Punitive Penal Preferences and Support for Welfare: Applying the Governance of Social Marginality Thesis to the Individual Level

Transcription

1 Punitive Penal Preferences and Support for Welfare: Applying the Governance of Social Marginality Thesis to the Individual Level Ashley T. Rubin Pre-print; Published in Punishment & Society (2011), Vol. 13, No. 2, pp INTRODUCTION A favorite topic among law and society scholars over the last few decades has been explaining the shift in the United States, the United Kingdom, and other developed countries away from rehabilitative penal policies toward more punitive policies that changed the way we punish, whom we punish, and for how long we punish them. Many have theorized about or shown correlations between punitive penal policies or trends, on the one hand, and, on the other, crime rates, whether property, violent, or homicide (e.g., Zimring and Hawkins 1997); the influence of race (Beckett and Western 2001, Parenti 2001, Wacquant 2001, Jackson and Carroll 1981); attitudes and responses to difference (Young 1999); socio-economic conditions and political frameworks common to neo-liberal states (Cavadino and Dignan 2006, see also Rusche and Kirchheimer 1939, Downes 2001, Sutton 2004, and Lappi-Seppälä 2008); legal cultures (Lappi-Seppälä 2008); confidence in various branches of government (e.g., the legislature and courts) as well as the salience of crime (Zimring and Johnson 2006, Costelloe, et al., 2009); and various combinations of these (Tonry 2009). Other scholars have looked to social solidarity based elements: Taylor and Boeckmann (1997) find correlations between popular support for three strikes laws and fear of declining moral cohesion, often exacerbated by growing diversity in society (cf. Putnam 2007). Likewise, Barker (2007) has shown that retributive policies are positively correlated with social polarization and negatively correlated with high 1

2 levels of social trust. A large body of scholarship has also examined political discourse, governance choices, rhetoric, and their respective influence on public opinion and penal trends (see, e.g., Beckett 1997; Simon 2007). One subset of this literature includes scholars who have empirically shown a qualitative or quantitative connection between penal trends and trends in welfare policy across the twentieth century, suggesting that penal severity is inversely correlated with assistance to the poor (welfare). This situation is variously recognized as penal welfarism, which describes the rehabilitative penal policies and welfare state of the United Kingdom and the United States in the middle of the twentieth century (Garland 1985, 1990), or the governance of social marginality, referring to the growing punitiveness of welfare and penal policies in the late twentieth century (Beckett and Western 2001). While we need not dispute theories suggesting the effect of crime rates, socio-economic conditions, racism, difference, diversity, confidence in government, moral and social cohesion, and political discourse on penal trends indeed, it may be that each of these together influence penal policy and general social punitiveness we shall focus here on the welfare-penality connection. Scholars have described a structural relationship between penal and welfare policies in which the penal system increasingly fulfills the role previously played by the welfare state, while state aid to the poor grows increasingly punitive (see Simon 2007). This welfare-penality pairing is by now well documented and often attributed to a variety or combination of social, economic, and political changes, including most notably the risks and consequences of modern deregulated (neo-liberal) economies, race relations or responses to the Civil Rights Movement, and levels of social solidarity, social trust, and confidence in government. Working on the national scale, Garland (2001, 199) explains that prisons are a ready-made penal solution to a new problem of social and economic exclusion while ostensibly serving other, more legitimate 2

3 ends (i.e., crime control). More specifically, they serve a newly necessary function in the workings of late modern, neo-liberal societies: the need of a civilized and constitutional means of segregating the problem populations created by today s economic and social arrangements, namely the poor excluded from an unregulated economy and left unaided by modern welfare programs, especially young urban minority males. Similarly, Wacquant (2001, 97) explains, the American prison has become an instrument for the management of the dispossessed and dishonored groups. [T]he increasing use of imprisonment to shore up caste division in American society partakes of a broader upsizing of the penal sector of the state which, together with the drastic downsizing of its social welfare sector, aims at imposing desocialized wage labor as a norm of citizenship for the deskilled fractions of the postindustrial working class. Thus, we see prisons adopt managerial styles of incarceration in the 1980s and 1990s in response to their role as keepers of the underclass (Feeley and Simon 1992). This trend is also apparent on the state level. Beckett and Western (2001) demonstrate a link between states incarceration rates and the generosity of their welfare programs and black populations of those states, while controlling for other factors, although it is not fully manifested until the 1990s. They characterize this development as a change during the 1980s and 1990s in what they call a single policy regime [of which welfare and penal policy are two components] aimed at the problems associated with deviance and marginality (2001, 44). Specifically, the change is one in the policy regime s commitment to social marginality (poverty and criminality) from inclusion and integration (characteristic of the early and middle parts of the twentieth century) to exclusion and separation (2001, 44). 3

