should further investigate whether the variables used in this study are predictive of changes in scores in each of the different categories. Given the results of this study, the relationship between the availability of affordable housing and public assistance to sanctions related to housing and public assistance may be particularly important to investigate further.

Conclusion

This article contributes to a broader understanding of criminal justice policy-making by expanding the focus of prior research beyond crime rates and recidivism. This work suggests that it is necessary to examine the relationship between collateral sanctions policies and resource allocation decisions. Collateral sanctions, promoted as a mechanism for public safety, should be neither the means nor the end of decisions to restrict resource allocation to more deserving populations. The LAC’s reporting project provides a unique opportunity to assess both the causes and the effects of state policy decisions in this arena, including key outcomes related to the integration or, alternatively, exclusion of socially marginalized populations including previously incarcerated persons. Further research should focus on whether or not state-level policies are actually hindering the ability of their residents to achieve financial and social independence. This research can ultimately provide avenues to implement alternative, empirically supported policies that make self-sufficiency and social integration a realistic opportunity for as many people as possible.

Footnotes

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

References

Alexander

(2012). The new Jim Crow: Mass incarceration in the age of colorblindness (rev. ed.). New York, NY: New Press.

American Bar Association. (2017). Standard 19-1.1: Collateral sanctions and discretionary disqualification of convicted persons. Retrieved from http://www.americanbar.org/publications/criminal_justice_section_archive/crimjust_standards_collateral_blk.html

Boire

R. G.

(2007). Life sentences: The collateral sanctions associated with marijuana offenses. Retrieved from Center for Cognitive Liberty and Ethics website: http://www.cognitiveliberty.org/rpts/col_sanctions.htm

Bruce

Crowley

Jeffcote

Coulston

(2014). Community DSPD pilot services in South London: Rates of reconviction and impact of supported housing on reducing recidivism. Criminal Behaviour & Mental Health, 24, 129–140. doi:10.1002/cbm.1892

Burton

V. S.

Cullen

F. T.

Travis

L. F.

(1987). The collateral consequences of a felony conviction: A national study of state statutes. Federal Probation, 51, 52–60.

Bushway

S. D.

Sweeten

(2007). Abolish lifetime bans for ex-felons. Criminology and Public Policy, 6, 697–706.

Carson

A. E.

(2015). Prisoners in 2014. Retrieved from the Bureau of Justice Statistics website: http://www.bjs.gov/index.cfm?ty=pbdetail&iid=5387

Chin

G. J.

(2002). Race, the war on drugs, and the collateral consequences of criminal convictions. Journal of Gender, Race, and Justice, 6, 253–254.

Clifasefi

S. L.

Malone

D. K.

Collins

S. E.

(2013). Exposure to project-based Housing First is associated with reduced jail time and bookings. International Journal of Drug Policy, 24, 291–296. doi:10.1016/j.drugpo.2012.10.002

10.

Davis

E. P.

Sanchez-Martinez

(2015). Economic theories of poverty. Retrieved May 1, 2016, from the Joseph Rowntree Foundation website: https://www.jrf.org.uk/report/economic-theories-poverty

11.

Durose

M. R.

Cooper

A. D.

Snyder

H. N.

(2014). Recidivism of prisoners released from 30 states in 2005: Patterns from 2005 to 2010. Retrieved from the Bureau of Justice Statistics website: http://www.bjs.gov/content/pub/pdf/rprts05p0510.pdf

12.

Ellison

Fox

Gains

Pollock

(2013). An evaluation of the effect of housing provision on re-offending. Safer Communities, 12, 27–37. doi:10.1108/17578041311293125

13.

Evans

D. N.

Porter

J. R.

(2014). Criminal history and landlord rental decisions: A New York quasi-experimental study. Journal of Experimental Criminology, 11, 21–42. doi:10.1007/s11292-014-9217-4

14.

Federal Bureau of Investigation. (2016). Crime statistics. Retrieved May 29, 2016, from the Federal Bureau of Investigation website: https://www.fbi.gov/stats-services/crimestats

15.

Grady

S. C.

(2012). Civil death is different: An examination of a post-Graham challenge to felon disenfranchisement under the Eighth Amendment. The Journal of Criminal Law and Criminology, 102, 441–470.

16.

Henry

J. S.

Jacobs

J. B.

(2007). Ban the box to promote ex offender employment. Criminology and Public Policy, 6, 755–762.

17.

Hoefer

Black

Salehin

(2012). Dating violence policy: Making the grade. Journal of Sociology and Social Welfare, 39, 9–24.

18.

Holtfreter

Reisig

M. D.

Morash

(2004). Poverty, state capital, and recidivism among women offenders. Criminology & Public Policy, 3, 185–208.

19.

Kutner

M. H.

Nachtsheim

C. J.

Neter

(2004). Applied linear regression models (4th ed.). Boston, MA: McGraw-Hill Irwin

20.

Labovitz

(1968). Criteria for selecting a significance level: A note on the sacredness of .05. The American Sociologist, 3, 220–222.

21.

Legal Action Center. (2004). After prison: Roadblocks to reentry. A report on state legal barriers facing people with criminal records. Retrieved from Legal Action Center website: http://www.lac.org/roadblocks-to-reentry/upload/lacreport/LACPrintReport.pdf

22.

