Perceptions of Prison and Punitive Attitudes

Abstract

The penal escalation hypothesis holds that people’s (mis)perception of prisons as being inadequately harsh will influence their punitive attitudes toward other punishments and components of the criminal justice system. In this study, I present the first test of the penal escalation hypothesis with survey data from residents of the United States of America. I find that people’s perceptions of life and conditions in prison are significantly related to their opinions about other punitive and progressive aspects of criminal justice policy and practice. I argue that these findings should lead scholars to reevaluate the importance of instrumental versus symbolic factors when attempting to explain variation in people’s opinions about punishment and justice.

Keywords

public opinion prison criminal justice punitive penal escalation

From the mid-1970s through the late 2000s, the rate of incarceration in the United States of America grew exponentially and unceasingly. This growth culminated in the fact that the United States presently has the highest incarceration rate of any country in the world (Walmsley, 2013). Numerous scholars find evidence that politics and policy choices played a major role in the growth of “mass incarceration” (National Research Council, 2014). Public anxieties about crime and demands for harsh punishment were one such political force that fueled mass incarceration, at least in part (Enns, 2014; Nicholson-Crotty, Peterson, & Ramirez, 2009). However, despite this evidence, very little research has examined the causes and consequences of American opinions about prisons, specifically.

In this study, I test Roberts and Hough’s (2005) penal escalation hypothesis, which contends that people who perceive prison environments to be insufficiently unpleasant or harsh will demand that other components of the criminal justice compensate with tougher punishments and stricter treatment of people under correctional supervision. This hypothesis was inspired by the fact that politicians frequently decried “country club conditions” in prisons during the height of the “tough-on-crime” era (Bidinotto, 1994). Even controlling for several common attitudinal predictors of public opinion about criminal justice, I find support for the penal escalation hypothesis. People’s perceptions of the prison environment are significantly related to their support for punishment, police, crime prevention, and prisoner reentry programs, although the precise nature of that relationship varies somewhat across measures. I argue that these findings suggest that scholars have underestimated the degree to which instrumental concerns about the functioning of the justice system affect public opinion about punishment and criminal justice.

Perceptions of Prison and Penal Escalation

A decade ago, Roberts and Hough (2005) reviewed the empirical literature in order to assess public opinion about prisons. They observed three common trends among survey respondents from several different nations in Europe and North America. First, most citizens have little knowledge of the environment or inner workings of prisons, and few have had direct contact with a correctional institution. Second, in spite of the scarcity of personal experience, many citizens believe that prison conditions are overly lenient or easy on inmates. Third, public opinion about prisons is not uniformly punitive; many citizens support inmate access to amenities, rehabilitation, and educational programs.

These latter two points seem to be contradictory. How can people support the amenities and programs that are part of the “easy” environment that they resent? Some evidence from qualitative studies suggests that a key factor is respondents’ perceptions of inmates’ activities; interview and focus group participants primarily objected to inmates’ ability to spend their days being lazy and unproductive, and they were more supportive of amenities, programs, and work opportunities that inmates used to rehabilitate themselves (e.g., Doble, 1987). Furthermore, public support varies from amenity to amenity (e.g., Applegate, 2001; Lenz, 2002). However, the precise relationship between the perception that life in prison is easy and support for amenities and programs is not terribly well understood because the few studies that have measured these constructs have tended to measure one or the other but not both within the same survey.

Beyond prison amenities, Roberts and Hough (2005) theorized that people’s perceptions of the nature of life in prison affect their attitudes toward other criminal justice policies and programs. It has been well documented that many Western, industrialized democracies expanded the scope of their criminal justice systems to varying degrees over the past 40 years (e.g., Baker, 1996; Garland, 2001; Roberts, Stalans, Indermaur, & Hough, 2003). Roberts and Hough argued that people’s (mis)perceptions of prisons may have partially fueled the politics and policies that harshened criminal punishment and increased the size of prison populations in many nations, the United States foremost among them. They stated,
A number of adverse consequences ensue from the public perception that prison life is not particularly unpleasant. First, people will be less likely to see prison as the punishment that it clearly is …. This perception (that prison is easy) depreciates the penal value of imprisonment and can only exert an inflationary pressure on sentence lengths; it will fuel public calls for longer sentences. (Roberts & Hough, 2005, p. 292)
They dub this relationship between perceptions of prison and punitive attitudes toward other facets of criminal justice penal escalation.

Unfortunately, the penal escalation hypothesis has been subjected to extremely little empirical testing. Indeed, the only published studies of which I am aware that test the relationship between people’s perceptions of prison and their broader attitudes about criminal justice preceded and informed Roberts and Hough’s essay. Analyzing data from a sample of respondents drawn from Montreal and Toronto, Brillon, Louis-Guerin, and Lamarche (1984) found that people who perceived prisons to be “veritable hotels” were also likely to believe that courts do not deliver sufficiently severe sentences and were less supportive of granting parole and access to halfway houses for all offenders. Analyzing data from a representative sample of the South African population, Glanz (1994) found that Black respondents who perceived conditions in prison to be very good, good, or fair were also likely to favor long prison sentences for all offenders; the relationship was not statistically significant for respondents of other races. To my knowledge, no scholar has tested the penal escalation hypothesis with data from the United States. This gap in our knowledge is glaring in light of the fact that the vast size of the United States’ prison population makes it an international outlier (Walmsley, 2013), and organizations such as Human Rights Watch, Amnesty International, and the American Civil Liberties Union have criticized overcrowding and inhumane conditions in American prisons. Furthermore, Simon (2014) argues that U.S. courts are beginning to craft contemporary jurisprudence that acknowledges and confronts the (negative) implications of prison conditions for inmates’ constitutional rights. In spite of these criticisms, (Wozniak, 2014) found that a plurality of American respondents believe that the living environment in prisons is “not harsh enough.” We do not yet know if this perception contributes to Americans’ punitive attitudes about other facets of the justice system as well.

Implications for Theories of Public Opinion About Criminal Justice

I argue that scholars should pay more attention to Roberts and Hough’s penal escalation hypothesis because it challenges a commonly accepted belief in the public opinion literature. Starting with Tyler and Boeckmann’s (1997) seminal paper, scholars have tested the ability of instrumental versus symbolic factors to explain variation in people’s opinions about criminal justice. Instrumental factors are typically operationalized as people’s perceptions of the actual crime rate, their personal victimization experience, and their fear of crime. The theory behind instrumental predictors of attitudes is that the criminal justice system is designed to prevent and reduce crime, so the more that people perceive that society is experiencing a crime problem and fear victimization, the more supportive they will be of punitive justice policies. On the other hand, symbolic factors are typically operationalized as people’s ideological beliefs and perceptions of abstract social forces or trends, such as a decline in moral values. The theory behind symbolic predictors is that the criminal justice system maintains social order, so the more that people perceive their desired social status quo to be under threat by immoral, chaotic forces, the more they will support punitive justice policies to reestablish the cultural status quo (e.g., Durkheim, 1893 /1964; Garland, 2001).

Most scholars conclude that symbolic factors are stronger predictors of people’s opinions about criminal justice than are instrumental factors (Frost, 2010; King & Maruna, 2009; Tyler & Boeckmann, 1997; Unnever & Cullen, 2010). This conclusion largely rests on the fact that the empirical relationship between fear of crime and punitive attitudes is mixed across studies, and actual victimization experience is rarely related to people’s opinions about criminal justice policy once you control for their symbolic beliefs (Frost, 2010). However, by operationalizing instrumental concerns almost exclusively through items related to perceptions of crime, previous studies largely fail to examine whether or not people’s evaluation of one facet of the criminal justice system affects their opinions about other facets of the system. This, too, would be an instrumental relationship; if people perceive that prisons are not adequately “doing their job” of punishing offenders, they may want other justice systems to compensate in order to fulfil the demands of retribution.

Contributions of the Present Study

In this study, I test the penal escalation hypothesis with data from an original public opinion survey administered to a random sample of residents of the United States. As the first empirical test of the penal escalation hypothesis using data from the United States, I test the generalizability of the theory. I also contribute to this literature by employing a wide range of measures. My key independent variables are three measures of people’s perceptions of the living environment in prisons. My dependent variables measure people’s support for several different punitive and progressive policy responses to crime. I include a variety of dependent variables in recognition of the growing body of literature that finds that people’s opinions about criminal justice are multidimensional; many people express support for both punitive punishments (like the death penalty and long prison sentences) and progressive alternatives (like rehabilitation and crime prevention; e.g., Duffee & Ritti, 1977; Green, Staerkle, & Sears, 2006; Maguire & Johnson, 2015; Mascini & Houtman, 2006; Pickett, Chiricos, & Gertz, 2014; Ramirez, 2014; Unnever, Cochran, Cullen, & Applegate, 2010). I control for a variety of symbolic, ideological predispositions that are frequently related to criminal justice opinions, such as respondents’ attribution of blame for offending, racism, and political ideology as well as the “standard” instrumental measures of fear of crime and victimization. Controlling for ideological attitudes in the analyses allows me to test whether or not people’s instrumental perceptions of prisons shape their punitive and progressive policy opinions above and beyond their symbolic concerns about society.