4 However, this penal-welfare pairing is restricted neither to punitiveness in both welfare and penal policy, nor to the United States. Garland (2001) has noted similar changes in the United Kingdom, although the changes are less severe than in the United States. Both Downes and Hansen (2006) and Lappi-Seppälä (2008) document late-twentieth-century inverse correlations between of punitive penal policies (measured by incarceration rates and/or penal expenditures as a proportion of gross domestic product) and generous welfare policies in a variety of other (mostly European) countries. Given the diversity of the international setting, some of these nations reveal lenience in penal policy and generousness in welfare policies reminiscent of mid-century Anglo penal-welfarism described in Garland (1985). As Lappi- Seppälä (2008, 314) explains, in countries like Scandinavia, [t]he welfare state has made it possible to develop workable alternatives to imprisonment. Welfare and social equality have promoted trust and legitimacy, which facilitate compliance with norms based on legitimacy and acceptance (instead of sentence severity). These characteristics reduce political pressures to resort to symbolic penal gestures. Low imprisonment rates are by-products of consensual, corporatist, and negotiative political cultures. Thus, we see a correlation between punitiveness in penal and welfare policies and practices, whether both lenient or both punitive, across the mid- and late twentieth century around the world. The correlation is by now considered firmly established (Greenberg 2001, 81). However, the attention paid primarily to the structural coherence between the welfare and penal policies and practices generally treats the individual as only an implicit actor. Scholars investigating the role of crime rates, crime salience, diversity, and levels of trust in institutions, social solidarity or social trust, economic insecurity, or other post-modern anxieties to explain 4

5 policy trends leave unstated the role of private individuals, whose views define or are affected by these elements. As individuals are not the direct makers or enforcers of policy, their exclusion generally makes sense. However, the place, and even influence, of the individual and public opinion are at the heart of many explanations for changes in penal and welfare policy. While dismissing the importance of public opinion in explaining penal trends (2009, 379), Tonry has recently noted that the arrangement of American political and criminal justice systems make policy particularly susceptible to public opinion, especially when it demands more severity. He explains, The structure of American government was meant to tie officials closely to community needs and beliefs, and democratic ideology celebrated the importance and influence of public opinion, even if it was ill informed, mercurial or mean-spirited. Constitutional draftsmen worried about the dangers of mobocracy but, with local notable exceptions, the problem did not fully take shape until late in the 20th century when ubiquitous electronic and broadcast media meant that detailed reports of horrible incidents anywhere, and ensuring emotionalism, could sweep across an entire continent (2009, 385-6). This is true as well of on-the-ground criminal justice work. Tonry (2009, 385) points out that the American criminal justice system is uniquely responsive to the public: If the public is anxious about crime or angry at criminals, or if particular cases become notorious, there is nothing to stop prosecutors from seeking personal political benefit by posturing before public opinion. Judges are elected in most states and know that highly unpopular decisions can lead to their defeat. Indeed, much of the literature focusing on the twin changes in penal and welfare policy towards punitiveness attribute their association to the characterization of the poor and the criminal in such a way that makes punitive treatment towards these groups easier options for the 5

6 general public (Beckett 1997; Beckett and Western 2001, 46-47). Political discourse is the means by which policy entrepreneurs gain approval and pass agendas that create coherent policy regimes like the penal-welfare regime. As Garland (2001) explains, A few decades ago public opinion functioned as an occasional brake on policy initiatives: now it operates as a privileged source (2001, 13). For a policy, let alone a policy regime, to exist there must be some level of support among voters for laws to pass and legislators to stay in office. It stands to reason that public approval is at least part of the purpose of politicians pessimistic rhetoric promoting the exclusion of the socially marginal since the 1960s (Beckett 1997; Beckett and Western 2001, 46; Garland 1990, 53-73): that rhetoric is intended to convince the voting public to support, directly or indirectly, the policies specified. To make the reigning theory more explicit: (1) For a variety of reasons, politicians find crime and poverty too troubling (and dangerous to their careers) to manage with previously traditional means (rehabilitation, welfare), and need to deploy new policies to manage these social problems. (2) To promote these new policies, politicians deploy a new characterization of the socially marginal as undeserving and responsible for their lot. (3) This new characterization convinces 1 or motivates the public to accept and (presumably) support politically the logical (according to the new descriptions) solutions to the problems of social marginality, viz., excessively punitive penal policies and less generous welfare policies. (4) The public does so (votes for these programs, does not vote out of office those who support them, votes out of office those who do not, etc.), these policies are adopted, and a relationship between welfare and penal policies becomes apparent. This article examines the third step in this process. As noted above, the scholarship in this area tends to examine why governance priorities changed (first step), the change in political rhetoric (second step), and the policy developments that followed (fourth 6