Legal Action Center. (2009). After prison: Roadblocks to reentry. A report on state legal barriers facing people with criminal records. 2009 update. Retrieved from Legal Action Center website: http://www.lac.org/roadblocks-to-reentry/upload/lacreport/Roadblocks-to-Reentry–2009.pdf

23.

Legal Action Center. (2016). Retrieved May 29, 2016, from the Legal Action Center website: http://lac.org/

24.

Luca

D. L.

(2011). The digital scarlet letter: The effect of online criminal records on crime. Retrieved May 29, 2016, from the Social Science Research Network website: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=1939589

25.

Morazes

Pintak

(2007). Theories of global poverty comparing developed world and developing world frameworks. Journal of Human Behavior in the Social Environment, 16, 105–121.

26.

Nally

J. M.

Lockwood

Knutson

(2014). Post-release recidivism and employment among different types of released offenders: A 5-Year follow-up study in the United States. International Journal of Criminal Justice Sciences, 9, 16–34.

27.

Nice

D. C.

(1994). Policy innovation in state government. Ames: Iowa State University Press.

28.

Olivares

K. M.

Burton

V. S.

Cullen

F. T.

(1996). The collateral consequences of a felony conviction: A national study of state legal codes 10 years later. Federal Probation, 60, 10–17.

29.

Petersilia

(2003). When prisoners come home: Parole and prisoner reentry. New York, NY: Oxford University Press.

30.

Pettus-Davis

(2012). Reverse social work’s neglect of adults involved in the criminal justice system: The intersection and an agenda. Social Work Research, 36, 3–8. doi:10.1093/swr/svs036

31.

Pinard

(2006). An integrated perspective on the collateral consequences of criminal convictions and reentry issues faced by formerly incarcerated individuals. Boston University Law Review, 86, 623, 651–652.

32.

Radice

(2012). Administering justice: Removing statutory barriers to reentry. University of Colorado Law Review, 83, 715–779.

33.

Roman

C. G.

Travis

(2004). Taking stock: Housing, homelessness and prisoner reentry. Retrieved from Urban Institute website: http://www.urban.org/research/publication/taking-stock

34.

Sliva

S. M.

(2016). Determinants of state innovations in American sentencing and corrections policy: A systematic review. Criminal Justice Policy Review, 27, 702–722.

35.

Sliva

S. M.

(In Press). Finally ‘changing lenses’? State level determinants of restorative justice laws. The Prison Journal.

36.

Sohoni

T. W. P.

(2013). The effect of collateral consequence laws on state rates of returns to prison. 3611943 Ph.D., University of Maryland, College Park, Ann Arbor. Retrieved from http://search.proquest.com/docview/1504840083?accountid=14608

37.

Somers

J. M.

Rezansoff

S. N.

Moniruzzaman

Palepu

Patterson

(2013). Housing first reduces re-offending among formerly homeless adults with mental disorders: Results of a randomized controlled trial. PLoS One, 8, e72946. doi:10.1371/journal.pone.0072946

38.

Stock

J. H.

Watson

M. W.

(2003). Introduction to econometrics. New York, NY: Prentice Hall.

39.

Tax Foundation. (2007). Federal Taxes Paid vs. Federal Spending Received by State, 1981–2005. Retrieved May 29, 2016, from Tax Foundation website: http://taxfoundation.org/article/federal-taxes-paid-vs-federal-spending-received-state-1981-2005

40.

Tax Foundation. (2016). About us. Retrieved May 29, 2016, from Tax Foundation website: http://taxfoundation.org/about-us

41.

Uggen

(2000). Work as a turning point in the life course of criminals: A duration model of age, employment, and recidivism. American Sociological Review, 65, 529.

42.

U.S. Census Bureau. (2016). American fact finder. Retrieved May 29, 2016, from United States Census Bureau website: http://factfinder.census.gov/faces/nav/jsf/pages/index.xhtml

43.

U.S. Department of Housing and Urban Affairs. (2016). Affordable housing. Retrieved May 17, 2016, from United States Department of Housing and Urban Affairs website: http://portal.hud.gov/hudportal/HUD?src=/program_offices/comm_planning/affordablehousing/

44.

U.S. House of Representatives: History Art and Archives. (2016). Election statistics, 1920 to present. Retrieved May 29, 2016, from United States House of Representatives: History, Art and Archives website: http://history.house.gov/Institution/Election-Statistics/Election-Statistics/

45.

West

H. C.

Sabol

W. J.

Greenman

S. J.

(2010). Prisoners in 2009. Retrieved May 29, 2016, from Bureau of Justice Statistics website: http://www.bjs.gov/content/pub/pdf/p09.pdf

46.

Whittle

T. N.

(2016). Felony collateral sanctions effects on recidivism: A literature review. Criminal Justice Policy Review. doi:10.1177/0887403415623328

47.

Whittle

T. N.

Parker

K. F.

(2014). Public ideology, minority threat, and felony collateral sanctions: A state-level analysis. Criminal Justice Review, 39, 432–454. doi:10.11770734016814543352.