I hypothesize that respondents who perceive inmates to be idle or life in prison to be easy or insufficiently harsh will be more supportive of other harsh punishments and punitive policies and less supportive of progressive, rehabilitative policies than respondents who do not perceive prisons to be insufficiently harsh.

Method

Data

The data for this study come from an original public opinion survey conducted between September 2010 and March 2011. The sampling frame for the study was purchased from the marketing research service InfoUSA, which maintains a regularly updated database of 230 million U.S. households. The frame was comprised of 3,000 addresses randomly sampled from InfoUSA’s database. I contracted with the Center for Social Sciences and Public Policy Research at Missouri State University to procure the sample and administer the survey. The study was funded with research grants from American University.

Given the racialized nature of the crime debate in the United States (Chiricos, Welch, & Gertz, 2004; Hurwitz & Peffley, 2005), I chose a paper, mail-based delivery mode in order to minimize race-related response biases; research indicates that in-person and phone surveys are more vulnerable to these biases than mail surveys (Krysan, 1998). I followed the “tailored design method” advocated by survey methodologists (Dillman, 1991; Dillman, Smyth, & Christian, 2009; Groves et al., 2004). The survey utilized several elements that increase response rates, such as multiple waves of mailings of reminder postcards and replacement questionnaires, envelopes with a university seal and hand-printed addresses, and prepaid return envelopes.

The survey included a total of 42 questions. In addition to the survey items analyzed in this study, the survey measured respondents’ opinions about a variety of different crime control policies and punishments and their political awareness. For more information about the questionnaire and survey design, see (Wozniak, 2012).

I received a total of 501 usable responses plus an additional 228 undeliverable packets. The response rate was 16.7%, and the cooperation rate was 18.1%.¹ The racial distribution of the sample is 76.9% White, 9.8% Black, and 8.1% Other. Five percent of respondents identified their ethnicity as Latino. The sample is 39.5% female and 60.5% male. The average age of respondents is 57.2, with a standard deviation of 15.5 years. Just over 50% of respondents are 60 years of age or older. Twenty percent of respondents have a high school diploma or less formal education, 36.7% have some college education or an associate’s degree, 22.7% have a bachelor’s degree, and 20.6% have a graduate or professional degree. The sample includes residents of 48 different states plus the District of Columbia.

While the response rate is low, survey methodologists note that response rates have been dramatically declining in recent years (Curtin, Presser, & Singer, 2005; de Leeuw & de Heer, 2002). For example, in 2012, the Pew Center conducted a standard, 5-day-long phone survey with no incentives and a high-effort phone survey that involved more than two dozen contacts over 2½ months plus financial incentives for participation. The standard method survey procured only a 9% response rate. Even the high-effort survey procured only a 22% response rate, which is not much higher than the response rate of this survey even though phone surveys typically yield higher response rates than mail surveys, and this survey did not possess the benefit of a financial incentive (Pew Center, 2012).

However, methodologists demonstrate that response bias is not a direct function of the response rate because a survey with few respondents could still theoretically possess sample statistics that match the population parameters (Groves, 2006; Groves & Peytcheva, 2008; Keeter, Miller, Kohut, Groves, & Presser, 2000). Indeed, the Pew Center found few significant differences between its standard and high-effort survey data sets, and both data sets possessed sample statistics that were comparable to benchmark estimates from the Census Bureau’s Current Population Survey (Pew Center, 2012). In light of these findings, survey methodologists recommend that generalizability be assessed through direct examination of the sample statistics rather than evaluated on the response rate alone.

To assess the external validity of these data, I compare the demographic characteristics of my sample to data from the 2010 census of the U.S. population in Table 1. This comparison shows that my sample is disproportionately non-Latino White, male, educated, and older compared to the national population; these response biases are relatively common in contemporary survey samples (Pew Center, 2012). I comment on the implications of these response biases for the generalizability of these data in the Discussion section.

Table 1.
Sample Versus 2010 U.S. Census Demographic Comparison.

Characteristic Survey Data (%) Census Data (%)

White/Caucasian 76.9 72.4

Black/African American 9.8 12.6

Latino 5.0 16.3

Other race 8.1 12.1

Male 60.5 49.2

Age 18–64 63.3 82.1^a

Age 65 and older 36.7 17.9^a

College, graduate, or advance professional degree 43.3 38.2

Note. ^aIn order to accurately compare the census data to the survey sample, which completely excludes respondents younger than 18 years of age, I recalculated the age percentages of the census data to similarly exclude respondents younger than 18 years of age. The true percentages of Americans age 18–64 and age 65 and older, when including people younger than 18 in the denominator, are 62.9% and 13.0%, respectively.

Dependent Variables

Punishment preference

I adapted one question from the 2000 National Election Study (American National Election Studies [ANES], 2015) to measure respondents’ preference for punishment versus addressing social causes of crime. The question is, “Some people say that the best way to reduce crime is to address the social problems that cause crime, like bad schools, poverty, and joblessness. Other people say the best way to reduce crime is to make sure that criminals are caught, convicted, and punished. How about you? Which approach to fighting crime do you think is better than the other?” This question had six response options: 1/2/3 = much/moderately/slightly better to fix social problems (BTFSPs); and 4/5/6 = slightly/moderately/much better to punish criminals (BTPCs). Thus, higher scores indicate a stronger preference for punishment, which I classify as the punitive response.

Government options for addressing crime

Four questions measured respondents’ support for crime control policies, all of which were embedded under the same root umbrella: “When it comes to fighting crime, the government can choose from a variety of different tactics. How much do you oppose or support …. (1) Building more prisons to house more offenders? (2) Funding programs to help former prisoners find jobs and housing after they have completed their prison sentence in order to reduce the chance that they will commit new crimes? (3) Funding programs to help prevent at-risk youths from committing crimes? (4) Hiring more police officers?” Each question had a 6-point Likert-type response option ranging from 1 = strongly oppose to 6 = strongly support. I classify support for building more prisons or hiring more police as punitive in nature, while support for funding reentry or crime prevention programs is progressive in nature.

Independent Variables

Perceptions of life in prison

Three questions measured respondents’ perceptions of life in prison. They shared the same root format, which was “Overall, do you think that life in prison is _, or is it _?” Respondents could choose between two contrasting adjectives, which were (a) hard or easy, (b) depressing or enjoyable, and (c) dangerous or safe. The response options were 6-point, bipolar scales. For example, the response option to the first question was 6 = very hard, 5 = moderately hard, 4 = slightly hard, 3 = slightly easy, 2 = moderately easy, and 1 = very easy.

I reverse coded these questions, so that higher values indicate perceptions that are more consistent with the image of plush prisons (i.e., higher values indicate perceptions of life as easy, enjoyable, and/or safe), and I combined these 3 items into an additive scale. The correlations among these 3 items are moderate to strong (ranging from 0.46 to 0.64), and the Cronbach’s α of the combined scale is .80.

Perception of inmate idleness

A single question asked respondents, “How do you think that most prisoners actually spend their time? Do you think that most prisoners spend their time being idle and lazy, or do you think that most prisoners spend their time being productively engaged in prison jobs or educational classes?” Similar to the previous items, this question had a 6-point, bipolar response scale ranging from very idle to very productive, which I coded so that higher values indicate a stronger belief that inmates are idle and lazy.

Prison punitiveness

One question measured respondents’ preferences for harsh prisons, which was “Overall, do you think that the living environment in prisons is too harsh, about right, or not harsh enough for inmates?” This question had a 3-point response scale, where 1 = too harsh, 2 = about right; neither too harsh nor too lenient, and 3 = not harsh enough. I recoded the measure so that 1 = not harsh enough and 0 = too harsh or about right. I collapsed the categories because only 11.8% of respondents stated that the prison environment is “too harsh” (as opposed to 41.6% saying “about right” and 46.6% saying “not harsh enough”).