7 step). Given the central yet implicit role of public opinion in this scholarship, it seems useful to explore its role in the policy regime change of the late twentieth century. While the structural coherence between welfare and penal policy may be an accurate account of national- and statelevel penal and welfare policy over time, it is valuable to know the extent to which it is also reflective of individual-level support for such policies over time given the public s tacit consent to these developments. Such knowledge may improve our understanding of the structural or state-level symbiosis between welfare and penal policy. Indeed, in two recent exceptions to the general neglect of the role of public opinion in the literature, we find suggestions that individual-level support is in fact an important step in the process of regime changes. Most recently, Lappi-Seppälä (2008, 351) has demonstrated a positive correlation between measures of public punitiveness and imprisonment rates in his study of European countries. He explains, Public perceptions and concern about crime shape how people treat crime and criminals. Fear and public sentiments may form an essential part of those explanations that combine penal changes, the rise of a culture of fear, and postmodern angst. Few of these assumptions have been empirically tested (Lappi-Seppälä 2008, 348, though see also Roberts 1992). Similarly, based on her examination of political and penal developments in California, Washington, and New York, Barker (2006) has suggested that citizen involvement in the political process can have very real effects on incarceration trends. Importantly, she notes that it is not the state alone that accounts for imprisonment variation, but rather the dynamic interaction between the state and civil society. In other words, I claim that the democratic process itself must be incorporated into our accounts of punishment (2006, 25). We take these studies to suggest the relevance of public opinion to the penal-welfare/social marginality scholarship. Additionally, however, we note that understanding public opinion and its 7

8 acquiescence to or influence on public policy (even when following the lead of political rhetoric) is particularly valuable in the U.S. context, as public opinion in the United States holds greater sway than in it does in other countries (Tonry 2009, ). We seek to answer three questions about the American welfare and penality complex. First, is there a correlation in punitiveness towards welfare and penal policy on the individual level? Second, does the correlation persist across time, or is it only a relatively recent trend that surfaced around the same time as state-level manifestations of the punitiveness pairing did? Third, what is the nature (i.e., punitive, welfarist) of this correlation over time, and does it reflect the larger trend toward cross-policy punitiveness documented in the literature? It is important to note that we are testing neither the correlation between public opinion and policy, as in Lappi- Seppälä (2008), nor the influence of individual preferences for punitiveness on policy punitiveness. Instead, we are simply investigating whether opposition to welfare and support for punitive penal policy are correlated on the individual level in general and across time during the same period as a correlation began developing on the state and national levels. The effect of any significant findings in this study will be mediated by the level of democracy in a society and the structural vulnerability of its penal or welfare policy and practice (see Zimring and Johnson 2006, 277): where citizens preferences have little effect on policy and judicial appointments, the effect of their punitiveness or lenience preferences has little relevance in determining state penal policy; where, however, structural vulnerability is high, or there is direct democracy (as is the case with California s initiative process), individual-level punitiveness should be highly relevant as it may indeed affect policy (see Barker 2006 for an alternative, counterintuitive role of public opinion on penal policy). Naturally, existing democracies fall somewhere in the middle of 8

9 this dyad; therefore, the significance of this study will vary depending on where on the spectrum a particular state lands. This article thus examines the nature of the link between welfare and penal punitiveness in the United States, both in general and over time, by changing the unit of analysis from states (and their penal policies and socio-cultural trends) to the individual (and his beliefs). As Beckett and Western (2001) is the most rigorous quantitative study of the U.S. welfare-penality symbiosis, we use their study as our main point of reference. We find that there is a strong correlation between opposition to welfare and various measures of punitiveness regarding penal policy. This relationship persists across time, though with significant variation. In general, the level of correlation declines toward the mid- to late 1970s and in the early and mid-1990s before climbing in the early 1980s and late 1990s. The resultant peaks generally correspond to the Nixon administration, the second Reagan administration, the second Clinton administration, and the second G.W. Bush administration. Moreover, we find that much of the correlation across time is due to cross-policy punitiveness, and that cross-policy welfarism is responsible for only a small (though growing since the late 1990s) portion of the correlation. This punitiveness also has peaks and troughs, but these do not correspond to the peaks and troughs in the level of overall correlation. Instead, we find peaks around the Carter and first Clinton administrations, and troughs in the George H.W. Bush and second George W. Bush administrations. We reflect briefly on the linkages between welfare and penality before discussing potential explanations for the difference between state-level and individual-level correlations. We close by suggesting that understanding the connection between punitive penal policy and punitive welfare policy on the individual level will better enable us to understand the mechanism behind this linkage at the state level, even if individual-level data do not correlate well with state policy. 9

10 METHOD Respondents This study is based on an analysis of data gathered in the General Social Survey (GSS) from 1972 to In these years, the GSS team was able to gather data on over 51,000 individuals. In 1972 to 1974, the data were collected through a modified probability sample, and since 1975, households were chosen based on a full-probability sampling of households (GSS 2007, Appendix A). GSS data consisted mostly of responses to in-person interviews, with Computer Assisted Personal Interviews (CAPI) since Table 1 describes general characteristics of GSS respondents in these years. Questionnaire Variables In this study, we focus on three opinions to determine the relationship between punitiveness and support for welfare. Punitiveness is determined by (1) one s belief that the courts do not treat criminals harshly enough 2 and (2) one s support for capital punishment. While popular opinion about how punitive courts are towards criminals is often inaccurate (see, e.g., Roberts and Doob 1989; Roberts 1992; Zimring and Johnson 2006), we assume that punitive individuals will be more prone to believing, and responding, that courts are not harsh enough. Both of these variables fall under Frost s (2008) penal intensity based definition of punitiveness: harsher treatment of criminals typically means longer sentences, though it may also include punishing one who may not be punished otherwise, and capital punishment is a particularly severe penal sanction. Additionally, incorporating opinion of courts treatment of 10