Control Variables

Attribution of blame

Following the work of Unnever, Cochran, Cullen, and Applegate (2010), I construct separate scales for a respondent’s dispositional and situational attributions of blame. These variables measure whether respondents believe that crime is predominantly caused by bad choices and personal failings (dispositional) or societal/environmental pressures (situational). I replicated the seven questionnaire items that operationalize these two variables directly from Unnever et al. (2010). Higher values on these scales indicate stronger agreement with each type of attribution. The Cronbach’s α of the dispositional attribution scale is .47, and the Cronbach’s α of the situational attribution scale is .56.²

Symbolic racism

This variable is comprised of 6 items that I adapted from the ANES and the Los Angeles County Social Survey (see Green et al., 2006; Unnever & Cullen, 2007). It includes statements such as “It’s really a matter of some people not trying hard enough; if Blacks would only try harder, they would be just as well off as Whites” and “Generations of slavery and discrimination have created conditions that make it difficult for Blacks to work their way out of the lower class.” Each of these items had a 6-point Likert-type scale to indicate strength of agreement or disagreement. All items were coded in such a manner that higher values on the combined scale indicate stronger agreement with the tenets of symbolic racism, which is the belief that higher rates of disadvantage among African Americans compared to other racial groups are due to Blacks’ inadequate work ethic and rejection of mainstream American values rather than systemic discrimination (see Tarman & Sears, 2005). The Cronbach’s α of the scale is .79.

Perception of fairness in the criminal justice system

This variable combines 3 items that measure a respondent’s level of agreement or disagreement with statements that the criminal justice system is fair regardless of race, the death penalty is applied fairly regardless of race, and a Black person is likely to receive a harsher sentence than a White person who committed the same crime (the last item is reverse coded). Higher values indicate a stronger belief that the justice system is fair and unbiased, which scholars find to be a strong predictor of people’s opinions about punishment (e.g., Bobo & Johnson, 2004; D. Johnson, 2008; Peffley & Hurwitz, 2010). The Cronbach’s α of the additive scale is .78.

Political ideology

This construct was measured with a single question: “Using this scale, how would you describe your political ideology?” This question had a 7-point response scale that ranged from extremely liberal (value of 1) through moderate/middle-of-the-road (value of 4) to extremely conservative (value of 7).

News consumption

This variable is an additive scale that measures the number of days during a typical week that the respondent consumes the news via TV, radio, Internet, and/or newspaper; higher values on this scale indicate a greater volume of news consumed from a greater variety of sources.

Fear of crime

A single-item variable measures this construct with a respondent’s answer to the question, ‘Within the past 6 months, have you ever felt afraid that you would become the victim of a crime?” The response options were yes (coded 1) or no (coded 0).

Victimization

Another single-item variable measures whether or not the respondent was the victim of a crime with the question, “Have you ever been the victim of a serious crime? This would include such things as someone breaking into your home, having your car stolen, or being physically assaulted or robbed.” This item had the same dichotomous response options as the fear of crime question.

Demographics

I control for a variety of demographic characteristics in order to partially address the response biases in the data. Dummy variables contrast the opinions of Democrats and Independents against those of Republicans (the omitted category). Dummy variables identify African Americans (1 = Black, 0 = all other races) and women. Age is a continuous variable. Education and household income are each ordinal variables coded so that higher scores indicate higher levels of educational achievement and income. Finally, dummy variables identify respondents who live in states in the Northeast, Central, and West regions of the country, contrasted to those who live in the South (the omitted category); I use the U.S. Census categorization of regions.

Plan of Analysis and Regression Diagnostics

I conducted all analyses with Stata version 13 (Statacorp, 2013). In order to test for multicollinearity, I first regressed each dependent variable on the independent and control variables using ordinary least squares (OLS) regression and calculated variance inflation factor (VIF) scores. No variable in any model generated a VIF score greater in magnitude than 2.70, which falls below the standard thresholds of concern (Fisher & Mason, 1981; Fox, 1991). However, the ordinal nature of the dependent variables violates the assumptions of OLS, so I ran preliminary analyses using ordinal logistic maximum likelihood estimation (Long, 1997). Postestimation Brant tests revealed that a small number of variables in each ordinal logistic model violated the parallel regression assumption, which means that the coefficients of those variables were not consistent across categories of the dependent variable. In order to overcome this violation, I reestimated the models as partial proportional odds models using the gologit2 command in Stata (Williams, 2006). This model allows the analyst to constrain variables that meet the parallel regression assumption so that their coefficients are identical across categories of the dependent variable while simultaneously allowing the coefficients of variables that violate the assumption to vary across categories.

Constrained variable effects can be interpreted like normal ordinal logistic regression (i.e., how a unit change in an independent variable affected the likelihood that a respondent would choose the “next highest” category of the dependent variable). The partial proportional odds model produces output for the unconstrained variables that is similar in nature to a multinomial logistic regression. Essentially, the partial proportional odds model simultaneously estimates a series of binary logistic regressions in which the categories of the dependent variable are iteratively compared to each other. Since the dependent variables in this study are 6-point scales, the first panel of results for the unconstrained variables contrasts category 1 (coded 0) of the dependent variable against Categories 2, 3, 4, 5, and 6 (coded 1). The second panel contrasts Categories 1 and 2 (coded 0) against Categories 3, 4, 5, and 6 (coded 1). The third panel contrasts Categories 1, 2, and 3 (coded 0) against Categories 4, 5, and 6 (coded 1), and so on. Thus, to interpret the present results, positive coefficients for unconstrained variables indicate that higher values on the explanatory variable made a respondent more likely to choose a more punitive category of the punishment preference question or more likely to express weaker opposition and/or stronger support for each crime prevention policy. Negative coefficients indicate that higher values on the explanatory variable decreased the likelihood of choosing a more punitive or more supportive answer (Williams, 2006).

Cases with missing data were eliminated via listwise deletion, which means that the analysis samples were lower than the total sample size, a tactic that Allison (2002) defends as no less robust than most alternative means of addressing missing data.

Results

Table 2 presents the results of the generalized ordered logit regression of people’s preferences for punishment versus fixing social problems as a way to reduce crime. Four of the constrained variables significantly affected respondents’ punishment preference, all of which are “symbolic,” attitudinal measures. Each scale unit increase toward stronger belief in a dispositional attribution of blame was associated with a 13% increased likelihood that a respondent would choose a response more supportive of punishment than fixing social problems. In contrast, each scale unit increase toward stronger belief in a situational attribution of blame was associated with an 8% decreased likelihood of choosing a more punitive response. Each scale unit increase toward stronger agreement with the tenets of symbolic racism was associated with a 6% increased likelihood of choosing a more punitive response. Lastly, each unit increase on the political ideology scale, moving from liberal to conservative, was associated with a 26% increased likelihood of more strongly preferring punishment to fixing social problems.

Table 2.
Results of Generalized Ordered Logit Regressions of Punishment Preference on Prison Perceptions and Control Variables.

Variables Punishment Preference

Constrained factors

Perception of inmate idleness 1.05 (.09)

Prison punitiveness 1.40 (.32)

Dispositional attribution of blame 1.13 (.04)*

Situational attribution of blame 0.92 (.03)

Symbolic racism 1.06 (.03)*

Perceived fairness of criminal justice system 1.03 (.04)

Political ideology (conservative) 1.26 (.13)*

News consumption 1.01 (.02)

Fear of crime 1.06 (.30)

Victimization 0.85 (.20)

Democrat 0.78 (.24)

Independent 0.62 (.18)

African American 0.73 (.35)

Female 1.00 (.24)

Age 1.00 (.01)

Education 0.90 (.07)

Household income 0.95 (.07)

Northeast 0.67 (.21)

Central 0.67 (.17)

West 0.76 (.22)

Unconstrained factors

[Much BTFSP] versus [Mod. BTFSP, Slt. BTFSP, Slt. BTPC, Mod. BTPC, Much BTPC]

Perception of life in prison scale 1.27 (.08)*

[Much BTFSP, Mod. BTFSP] versus [Slt. BTFSP, Slt. BTPC, Mod. BTPC, Much BTPC]

Perception of life in prison scale 1.25 (.06)*

[Much BTFSP, Mod. BTFSP, Slt. BTFSP] versus [Slt. BTPC, Mod. BTPC, Much BTPC]

Perception of life in prison scale 1.22 (.05)*

[Much BTFSP, Mod. BTFSP, Slt. BTFSP, Slt. BTPC] versus [Mod. BTPC, Much BTPC]

Perception of life in prison scale 1.15 (.04)*

[Much BTFSP, Mod. BTFSP, Slt. BTFSP, Slt. BTPC Mod. BTPC] versus [Much BTPC]

Perception of life in prison scale 1.03 (.04)

Log pseudolikelihood −528.40

N 365

McFadden’s pseudo R ² .17

Note. Odds ratios with robust standard errors are given in parentheses. Slt. = slightly; Mod. = moderately; BTFSP = better to fix social problems; BTPC = better to punish criminals.

^† p ≤ .10. p ≤ .05. p ≤ .01. p ≤ .001.

Only one measure related to the penal escalation hypothesis significantly affected respondents’ preference for punishment versus fixing social problems, and this variable had to be unconstrained because it violated the parallel regression assumption. Examining the unconstrained odds ratios, we see that the more strongly that respondents perceived life in prison to be easy, enjoyable, and/or safe, the more likely they were to support punishing offenders. This effect was most pronounced in reducing the odds that a respondent would say that it is “much BTFSPs” versus all other response options and least pronounced in increasing the odds that a respondent would say that it is “moderately” or “much BTPCs” versus all other response options. In fact, people’s perceptions of life in prison did not significantly affect their odds of choosing “much BTPCs” contrasted to all other categories. Overall, then, these results suggest that perceptions of life in prison as easy, enjoyable, and/or safe make people resistant to fixing social problems as a response to crime, but they do not increase a preference for punishment to the same degree.

Examining the odds ratios, we see that people’s perceptions of life in prison and their political ideology exerted the largest substantive effects on their preference for punishment versus fixing social problems; depending upon the level of the dependent variable, the effects of these two variables were comparable. People’s attributions of blame and agreement with the tenets of symbolic racism were associated with comparatively smaller shifts in their punishment preferences.

Table 3 presents the results of generalized ordered logit regressions of people’s support for punitive and progressive policies to fight crime. I will first discuss the effects of respondents’ perceptions of and attitudes toward prisons, then I will discuss the effects of their symbolic attitudes, and lastly I will discuss the effects of demographic characteristics.

Table 3.
Results of Generalized Ordered Logit Regressions of Punitive and Progressive Crime Policy Opinions on Prison Perceptions and Control Variables.

Variables More Prisons More Police Prisoner Reentry Crime Prevention

Constrained factors

Perception of life in prison scale 1.08 (.03)* 0.93 (.03)* 0.81 (.03)*** (unconstrained)

Perception of inmate idleness 0.82 (.06)* 0.95 (.08) 1.22 (.10)* (unconstrained)

Prison punitiveness (unconstrained) (unconstrained) 1.01 (.25) 2.06 (.52)

Dispositional attribution of blame 1.15 (.04)* 1.07 (.04)* 1.02 (.04) 1.04 (.04

Situational attribution of blame 1.01 (.03) 1.02 (.03) 1.11 (.03)*** 1.12 (.04)*

Symbolic racism 1.07 (.03)* 1.05 (.03)* 1.00 (.03) 0.97 (.02)

Perceived fairness of criminal justice system 1.01 (.04) 0.98 (.03) 0.96 (.03) 0.97 (.04)

Political ideology (conservative) 1.10 (.12) 1.13 (.10) 0.98 (.10) 1.21 (.12)^†

News consumption 0.97 (.02) 1.05 (.02) 1.03 (.02) 1.01 (.02)

Fear of crime 1.28 (.30) 1.18 (.27) 0.68 (.19) 0.74 (.20)

Victimization 0.93 (.20) 1.31 (.28) 1.28 (.30) 0.92 (.23)

Democrat 0.61 (.19) 1.22 (.39) (unconstrained) 1.81 (.58)^†

Independent 0.52 (.14)* 0.90 (.25) 0.95 (.25) 1.48 (.39)

African American 0.88 (.34) 2.11 (.81)^† 1.94 (1.02) 2.54 (1.44)^†

Female 1.01 (.22) 1.83 (.43) (unconstrained) 2.04 (.51)

Age 1.00 (.01) 1.00 (.01) 1.02 (.01)* 1.02 (.01)*

Education 1.03 (.08) 1.08 (.08) 0.98 (.07) 1.09 (.08)

Household income 1.06 (.08) 1.02 (.07) 0.96 (.06) 1.00 (.07)

Northeast 1.37 (.38) (unconstrained) (unconstrained) 0.86 (.25)

Central 1.42 (.39) 1.26 (.32) 0.85 (.24) 1.03 (.29)

West 1.05 (.27) 0.78 (.24) 0.71 (.20) 1.47 (.46)

Unconstrained factors

[Str. Opp.] versus [Mod. Opp., Slt. Opp., Slt. Supp., Mod. Supp., Str. Supp.]

Perception of life in prison scale — — — 0.91 (.08)

Perception of inmate idleness — — — 0.85 (.22)

Prison punitiveness .65 (.22) 0.69 (.46) — —

Democrat — — 2.23 (1.65) —

Female — — 3.07 (1.92)^† —

Northeast — 1.19 (1.07) 0.19 (.14)* —

[Str. Opp., Mod. Opp.] versus [Slt. Opp, Slt. Supp., Mod. Supp., Str. Supp.]

Perception of life in prison scale — — — 0.79 (.05)*

Perception of inmate idleness — — — 0.79 (.18)

Prison punitiveness .89 (.06) 0.61 (.26) — —

Democrat — — 3.02 (2.08) —

Female — — 1.43 (.70) —

Northeast — 2.03 (1.29) 0.15 (.11) —

[Str. Opp., Mod. Opp., Slt. Opp.] versus [Slt. Supp., Mod. Supp., Str. Supp.]

Perception of life in prison scale — — — 0.81 (.04)*

Perception of inmate idleness — — — 1.48 (.17)*

Prison punitiveness 1.26 (.32) 1.09 (.33) — —

Democrat — — 0.79 (.29) —

Female — — 0.83 (.26) —

Northeast — 0.66 (.22) 1.53 (.59) —

[Str. Opp., Mod. Opp., Slt. Opp., Slt. Supp.] versus [Mod. Supp., Str. Supp.]

Perception of life in prison scale — — — 0.80 (.03)***

Perception of inmate idleness — — — 1.24 (.12)*

Prison punitiveness 1.48 (.39) 1.51 (.41) — —

Democrat — — 0.92 (.32) —

Female — — 1.20 (.31) —

Northeast — 1.01 (0.29) 1.42 (.45) —

[Str. Opp., Mod. Opp., Slt. Opp., Slt. Supp., Mod. Supp.] versus [Str. Supp.]

Perception of life in prison scale — — — 0.70 (.03)***

Perception of inmate idleness — — — 1.12 (.11)

Prison punitiveness 1.80 (.61)^† 1.88 (.54)* — —

Democrat — — 1.75 (.60)^† —

Female — — 2.12 (.56)** —

Northeast — 1.73 (.53)^† 1.23 (.43) —

Log pseudolikelihood −585.74 −531.43 −470.01 −455.37

N 365 359 365 363

McFadden’s pseudo R ² .09 .06 .16 .19

Note. Odds ratios with robust standard errors are given in parentheses. Slt. = slightly; Mod. = moderately; Str. = strongly; Opp. = oppose; Supp. = support.

^† p ≤ .10. p ≤ .05. p ≤ .01. p ≤ .001.

All three measures of prison opinions exerted significant effects, although these effects varied from policy to policy. A one-unit increase on the perception of life in prison scale, indicating a stronger perception that life in prison is easy, enjoyable, and/or safe, was associated with an 8% greater likelihood of expressing stronger support for building more prisons but a 7% and a 19% reduction in the likelihood of expressing stronger support for hiring more police or funding prisoner reentry programs, respectively. These effects were consistent across categories of the dependent variables. In contrast, the effect of respondents’ perceptions of life in prison on their support for crime prevention programs violated the parallel regression assumption. Overall, perceptions of life in prison as easy, enjoyable, and/or safe decreased respondents’ likelihood of supporting crime prevention, but the effect was strongest in reducing the likelihood that respondents would “strongly support” crime prevention programs (associated with a 30% reduced likelihood). On the other end of the spectrum, perceptions of life in prison did not significantly affect the likelihood that respondents would choose to “strongly oppose” crime prevention contrasted to all other response options. In between, perceptions of life in prison as easy, enjoyable, and/or safe decreased respondents’ likelihood of expressing moderate support versus moderate opposition by about 20%.

A one-unit increase in the degree to which respondents perceived inmates to spend their time idly was associated with an 18% reduction in the likelihood that a person would express stronger support for building more prisons but a 22% increase in the likelihood of expressing stronger support for prisoner reentry programs. Respondents’ perceptions of inmates’ idleness were not significantly related to their support for hiring police, and the effect on their support for crime prevention policies violated the parallel regression assumption. The unconstrained odds ratios show that perceptions of inmate idleness significantly increased the likelihood that the respondents would support crime prevention, but this effect was restricted to the choice between overall support versus opposition (increasing the likelihood of choosing any of the supportive responses by 48%) as well as the choice between moderate or strong support and the other responses (increasing the likelihood of support by 24%); the variable did not significantly predict the likelihood of choosing either strong support or opposition versus the more moderate choices.