11 criminals, rather than relying solely on capital punishment support, helps control for race-based differences in support for capital punishment (see Unnever, et al., 2008). By contrast, one s support for welfare is measured by (3) one s view on welfare spending. Support for welfare is assumed when the respondent responds that we spend too little on welfare, and a lack of support if one responds that we spend too much on welfare. (Those who respond the amount is about right are excluded.) While studies have shown that popular estimates of governmental spending are often wildly inaccurate, we may assume that people will believe we spend too much on agenda items they care little about and too little on agenda items they care much about, regardless of the dollar amount they believe is an accurate description of governmental spending. Additionally, we need not worry that many individuals censored their answer to give what they may have perceived to be a more humane or socially acceptable answer: almost 50% of respondents answered that we spend too much on welfare, while only 20% answered that we spend too little on welfare. Presumably these answers reflect the respondents true perceptions and preferences about welfare, but they remain imperfect measures. Responses to each of these three issues are represented by binary variables. The variables and the associated question texts are summarized in Table 2. In addition to the above opinion variables, we include control variables for standard demographic and socio-economic information. Demographic variables are age (a continuous variable with values from 18 to 89+), sex (a binary variable to identify women), and race (as GSS only asked whether one was White or Black until recently, we use a binary variable to identify Black respondents). Socio-economic variables include level of education (a continuous variable measured in years, 0 to 20) and kind of dwelling 3 (measured by a dummy variable that identifies those who live in detached single-family homes rather than dwellings that range from 11

12 trailer to apartment to multi-unit building). Finally, we control for political leaning (measured by a binary variable signaling those who self-identify as strong Democrat, not strong Democrat, or independent, near Democrat, rather than some level Republican). Behavior In the twenty-five (non-consecutive) years for which the GSS produced data, most respondents supported punitive penal policies and opposed welfare. Table 3 summarizes the responses on relevant questions. Between 1972 and 2006, a clear majority of respondents said they believed courts were not harsh enough on criminals, supported the death penalty, and thought spending on welfare was too much. Furthermore, over 64% of respondents who responded to both questions about punishment responded punitively, i.e., in support of capital punishment and claiming that courts are not harsh enough on criminals. Only around 8% of such respondents who answered both questions were both opposed to capital punishment and thought courts were too harsh. The statistical analyses in this study describe the extent of the correlation between penal punitiveness and welfare opposition. RESULTS We seek to determine the relationship between individuals penal punitiveness and support for welfare and the nature of this relationship over time. By using a series of statistical analyses, we find a statistically significant negative correlation between punitiveness and support for welfare and that this correlation has existed with greater and lesser strength since at least 1973, well before the years to which Beckett and Western (2001) attribute the state-level 12

13 correlation. We further find that much, though not all, of this correlation is due to cross-policy punitiveness, a trend that is currently declining. Relationship Between Punitiveness and Support for Welfare The first step of our analysis is to identify the general relationship between, on the one hand, support for welfare and, on the other, punitiveness measured by respondents (1) perception of judicial lenience towards criminals and (2) support for the death penalty. We begin first by looking at the simple correlation among the variables. Next, we establish and test the null hypothesis that there exists no difference in support for welfare among punitive respondents. This is followed by ordinary least squares (OLS) and fixed effects regression analyses of support for welfare on our measures of punitiveness. A. Simple Correlations Among Variables If we look at the relationship among the three variables without controlling for other factors, we find a strong (negative) correlation between punitiveness and welfare support. First, we find that the correlation between the belief in unduly lenient courts and support for the death penalty are not perfectly collinear (r is only ). This suggests that our measures of punitiveness reflect differences in punitiveness (or types of punitiveness). However, both measures are negatively and significantly correlated with support for welfare (r = and , respectively). It should be noted that Table 4 shows a degree of collinearity among the variables, but it is not so great as to preclude statistical analyses. The following sections examine the nature of this correlation in more detail. B. Testing the Null Hypothesis Via Matching Here we establish and test a null hypothesis that support for welfare is not correlated with punitiveness within otherwise homogenous groups. Given this hypothesis, we should expect no 13