Finally, respondents who judged that the living environment in prisons is not harsh enough were 106% more likely to express greater support for crime prevention policies than respondents who held a less punitive attitude toward prison conditions, but this variable did not significantly affect support for prisoner reentry programs. The effects of this independent variable on support for building more prisons and hiring more police violated the parallel regression assumption. The unconstrained odds ratios show that the perception that the living environment in prisons is not harsh enough only significantly increased the likelihood that respondents would express strong support for more prisons (by 80%) and more police (by 88%) compared to all other responses, which indicates that this perception of prisons only affects the strongest of punitive criminal justice policy preferences, not more weakly held attitudes.

Several measures of “symbolic” attitudes were also related to respondents’ support for punitive and progressive criminal justice policies. Each scale unit increase toward stronger belief in a dispositional attribution of blame was associated with a 15% increased likelihood in expressing greater support for building prisons and a 7% increased likelihood in expressing greater support for hiring more police officers. Each scale unit increase toward stronger belief in a situational attribution of blame was associated with an 11% increased likelihood of expressing greater support for prisoner reentry programs and a 12% increased likelihood of expressing greater support for crime prevention policies. Each scale unit increase toward stronger agreement with the tenets of symbolic racism was associated with a 7% increased likelihood of expressing greater support for building more prisons and a 5% increased likelihood of expressing greater support for hiring police. Lastly, each unit increase on the political ideology scale, moving from liberal to conservative, was associated with a 21% increased likelihood of expressing greater support for crime prevention. The relationships between all of the symbolic attitudinal variables and the dependent variables conformed to the parallel regression assumption.

Finally, several demographic factors were related to people’s crime policy opinions. Democrats were 81% more likely to express stronger support for crime prevention than Republicans, and Independents were 48% more likely to express stronger opposition to building more prisons than Republicans. African Americans were 111% more likely to express stronger support for hiring more police and 154% more likely to express stronger support for crime prevention programs than were respondents of other races (predominantly Whites). Similarly, women were 83% more likely to express stronger support for hiring more police and 104% more likely to express stronger support for crime prevention programs than were men. The relationship between gender and support for prisoner reentry programs is more complex because it did not conform to the parallel regression assumption. The unconstrained odds ratios show that women were much more likely to express both strong opposition (207% increased likelihood) and strong support (112% increased likelihood) for reentry programs than men, but there was no significant difference across gender in regard to moderate or weak attitudes toward reentry. Each additional year of age was associated with a 2% increased likelihood of more strongly supporting both prisoner reentry and crime prevention policies. Lastly, the relationships between region of residence and support for the police and prisoner reentry violated the parallel regression assumption. Residents of northeastern states were 73% more likely to express strong support for hiring more police than residents of the south, and there were no significant regional differences across moderate support or opposition. Residents of the northeast were 81% more likely to say that they “strongly oppose” and 85% more likely to say that they “strongly” or “moderately oppose” prisoner reentry programs than residents of the south, but there were no significant regional differences across levels of support.

Assessing the odds ratios of the variables in these models, we see that the substantive magnitude of the relationships between people’s perceptions of and attitudes toward prisons and their support for punitive and progressive criminal justice policies was often comparable to or larger than the substantive effects of their “symbolic” attitudes and ideologies, often causing 10–20% changes in respondents’ likelihood of choosing particular responses. Furthermore, the relationships between people’s punitive preferences for the prison environment and their support for other criminal justice policies were among the largest in magnitude in this study, changing the likelihood of respondents’ responses by 80% to about 100%. These findings suggest that people’s instrumental attitudes toward prisons are just as influential as their symbolic attitudes when it comes to shaping their broader opinions about criminal justice policy. However, it must be noted that it was people’s demographic characteristics that exerted the substantively largest effects in most of these models, which suggests that the influence of race and gender on people’s crime policy attitudes may be quite significant. However, given the fact that this sample is not perfectly representative of the population, I believe that these effects of race and gender should be interpreted with a degree of caution.

Discussion

It was my purpose in this study to empirically test Roberts and Hough’s (2005) penal escalation hypothesis. I find support for their theoretical argument that people’s global opinions about punishment and criminal justice are significantly related to their judgments about whether or not prisons deliver sufficiently harsh punishment, although the extent and nature of that relationship varied from measure to measure. Some of the present findings are consistent with the directional predictions of their hypothesis. Perceptions of life in prison as easy, enjoyable, and/or safe increased respondents’ support for building more prisons (a punitive response) and decreased their support for prisoner reentry and crime prevention programs (progressive responses). People who judged the living environments in prison to be “not harsh enough” were also significantly more likely to support the punitive options of building more prisons and hiring more police.

Other findings were opposite to those hypothesized at the beginning of this study. I categorized support for hiring more police as a punitive response, but perceptions of life in prison as easy, enjoyable, and/or safe decreased respondents’ support for hiring police. This effect suggests that it may be reductive to classify support for police as a punitive response, which would make sense given that police are capable of protective crime prevention and community building as well as oppressive surveillance and rigid social control. The more that respondents perceived inmates to spend their time idly, the less likely they were to support building more prisons and the more likely they were to support prisoner reentry and crime prevention programs. I interpret these findings as consistent with previous studies which found that people desire for prisons to give inmates the means to rehabilitate themselves, yet largely perceive that prisons fail to accomplish that goal (e.g., Doble, 1987). It seems as if these respondents see little reason to build more prisons if they give inmates nothing to do but sit idly all day. In contrast, these results suggest that respondents who perceive that prisons give inmates insufficient opportunities to rehabilitate themselves instead “place responsibility” for that goal upon reentry programs or programs designed to keep offenders out of prison in the first place.

Two points bear mentioning. First, in a related article (Wozniak, 2015), I find that all three prison perception items are significantly related to people’s preferences for the death penalty over life imprisonment without the possibility of parole; the more strongly that respondents perceived life in prison to be easy, perceived inmates to be idle, and believed that life in prison is not harsh enough, the more likely they were to prefer the death penalty. These findings are consistent with the penal escalation hypothesis, which suggests Roberts and Hough were not incorrect, but rather scholars need to further explore potential differential relationships between different facets/measures of people’s perceptions of prisons and their opinions about a variety of criminal justice policies and practices.

Second, even though the present findings did not unvaryingly conform to the predictions of the penal escalation hypothesis, it is important to recognize that measures of people’s perceptions of prisons were significantly related to their opinions about other aspects of criminal justice even when controlling for other instrumental, symbolic, and ideological measures that are common in this empirical literature. Indeed, the magnitude of the relationships between the prison perception items and the dependent variables was frequently comparable to or greater than the magnitude of the effects of attribution of blame, symbolic racism, and political ideology. These findings suggest that people’s instrumental evaluations of different parts of the justice system may play a more important role in explaining their overall opinions about punishment and criminal justice than scholars have typically acknowledged in this literature. If we have, indeed, underestimated the true importance of instrumental concerns, I argue that we have done so because scholars have all too often relied upon fear of crime and victimization as the sole operational definitions of the instrumentalism construct. The present findings indicate that scholars should explore broader, more diverse ways of measuring people’s instrumental judgments about the functioning of the justice system. Thus, these findings should contribute to both the theory and measurement of public opinion about criminal justice.

In addition to the findings directly related to the penal escalation hypothesis, the pattern of relationships between the symbolic and ideological measures and the dependent variables is notable. Dispositional attribution and symbolic racism are hypothesized to increase punitiveness, and here they affected public opinion about policies that are typically deemed to be punitive in nature (building more prisons and hiring more police), but they were not significantly related to progressive policies (funding reentry and prevention programs). We see the exact opposite pattern for situational attribution. The only dependent variable that was significantly related to both dispositional and situation attribution was the “bipolar” punishment preference item, which encompasses both punitive and progressive sentiments. These results affirm other scholars’ recent arguments that valid measures of the full scope of public opinion about criminal justice must be multidimensional (i.e., separately measure both punitive and progressive policy opinions; Maguire & Johnson, 2015; Mascini & Houtman, 2006; Unnever et al., 2010; for a counter argument, see Pickett & Baker, 2014); here, we see that this advice applies not only to the measurement of dependent variables but also to the attitudinal predictors that explain variation in those different types of variables.

The findings of this study have implications for the politics of crime and punishment. In the heyday of “tough on crime” policy making during the 1980s and 1990s, politicians and pundits in the United States frequently decried “country club conditions” in prisons (see Wozniak, 2014 for a review). Even though that specific rhetoric has waned somewhat in recent years, politicians’ fear of being labeled “soft on crime” has not entirely vanished, and political attack ads centered on criminal justice issues are still fielded in contemporary elections (Litton, 2015; Schwartzapel & Keller, 2015). As Roberts and Hough (2005, p. 292) argued,
… this public view of a congenial prison environment may encourage politicians and correctional officials to further restrict the amenities available to prisoners. It is critical, therefore, that members of the public have a realistic idea of the nature of life in prison.
The present findings suggest that public (mis)perceptions of the nature of life in prison may lead policy makers to intervene in justice policies well beyond the allowance and funding of prison amenities if they perceive that doing so is the only way to convince their constituents that justice system punishments are “up to the task.” As such, I echo Roberts and Hough’s call for a public education campaign to break down popular stereotypes about prisons that minimize the true pains of imprisonment lest those stereotypes fuel broader punitive sentiment (R. Johnson, 2002). A good first step in such a campaign would be for wardens and correctional officers to share their perspectives on prisons with the public. Numerous surveys find that corrections professionals value amenities and leisure time for inmates as tools to maintain order within prisons (DeWine, 1997; Finn, 1996; W. W. Johnson, Bennett, & Flanagan, 1997; Tewksbury & Mustaine, 2005). One can hope that voters’ anger toward “plush” conditions in prisons could be tempered if the professionals who run those prisons explain that amenities are not “perks;” they are tools that can be used to help inmates rehabilitate themselves, and most Americans say that rehabilitation should be the number one goal of incarceration (e.g., Cullen, Pealer, Fisher, Applegate, & Santana, 2002).

In the absence of such “counterframing” by corrections professionals and other reform advocates, these findings indicate that populist demagoguery attacking “country club prisons” could negatively impact public support for contemporary criminal justice reforms. For the first time in decades, numerous states and the federal government are seriously considering sentencing reform bills designed to reduce the size of the nation’s incarcerated population (Beitsch, 2016; Horwitz, 2016). However, despite consistent survey evidence of strong public support for rehabilitation and other progressive criminal justice policies, Wozniak (2016) cautions that limitations in extant research preclude a full and proper understanding of the degree to which public support for criminal justice reform would be vulnerable to new “tough on crime” attacks. This remains an important, open question that needs to be empirically answered.

Limitations

As discussed in the Methods section, this sample overrepresents non-Latino Whites, men, people with more formal education, and the elderly. As such, these data cannot be said to be perfectly representative of the U.S. population. However, it is difficult to say whether or how these response biases may affect the opinions and attitudes under study. Some studies indicate that African Americans oppose punitive punishment more frequently than Whites (Borg, 1998; Chiricos et al., 2004; Unnever & Cullen, 2005), so the overrepresentation of Whites might make my sample more punitive than the population. On the other hand, college-educated respondents are overrepresented, and studies often reveal a negative relationship between education and punitiveness (Chiricos et al., 2004; Green et al., 2006; Unnever & Cullen, 2010). The age and gender disparities in my sample relative to the population are larger in magnitude than race, but past studies find inconsistent or null relationships between these variables and punitiveness (Borg, 1998; Green et al., 2006; Jacoby & Cullen, 1999; Tyler & Weber, 1982). Thus, it is difficult to theorize how these results would have been different had the sample been more perfectly representative of the population, but I will note that the relationships between gender, race, and punitive and progressive policy support observed in these data should be interpreted with a degree of caution, given the overrepresentation of men and the underrepresentation of African Americans in the sample.

Although these data do not possess perfect generalizability, I contend that they make a contribution to our knowledge since no other previous survey measured both people’s perceptions of prisons and their attitudes toward a variety of different punitive and progressive justice policies. I hope that this article will merely be the first step in a broader line of inquiry that revisits Roberts and Hough’s (2005) penal escalation hypothesis by directly testing the relationships between people’s perceptions of many different components of the criminal justice system. Only through replication will we be able to judge the generalizability of the present findings with certainty. In particular, it would be valuable to test the penal escalation hypothesis with new data from European countries. Many European correctional systems are held up as paragons of human rights in comparison to U.S. prisons (Benko, 2015; Chammah, 2015). The fact that the findings of Brillon et al. (1984) from Canada and Glanz (1994) from South Africa mirror many of the relationships uncovered in these data from the United States suggest that the penal escalation hypothesis is valid cross nationally, but this proposition should be tested in countries with low incarceration rates that emphasize the civil and human rights of inmates.

It is also important to note that this study measures global attitudes. The prison perception questions did not ask respondents to distinguish between different types of prisons (i.e., minimum vs. maximum security) or prisons in different areas of the country. It is quite plausible to imagine a sophisticated respondent saying, “Well, some prisons are dangerous, but others are safe. It depends.” As such, these measures are inherently reductive. It is perhaps more accurate to think of these data more as people’s stereotypes of prisons rather than their perceptions. However, people form opinions about politics and public policy “off the cuff,” using whatever impressions jump to mind, all the time (Zaller, 1992). The fact that many of these opinions do not match “empirical reality” hardly mitigates their influence when it comes to electoral politics (Beckett, 1997; Brown, 2011). Future research should test whether or not people meaningfully distinguish between different types of prisons when forming their opinions, and if so, how those different perceptions relate to their opinions about other facets of the justice system.

Because of the fact that these data are cross sectional, I cannot conclusively establish temporal order between people’s perceptions of prisons and their opinions about other facets of criminal justice policy and practice. As such, these findings are correlational, not causal, in nature. Future research could use experimental methods to prime respondents with pictures of prison environments that are more or less austere to determine whether or not perceptions of prisons as insufficiently harsh cause people’s punitive attitudes about other facets of the justice system to change.

Conclusion

Analyzing data from an original public opinion survey administered to a randomly selected sample of adults in the United States, I find that people’s perceptions of life and conditions in prison are significantly related to their opinions about other, punitive and progressive aspects of criminal justice policy and practice. These results partially support Roberts and Hough’s (2005) penal escalation hypothesis and highlight the need for scholars to reevaluate the importance of instrumental versus symbolic factors when attempting to explain variation in people’s opinions about punishment and justice. Future research should continue to explore the relationships between people’s perceptions of numerous different parts of the criminal justice system rather than studying them in isolation.

Characteristic	Survey Data (%)	Census Data (%)
White/Caucasian	76.9	72.4
Black/African American	9.8	12.6
Latino	5.0	16.3
Other race	8.1	12.1
Male	60.5	49.2
Age 18–64	63.3	82.1^a
Age 65 and older	36.7	17.9^a
College, graduate, or advance professional degree	43.3	38.2

Variables	Punishment Preference
Constrained factors
Perception of inmate idleness	1.05 (.09)
Prison punitiveness	1.40 (.32)
Dispositional attribution of blame	1.13 (.04)***
Situational attribution of blame	0.92 (.03)**
Symbolic racism	1.06 (.03)*
Perceived fairness of criminal justice system	1.03 (.04)
Political ideology (conservative)	1.26 (.13)*
News consumption	1.01 (.02)
Fear of crime	1.06 (.30)
Victimization	0.85 (.20)
Democrat	0.78 (.24)
Independent	0.62 (.18)
African American	0.73 (.35)
Female	1.00 (.24)
Age	1.00 (.01)
Education	0.90 (.07)
Household income	0.95 (.07)
Northeast	0.67 (.21)
Central	0.67 (.17)
West	0.76 (.22)
Unconstrained factors
[Much BTFSP] versus [Mod. BTFSP, Slt. BTFSP, Slt. BTPC, Mod. BTPC, Much BTPC]
Perception of life in prison scale	1.27 (.08)***
[Much BTFSP, Mod. BTFSP] versus [Slt. BTFSP, Slt. BTPC, Mod. BTPC, Much BTPC]
Perception of life in prison scale	1.25 (.06)***
[Much BTFSP, Mod. BTFSP, Slt. BTFSP] versus [Slt. BTPC, Mod. BTPC, Much BTPC]
Perception of life in prison scale	1.22 (.05)***
[Much BTFSP, Mod. BTFSP, Slt. BTFSP, Slt. BTPC] versus [Mod. BTPC, Much BTPC]
Perception of life in prison scale	1.15 (.04)***
[Much BTFSP, Mod. BTFSP, Slt. BTFSP, Slt. BTPC Mod. BTPC] versus [Much BTPC]
Perception of life in prison scale	1.03 (.04)
Log pseudolikelihood	−528.40
N	365
McFadden’s pseudo R ²	.17