14 difference in support for welfare among those who favor the death penalty and those who oppose it, and among those who find courts not harsh enough and those who find courts too harsh in their treatment of criminals. We limit the test of this hypothesis to white males separated by reported political party affiliation. We look first at white male conservatives partitioned by their views on courts and compare their support for welfare spending, and then at white male liberals also disaggregated by their views on courts. This process is repeated with views on capital punishment. As is evident from our results in Table 5, in each case, however, we find a statistically significant difference (z = to 10.34) in support for welfare between punitive and nonpunitive white men that leads us to reject the null hypothesis. For example, white male Republicans who oppose the death penalty are more than twice as likely to support welfare as those who support the death penalty. Among white male Democrats, views on the death penalty and on courts harshness show a statistically significant difference in welfare support of roughly 20%. Taking the analysis a step further and comparing welfare support among those who both support capital punishment and find courts too lenient and those who both oppose capital punishment and find courts too harsh, we find even more significant results. White male Democrats who are lenient by both measures are more than twice as likely to support welfare as those who are punitive by these two measures. This pattern is true for white male Republicans as well. Together, these analyses suggest the null hypothesis is false, and that we must investigate the relationship further. C. Multivariate Regression Analyses 14

15 This section uses six models based on uncontrolled ordinary least squares (OLS) regression and controlled multivariate fixed effects regression analyses to further explore the relationship between welfare support and punitiveness. As displayed in Table 6, all models reveal a statistically significant (p < 0.001) relationship between welfare and penal punitiveness net of other factors. First, using our two measures of punitiveness jointly, we find that, without controlling for standard characteristics or year-based fixed effects, those who believe courts are not harsh enough in their treatment of criminals are roughly 13.9% less likely to support welfare than those who believe courts are too harsh, and those who support the death penalty are roughly 18.8% less likely to support welfare than death penalty opponents (model I). Upon controlling for gender, race, political leaning, education, and age, the relationship still exists: support for welfare is 11.8% less likely among those who believe courts are too lenient and 11.0% less likely among those who support the death penalty (model II). The relationship is slightly stronger when we look at only one measure of punitiveness rather than both at the same time. We find that individuals who believe courts are unduly lenient are 14.1% less likely to support welfare (model III) and supporters of capital punishment are 13.9% less likely to support welfare (model IV). The final two models in this section involve a punitiveness index based on support for both, only one of the two, or neither of the two punitive penal policies. 4 High scores on the index thus correspond to strong punitive penal preferences. The results of these analyses are, again, enlightening. The uncontrolled version of this model suggests that those who support both penal policies are 33.6% less likely to support welfare than those who support neither (model V). The controlled version suggests such persons are 22.6% less likely to support welfare than their less punitive counterparts. 5 15

16 From these analyses, we can conclude that the correlation between level of punitiveness in welfare and penal policies identified by Beckett and Western (2001) strongly exists on the individual level as well. Those who are less punitive in their penal policy preferences are more likely to support welfare spending, and, by extension, welfare, than those who are more punitive in their penal policy preferences. Now we must determine if this relationship exists over time or, as Beckett and Western (2001) find for the state level, only develops in the late 1980s. Determine the Relationship over Time In this section, we perform multivariate OLS regression analyses controlling for standard characteristics to determine the relationship between welfare support and our measures of punitiveness by year starting in 1973 and again by presidency. These regressions generally (and necessarily) have fewer observations and lower significance levels, particularly in our by-year analyses using views on courts treatments of criminals as the dependent variable. Nevertheless, we still find largely negative, sizeable correlations that exist over time, regardless of the decade. Moreover, the pattern of correlation between welfare support and penal punitiveness are generally consistent across our three measures of punitiveness: the level of correlation declines toward the mid- to late 1970s and in the early and mid-1990s before climbing in the early 1980s and late 1990s. We find similar results in our analyses by presidential administration: peaks around the administrations of Richard Nixon, Ronald Reagan (generally second term), Bill Clinton (second term), and George W. Bush (second term), and troughs around the Jimmy Carter and George H.W. Bush administrations and G.W. Bush s first term. A. By Year First, excluding the non statistically significant data, we find a strong negative correlation between welfare support and support of punitive courts that persists throughout all 16

17 four decades. There are above-average correlations in the early 1970s that wane as we approach the 1980s before rising in The 1980s and 1990s correlations show great volatility and diminished statistical significance. The 2000s again show above-average inverse correlations between welfare support and penal punitiveness. Thus, rather than developing only in the 1980s and 1990s, we see weakened significance levels in these years (though these are also years with fewer observations) and the strongest significance levels in the 1970s and 2000s. This shows that, at least for the relationship between individuals support for welfare and for punitive courts, the correlation manifested itself much earlier than the late 1980s. The picture is clearer when it comes to support for capital punishment and welfare: here, only one year (1989) lacks significance. Again, we begin with high correlations in the 1970s that decline as we approach the 1980s, which witness a rise in the level of correlation before dropping to average levels in the 1990s and lower in The level of correlation rises as we approach the Millennium. The 2000s again generally show above-average levels of correlation. Thus, as with the relationship with support of punitive courts, support for capital punishment is significantly correlated as early as 1973 and continues to be so. The pattern is the same with the punitiveness index, though with stronger negative correlations between penal punitiveness and welfare support. As with the other two single measures of punitiveness, the correlation begins strong in the early 1970s, declines as it reaches the end of the decade, inclines with volatility in the 1980s, drops in the middle of the 1990s before rising as it reaches the 2000s with some decline in B. By Presidential Administration By performing these analyses by presidential administration rather than by year, we can see more clearly the trend over time (Table 8). First, it is important to note that there is variation 17

18 over time: each set displays high year to low year variations of between 10% and 16%. Described differently, the highest-level correlation in each set is between two and three times the magnitude of lowest-level correlation in that set. Thus, there is definite variation across presidencies. In the regressions of welfare support on support for harsher courts, the four largest correlations are found, in order of magnitude, under Richard Nixon (21.1%), George W. Bush s second term (18.8%), Ronald Reagan s first term (16.5%), and Bill Clinton s second term (15.4%). We see a similar pattern in the regression on support for capital punishment: G.W. Bush s second term leads (20.2%), followed by Reagan s second term (19.6%), Nixon (18.1%), and Clinton s second term (16.7%). Similarly for the punitive index, we find the strongest correlation under G.W. Bush s second term (31.3%), Nixon (30.3%), Reagan s second term (27.8%), and Clinton s second term (25.3%). Thus, rather than seeing a continuous upward trend in the correlation which is what we would expect if individual-level correlations mirrored the state-level correlations in Beckett and Western s (2001) analysis we find a series of peaks and troughs: peaks around Nixon, Reagan (generally second term), Clinton (second term), and G.W. Bush (second term), and troughs around the Jimmy Carter and George H.W. Bush administrations and G.W. Bush s first term. C. Tests of Real Significance Are these findings spurious? Either one of two situations would suggest they are: (1) Punitiveness is constant, but generosity regarding welfare spending fluctuates. The peaks and troughs reflect periods of agreement and disagreement, respectively. (2) Generosity in welfare spending is constant, but punitiveness fluctuates. Again, the peaks and troughs reflect periods of agreement and disagreement, respectively. 18

19 By examining the aggregate trends in the GSS data, we find evidence against both of these possibilities; instead, it appears that punitiveness and generousness in welfare spending both fluctuate (though by different magnitudes), and they fluctuate together (with negative correlation). We find that our measures of punitiveness are both generally consistently high across the period of interest (averages of 82.6% and 72.1%, respectively), welfare support seems to fluctuate quite substantially and in recognizable periods, despite a generally low support level (average of 29.2%). However, upon putting the aggregate trends in standard units (the average is subtracted from each value, and is then divided by the standard deviation), we can see more clearly the trends in the levels of support for each (Graph 1). While the three levels of support do not produce mirror images with this new calculation, we can see that peaks in support for capital punishment or harsh courts generally correspond with troughs in welfare support of roughly the same magnitude, and vice-versa. We see clear peak-trough combinations developing most fully in the 1980s, and reaching their apex in the years 1994 (high punitiveness, low welfare support) and 2006 (low punitiveness, high welfare support). Peak-trough combinations are apparent in the 1970s, but not as frequently or as clearly. The state-level manifestation of the correlation in the 1990s may reflect the general upward trend in individual-level correlations around Clinton. Similarly, the peaks around Reagan s second term may be a product of the rhetoric of the Reagan revolution referred to in Beckett and Western (2001, 55). However, the peak under Nixon suggests that Nixon-era rhetoric that began the association between the undeserving poor and criminal also had an effect on public opinion, even if it took state-level policies several decades to manifest what Beckett and Western (2001) call the penal-welfare regime. 19

20 From these several analyses, we can conclude that support for welfare and punitiveness were (negatively) correlated for individuals well before it became apparent at the state level. Additionally, we may note that these peaks appear to gain the most strength under conservative political rhetoric Nixon, Reagan, Clinton, and George W. Bush. We note that, although President Clinton was a democrat, his administration witnessed a war on welfare and displayed levels of punitiveness typical of late-century liberal politicians (i.e., indistinguishable from conservatives levels; see Simon 2007). To further understand these correlations, it is now necessary to examine the nature of the correlation over time. In the following section, we determine whether correlated views are primarily cross-policy punitiveness, cross-policy welfarism, or a combination of these. Determine the Nature of the Relationship over Time Given that the general direction of welfare and penal policy in the last third of the twentieth century was towards punitiveness, we can make better sense of the trends if we look to the portion of the public that shows punitive preferences in these policy areas and determine whether that portion increases over time as well. In this section, we first step away from regression methods and simply look at the aggregate portion of GSS respondents who support punitive punishments (harsher courts, capital punishment, or both) and oppose welfare. As is evident in Graph 2, we do not find a continuous increase in levels of cross-policy punitiveness over time, but instead find variation over time including peaks and troughs that do not mirror the general trend for correlated views. As a check on these uncontrolled analyses, we then employ regression methods again to determine the level of cross-policy punitiveness associated with each presidency net of our standard controls. We find very strong peaks of cross-policy 20