Variables	More Prisons	More Police	Prisoner Reentry	Crime Prevention
Constrained factors
Perception of life in prison scale	1.08 (.03)*	0.93 (.03)*	0.81 (.03)***	(unconstrained)
Perception of inmate idleness	0.82 (.06)*	0.95 (.08)	1.22 (.10)*	(unconstrained)
Prison punitiveness	(unconstrained)	(unconstrained)	1.01 (.25)	2.06 (.52)**
Dispositional attribution of blame	1.15 (.04)***	1.07 (.04)*	1.02 (.04)	1.04 (.04
Situational attribution of blame	1.01 (.03)	1.02 (.03)	1.11 (.03)***	1.12 (.04)*
Symbolic racism	1.07 (.03)*	1.05 (.03)*	1.00 (.03)	0.97 (.02)
Perceived fairness of criminal justice system	1.01 (.04)	0.98 (.03)	0.96 (.03)	0.97 (.04)
Political ideology (conservative)	1.10 (.12)	1.13 (.10)	0.98 (.10)	1.21 (.12)^†
News consumption	0.97 (.02)	1.05 (.02)	1.03 (.02)	1.01 (.02)
Fear of crime	1.28 (.30)	1.18 (.27)	0.68 (.19)	0.74 (.20)
Victimization	0.93 (.20)	1.31 (.28)	1.28 (.30)	0.92 (.23)
Democrat	0.61 (.19)	1.22 (.39)	(unconstrained)	1.81 (.58)^†
Independent	0.52 (.14)*	0.90 (.25)	0.95 (.25)	1.48 (.39)
African American	0.88 (.34)	2.11 (.81)^†	1.94 (1.02)	2.54 (1.44)^†
Female	1.01 (.22)	1.83 (.43)**	(unconstrained)	2.04 (.51)**
Age	1.00 (.01)	1.00 (.01)	1.02 (.01)*	1.02 (.01)*
Education	1.03 (.08)	1.08 (.08)	0.98 (.07)	1.09 (.08)
Household income	1.06 (.08)	1.02 (.07)	0.96 (.06)	1.00 (.07)
Northeast	1.37 (.38)	(unconstrained)	(unconstrained)	0.86 (.25)
Central	1.42 (.39)	1.26 (.32)	0.85 (.24)	1.03 (.29)
West	1.05 (.27)	0.78 (.24)	0.71 (.20)	1.47 (.46)
Unconstrained factors
[Str. Opp.] versus [Mod. Opp., Slt. Opp., Slt. Supp., Mod. Supp., Str. Supp.]
Perception of life in prison scale	—	—	—	0.91 (.08)
Perception of inmate idleness	—	—	—	0.85 (.22)
Prison punitiveness	.65 (.22)	0.69 (.46)	—	—
Democrat	—	—	2.23 (1.65)	—
Female	—	—	3.07 (1.92)^†	—
Northeast	—	1.19 (1.07)	0.19 (.14)*	—
[Str. Opp., Mod. Opp.] versus [Slt. Opp, Slt. Supp., Mod. Supp., Str. Supp.]
Perception of life in prison scale	—	—	—	0.79 (.05)***
Perception of inmate idleness	—	—	—	0.79 (.18)
Prison punitiveness	.89 (.06)	0.61 (.26)	—	—
Democrat	—	—	3.02 (2.08)	—
Female	—	—	1.43 (.70)	—
Northeast	—	2.03 (1.29)	0.15 (.11)**	—
[Str. Opp., Mod. Opp., Slt. Opp.] versus [Slt. Supp., Mod. Supp., Str. Supp.]
Perception of life in prison scale	—	—	—	0.81 (.04)***
Perception of inmate idleness	—	—	—	1.48 (.17)***
Prison punitiveness	1.26 (.32)	1.09 (.33)	—	—
Democrat	—	—	0.79 (.29)	—
Female	—	—	0.83 (.26)	—
Northeast	—	0.66 (.22)	1.53 (.59)	—
[Str. Opp., Mod. Opp., Slt. Opp., Slt. Supp.] versus [Mod. Supp., Str. Supp.]
Perception of life in prison scale	—	—	—	0.80 (.03)***
Perception of inmate idleness	—	—	—	1.24 (.12)*
Prison punitiveness	1.48 (.39)	1.51 (.41)	—	—
Democrat	—	—	0.92 (.32)	—
Female	—	—	1.20 (.31)	—
Northeast	—	1.01 (0.29)	1.42 (.45)	—
[Str. Opp., Mod. Opp., Slt. Opp., Slt. Supp., Mod. Supp.] versus [Str. Supp.]
Perception of life in prison scale	—	—	—	0.70 (.03)***
Perception of inmate idleness	—	—	—	1.12 (.11)
Prison punitiveness	1.80 (.61)^†	1.88 (.54)*	—	—
Democrat	—	—	1.75 (.60)^†	—
Female	—	—	2.12 (.56)**	—
Northeast	—	1.73 (.53)^†	1.23 (.43)	—
Log pseudolikelihood	−585.74	−531.43	−470.01	−455.37
N	365	359	365	363
McFadden’s pseudo R ²	.09	.06	.16	.19

Footnotes

Declaration of Conflicting Interests

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

Funding

The author disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The collection of the data analyzed in this study was funded by dissertation research grants from American University and the Gill Family Foundation.

Notes

References

Allison

P. D.

(2002). Missing data. Thousand Oaks, CA: Sage.

American National Election Studies and Stanford University. (2015). ANES Time Series Cumulative Data File (1948–2012). ICPSR08475-v15. Ann Arbor, MI: Interuniversity Consortium for Political and Social Research. doi:http://doi.org/10.3886/ICPSR08475.v15

Applegate

B. K.

(2001). Penal austerity: Perceived utility, desert, and public attitudes toward prison amenities. American Journal of Criminal Justice, 25, 253–268.

Baker

(1996). From ‘making bad people worse’ to ‘prison works’: Sentencing policy in England and Wales in the 1990s. Criminal Law Forum, 7, 639–671.

Beckett

(1997). Making crime pay: Law and order in contemporary American politics. New York, NY: Oxford.

Beitsch

(2016, 1 28). States at a crossroads on criminal justice reform. Stateline/Pew Center. Retrieved from http://www.pewtrusts.org/en/research-and-analysis/blogs/stateline/2016/01/28/states-at-a-crossroads-on-criminal-justice-reform

Benko

(2015, 3 26). The radical humaneness of Norway’s Halden Prison. New York Times Magazine. Retrieved from http://www.nytimes.com/2015/03/29/magazine/the-radical-humaneness-of-norways-halden-prison.html?_r=0

Bidinotto

R. J.

(1994, 11). Must our prisons be resorts? Reader’s Digest, 145, 65–71.

Bobo

L. D.

Johnson

(2004). A taste for punishment: Black and White Americans’ views on the death penalty and the war on drugs. Du Bois Review, 1, 151–180.

10.

Borg

M. J.

(1998). Vicarious homicide victimization and support for capital punishment: A test of Black’s theory of law. Criminology, 36, 537–568.

11.

Brillon

Louis-Guerin

Lamarche

M. C.

(1984). Attitudes of the Canadian public toward crime policies. Montreal, Canada: University of Montreal.

12.

Brown

(2011). Constructing the public will: How political actors in New York State construct, assess, and use public opinion in penal policy making. Punishment and Society, 13, 424–450.

13.

Chammah

(2015, 9 25). Prison without punishment. The Marshall Project. Retrieved from https://www.themarshallproject.org/2015/09/25/prison-without-punishment?utm_medium=email&utm_campaign=newsletter&utm_source=opening-statement&utm_term=newsletter-20150925-281

14.

Chiricos

Welch

Gertz

(2004). Racial typification of crime and support for punitive measures. Criminology, 42, 359–389.

15.

Cullen

F. T.

Pealer

J. A.

Fisher

B. S.

Applegate

B. K.

Santana

S. A.

(2002). Public support for correctional rehabilitation in America: Change or consistency? In Roberts

J. V.

Hough

(Eds.), Changing attitudes to punishment: Public opinion, crime, and justice (pp. 128–147). Portland, OR: Willan.

16.

Curtin

Presser

Singer

(2005). Changes in telephone survey nonresponse over the past quarter century. Public Opinion Quarterly, 69, 87–98.

17.

de Leeuw

de Heer

(2002). Trends in household survey nonresponse: A longitudinal and international comparison. In Groves

R. M.

Dillman

D. A.

Eltinge

J. L.

Little

R. J. A.

(Eds.), Survey nonresponse (pp. 41–54). New York, NY: Wiley.

18.

DeWine

(1997). Public opinion and corrections: A need to be proactive. Corrections Management Quarterly, 1, 6–9.

19.

Dillman

D. A.

(1991). The design and administration of mail surveys. Annual Review of Sociology, 17, 225–249.

20.

Dillman

D. A.

Smyth

J. D.

Christian

L. M.

(2009). Internet, mail, and mixed-mode surveys: The tailored design method. Hoboken, NJ: John Wiley.

21.

Doble

(1987). Crime and punishment: The public’s view. New York, NY: Public Agenda Foundation for the Edna McConnell Clark Foundation.

22.

Duffee

Ritti

R. R.

(1977). Correctional policy and public values. Criminology, 14, 449–460.

23.

Durkheim

(1964). The division of labor in society (Trans. Simpson

). New York, NY: Free Press of Glencoe. (Original work published 1893)

24.