21 punitiveness that are consistent with our uncontrolled annual levels of aggregate support for punitive policies. A. Correlated Views and Cross-Policy Punitiveness in Annual Aggregate Views In general, we observe peaks of cross-policy punitiveness in the late 1970s and the mid- 1990s, and troughs in the mid-1970s, the 1980s, and the 2000s. In each of our three measures, we find the lowest proportion of cross-policy punitiveness in 2006 (46.9% where we use harsher courts as a measure of penal punitiveness, 46.4% when we use capital punishment, and 39.6% when we use both). Similarly, we find the highest proportion in 1994 (76.5%, 70.1%, and 66.4%, respectively). The punitiveness peak in the mid-1990s occurs around the same time that harsher welfare-penality policy regimes were appearing on the state level. However, the peaks in the late 1970s (with or without death penalty support as a measure) suggest that the public was ripe then for supporting a harsher welfare-penality policy regime. As a proportion of the overall correlation, cross-policy punitiveness tends to be relatively large; however, it does not encompass the entire correlation (see Graph 3). Where our measure of penal punitiveness is support for capital punishment, we see that penal punitiveness and welfare support are generally inversely correlated, but a sizeable proportion of the set of correlated views come from a lenience grouping (pro-welfare, anti death penalty correlations). The size of the lenience grouping is much smaller, though it still exists, when our measure of penal punitiveness is support for harsher courts or is both support for harsher courts and capital punishment. It is important to note that in all three cases, beginning around 2000, we see an increase in the size of the lenience grouping. Though never large, this grouping has nevertheless nearly quadrupled from its 1994 levels. Between the decline in the proportion of the public who support cross-policy punitiveness, and the increase of the lenience grouping around the same time, this 21

22 may indicate that the public is growing ready for a policy regime change. Given the relatively large portion of the punitive grouping operating since the mid-1970s, the recent divergence may be of interest for further research. B. Cross-Policy Punitiveness by Presidency In this section, we create three variable signifying cross-policy punitiveness (one for each measure of penal punitiveness) and then perform three controlled regressions of this variable on each presidency (with the last presidency George W. Bush s second term as our comparison group). Each variable measuring cross-policy punitiveness enjoys a clear majority of respondents who answered all relevant questions: where death penalty support is our measure of penal punitiveness, 57.4% show cross-policy punitiveness; where it is support for harsher courts, 64.1%; where it is both, 53.7%. However, the results of our regression show substantial variation over time (see Table 9 and Graph 4). The coefficient on each presidency (except the first administration of George W. Bush in our third regression) is statistically significant and each regression shows remarkably similar results. We find the largest coefficients of cross-policy punitiveness on presidency in the Carter Administration and first term of the Clinton administration. We see declines in each analysis following the Carter Administration through the George H.W. Bush Administration. Moreover, in each case, the first term of the George W. Bush Administration reveals lower cross-policy punitiveness than the corresponding levels under President Nixon. Finally, as the second term of George W. Bush is our reference group, we may note that each of the previous administrations (except Bush s first term in our third regression) had (statistically significant) higher levels of cross-policy punitiveness, which reflects the decline (described above) in the portion of individuals supporting these views. 22

23 DISCUSSION Summary We have shown that, net of other factors, individuals support for welfare and support for penal punitiveness (harsher courts, capital punishment, and an index of support for both) are strongly inversely correlated, and that this relationship persists across time with some important variation. In general, the level of correlation declines toward the mid- to late 1970s and in the early and mid-1990 before climbing in the early 1980s and late 1990s. The resultant peaks generally correspond to the Nixon administration, Reagan s second administration, Clinton s second administration, and G.W. Bush s second administration. Further, we have shown that the aggregate inverse correlation between welfare support and punitiveness is largely, but not wholly, composed of cross-policy punitiveness. This punitive portion also has peaks and troughs: cross-policy punitiveness peaks in the late 1970s and the mid-1990s, while troughs appear in the mid-1970s, the early and late 1980s (but not the middle of that decade), and the 2000s. These correspond to peaks in the Carter and first Clinton administrations, and troughs under the George H.W. Bush and the second George W. Bush administrations. We speculate that the large portion of cross-policy punitive views is a result of the rhetoric of the 1960s discussed by Beckett (1997) and others. While our data do not permit an empirical test of this speculation, we suggest that, were it possible to examine data from the 1950s and 1960s, we would initially see cross-policy welfarism that would begin to transform in the 1960s toward cross-policy punitiveness. We further suggest that these more punitive views, responding to political rhetoric, supported the state-level policies that (for various reasons to be described below) did not manifest in Beckett and Western (2001) until the late 1980s and mid- 23