Enns

P. K.

(2014). The public’s increasing punitiveness and its influence on mass incarceration in the United States. American Journal of Political Science, 58, 857–872.

25.

Finn

(1996). No-frills prisons and jails: A movement in flux. Federal Probation, 60, 35–44.

26.

Fisher

J. E.

Mason

R. L.

(1981). The analysis of multicollinear data in criminology. In Fox

J. A.

(Ed.), Methods in quantitative criminology (pp. 99–125). New York, NY: Academic Press.

27.

Fox

(1991). Regression diagnostics. Newbury Park, CA: Sage.

28.

Frost

N. A.

(2010). Beyond public opinion polls: Punitive public sentiment and criminal justice policy. Sociology Compass, 4/3, 156–168.

29.

Garland

(2001). The culture of control: Crime and social order in contemporary America. Chicago, IL: University of Chicago Press.

30.

Glanz

(1994). The South African public’s attitudes toward imprisonment and the release of offenders. Acta Criminologica, 7, 64–79.

31.

Green

E. G. T.

Staerkle

Sears

D. O.

(2006). Symbolic racism and whites’ attitudes towards punitive and preventive crime policies. Law and Human Behavior, 30, 435–454.

32.

Groves

R. M.

(2006). Nonresponse rates and nonresponse bias in household surveys. Public Opinion Quarterly, 70, 646–675.

33.

Groves

R. M.

Fowler

F. J.

Couper

M. P.

Jepkowski

J. M.

Singer

Tourangeau

(2004). Survey methodology. Hoboken, NJ: Wiley-Interscience.

34.

Groves

R. M.

Peytcheva

(2008). The impact of nonresponse rates on nonresponse bias: A meta-analysis. Public Opinion Quarterly, 72, 167–189.

35.

Horwitz

(2016, 2 10). Utah’s Republican Sen. Mike Lee: Criminal justice reform legislation has momentum. Washington Post. Retrieved from https://www.washingtonpost.com/news/post-nation/wp/2016/02/10/utahs-republican-sen-mike-lee-criminal-justice-reform-legislation-has-momentum/

36.

Hurwitz

Peffley

(2005). Playing the race card in the post-Willie Horton era. Public Opinion Quarterly, 69, 99–112.

37.

Jacoby

J. E.

Cullen

F. T.

(1999). The structure of punishment norms: Applying the Rossi-Berk model. The Journal of Criminal Law and Criminology, 89, 245–312.

38.

Johnson

(2002). Hard time: Understanding and reforming the prison (3rd ed.). Belmont, CA: Wadsworth.

39.

Johnson

(2008). Racial prejudice, perceived injustice, and the Black-White gap in punitive attitudes. Journal of Criminal Justice, 36, 198–206.

40.

Johnson

W. W.

Bennett

Flanagan

T. J.

(1997). Getting tough on prisoners: Results from the national corrections executive survey, 1995. Crime and Delinquency, 43, 24–41.

41.

Keeter

Miller

Kohut

Groves

R. M.

Presser

(2000). Consequences of reducing nonresponse in a national telephone survey. Public Opinion Quarterly, 64, 125–148.

42.

King

Maruna

(2009). Is a conservative just a liberal who has been mugged? Exploring the origins of punitive views. Punishment and Society, 11, 147–169.

43.

Krysan

(1998). Privacy and the expression of white racial attitudes: A comparison across three contexts. Public Opinion Quarterly, 62, 506–544.

44.

Lenz

(2002). “Luxuries” in prison: The relationship between amenity funding and public support. Crime and Delinquency, 48, 499–525.

45.

Litton

(2015, 10 15). David Vitter ad accuses John Bel Edwards of plan to release ‘thugs’ from prison. Times-Picayune. Retrieved from http://www.nola.com/politics/index.ssf/2015/10/expert_david_vitter_ad_accusin.html#incart_m-rpt-1

46.

Long

J. S.

(1997). Regression models for categorical and limited dependent variables. Thousand Oaks, CA: Sage.

47.

Maguire

E. R.

Johnson

(2015). The structure of public opinion on crime policy: Evidence from seven Caribbean nations. Punishment and Society, 17, 502–530.

48.

Mascini

Houtman

(2006). Rehabilitation and repression: Reassessing their ideological embeddedness. British Journal of Criminology, 46, 822–836.

49.

National Research Council. (2014). The growth of incarceration in the United States. Washington, DC: The National Academies Press.

50.

Nicholson-Crotty

Peterson

D. A. M.

Ramirez

M. D.

(2009). Dynamic representation(s): Federal criminal justice policy and an alternative dimension of public mood. Political Behavior, 31, 629–655.

51.

Peffley

Hurwitz

(2010). Justice in America: The separate realities of Blacks and Whites. New York, NY: Cambridge University Press.

52.

Pew Research Center. (2012). Assessing the representativeness of public opinion surveys. Washington, DC: Author.

53.

Pickett

J. T.

Baker

(2014). The pragmatic American: Empirical reality or methodological artifact? Criminology, 52, 195–222.

54.

Pickett

J. T.

Chiricos

Gertz

(2014). The racial foundations of whites’ support for child saving. Social Science Research, 44, 44–59.

55.

Ramirez

M. D.

(2014). Racial discrimination, fear of crime, and variability in blacks’ preferences for punitive and preventative anti-crime policies. Political Behavior. Epud ahead of print. doi:10.1007/s11109-014-9285 -1

56.

Roberts

J. V.

Hough

(2005). The state of the prisons: Exploring public knowledge and opinion. Howard Journal of Criminal Justice, 44, 286–306.

57.

Roberts

J. V.

Stalans

L. J.

Indermaur

Hough

(2003). Penal populism and public opinion: Lessons from five countries. New York, NY: Oxford University Press.

58.

Schwartzapel

Keller

(2015). Willie Horton revisited. The Marshall Project, 13 May. Retrieved from https://www.themarshallproject.org/2015/05/13/willie-horton-revisited?ref=hp-4-121

59.

Simon

(2014). Mass incarceration on trial: A remarkable court decision and the future of prisons in America. New York, NY: The New Press.

60.

StataCorp. (2013). Stata Statistical Software: Release 13. College Station, TX: StataCorp LP.

61.

Tarman

Sears

D. O.

(2005). The conceptualization and measurement of symbolic racism. Journal of Politics, 67, 731–761.

62.

Tewksbury

Mustaine

E. E.

(2005). Insiders’ views of prison amenities: Beliefs and perceptions of correctional staff members. Criminal Justice Review, 30, 174–188.

63.

Tyler

T. R.

Boeckmann

R. J.

(1997). Three strikes and you’re out, but why? The psychology of public support for punishing rule breakers. Law and Society Review, 31, 237–265.

64.

Tyler

T. R.

Weber

(1982). Support for the death penalty: Instrumental response to crime, or symbolic attitude? Law and Society Review, 17, 21–45.

65.

Unnever

J. D.

Cochran

J. K.

Cullen

F. T.

Applegate

B. K.

(2010). The pragmatic American: Attributions of crime and the hydraulic relation hypothesis. Justice Quarterly, 27, 431–457.

66.

Unnever

J. D.

Cullen

F. T.

(2005). Executing the innocent and support for capital punishment: Implications for public policy. Criminology and Public Policy, 4, 3–38.

67.

Unnever

J. D.

Cullen

F. T.

(2007). The racial divide in support for the death penalty: Does white racism matter? Social Forces, 85, 1281–1301.

68.

Unnever

J. D.

Cullen

F. T.

(2010). The social sources of Americans’ punitiveness: A test of three competing models. Criminology, 48, 99–130.

69.

Walmsley

(2013). World prison population list (10th ed.). London, England: International Centre for Prison Studies.

70.

Williams

(2006). Generalized ordered logit/partial proportional odds models for ordinal dependent variables. The Stata Journal, 6, 58–82.

71.

Wozniak

K. H.

(2012). The effect of exposure to political rhetoric on public opinion about criminal justice (Doctoral dissertation). Retrieved from ProQuest Dissertations and Theses (Order Number 3523746).

72.

Wozniak

K. H.

(2014). American public opinion about prisons. Criminal Justice Review, 39, 305–324.

73.

Wozniak

K. H.

(2015). The relationship between perceptions of prison and support for the death penalty versus life without parole. Journal of Crime and Justice. Online ahead of print. doi:10.1080/0735648X.2015.1050598

74.

Wozniak

K. H.

(2016). Public opinion and the politics of criminal justice policy making: Reasons for optimism, pessimism, and uncertainty. Criminology & Public Policy, 15, 179–186.

75.

Zaller

J. R.

(1992). The nature and origins of mass opinion. New York, NY: Cambridge.