24 1990s. The decline in the portion of correlated views that are punitive since the 1990s and the concurrent rise in the portion of cross-policy welfarist views may suggest another regime change. Let us hope that it does not take as long to manifest itself on the state level as (we believe) the cross-policy punitiveness did. Our findings leave us with two questions: First, what is it that links welfare and penal policy that lends itself to individual preferences for punitiveness in both or welfarism in both? Second, why does the individual-level correlation differ somewhat from the state-level correlation demonstrated in Beckett and Western (2001)? Reflecting on the Welfare-Penality Complex among Individuals As Greenberg suggests, If imprisonment and welfare are both products of the same mind-set, the critical issue is to understand the sources of this mind-set, and why it changed between the 1960s and subsequent decades, with consequences for both imprisonment and welfare (2001, 84). What is it that links welfare and punishment in the minds of individuals? The historical marriage of crime and poverty (for example, the original use of the prison as workhouse and a place for debtors; see, e.g., Ignatieff 1979) suggests that poverty and crime have long been associated with each other in more ways than we have room to discuss (see Garland 1985, 1990; see also Rusche and Kirchheimer 1939). Turning to cognitive science, we find that Lakoff (2002) suggests that people tend to think in certain frames the strict father frame or the nurturing parent frame that dictate and unify preferences in all areas of life. Following Lakoff (2002), we may suggest that perhaps there is something intrinsic in one s cognition that will dispose one to believe (the strict father notion that) nature, not nurture, is the cause of various kinds of social marginality such as poverty and criminality and, as such, these behaviors require a more punitive response (a strict father response). Similarly, one who believes 24

25 nurture and other environmental causes are at fault would favor a more helpful, lenient approach (the nurturing parent approach) (Lakoff 2002, ). This would suggest that these two policy areas will inevitably be linked and will only be disrupted by external forces like the economy, new and persuasive discourses, racism, etc. However, it seems plausible that, whether there exists a natural linkage between views of the poor and criminal, punishment and welfare, in human thinking, political discourse and racism did much to manipulate these views to become more punitive over the last half century. As has been noted, political rhetoric in the second half of the twentieth century has spoken about poverty and criminality in similar terms. Rather than suggesting theories of external causes of poverty and criminality such as unemployment rates, sub-standard education, or discrimination for various reasons, politicians (and citizens who listen) have described these as products of personal responsibility, or rather lack thereof. Being poor or criminal is a person s fault for making bad choices and being lazy, bad, or otherwise unworthy; there are no mitigating factors. Although this rhetoric has been deployed since the 1960s (Beckett 1997, 10), it did not produce a state-level correlation between welfare and penal policy until the late 1980s and 1990s, suggesting that declining support for welfare and increased punitiveness are results of the Reagan administration (Beckett and Western 2001, 55). Nevertheless, the result has been that penal and welfare policy share the same assumptions, harbor the same anxieties, deploy the same stereotypes, and utilize the same recipes for the identification of risk and the allocation of blame (Garland 2001, 201), each of which is consistent with this political rhetoric. Whether purposefully or not, the political rhetoric emphasizing personal responsibility and the policies that followed have racial overtones. The description of personal responsibility for one s poverty, for example, fits with widespread American racial stereotypes of blacks as 25

26 lazy (Gilens 1995). Indeed, Gilens (1996, 601) has shown that racial considerations are the single most important factor shaping whites' views of welfare. Similarly, views towards punishment are likewise racialized. Unnever, et al., (2008) for example, have shown that racism is a powerful explanation for the difference in black and white rates of support for capital punishment and that the effect of racial or ethnic animosity on death penalty support is not limited to the United States. Indeed, many have argued that it is not coincidental that political rhetoric and policy (particularly policy that had significant consequences for black Americans) changed not long after the Civil Rights Movement peaked (see, e.g., Wacquant 2001). However, as Garland (2001) notes, the changes in penal and welfare policy and related rhetoric was likely in response to a combination social change that had built up across the middle decades of the twentieth century (Garland 2001, 195). Our data cannot say whether the correlation between welfare opposition and support for punitive punishments in public opinion since 1973 is a response to political rhetoric, a product of racist attitudes (or whether or not such attitudes are exacerbated by political rhetoric), or a manifestation of human cognition patterns. 6 Further testing may reveal that one or more of these theories may explain the fluctuation in the levels of correlation (and cross-policy punitiveness) over time. However, our results do not rule out the possibility that public opinion responded early to whatever mindset anticipated the changes in both penal and welfare policy later in the century. Explaining the Difference in State-Level and Individual-Level Findings In their analysis of state-level data from 1975, 1985, and 1995, Beckett and Western (2001) find that a strong negative relationship between welfare generosity and penal punitiveness did not come into existence until The penal-welfare regime thus appears to 26

27 have crystallized relatively recently (2001, 52). By contrast, we have shown strong correlations between opposition to welfare and support for punitive penal policies on the individual level since If the reconceptualization of the nature and causes of the problems associated with social marginality [i.e., crime and poverty] has legitimated the shift away from welfarism in both the penal and social spheres as Beckett and Western (2001, 46) conclude (rightly, we believe), and this discourse has been operating since the 1960s, why has the relationship not been apparent at the state level until the 1990s even while individuals have accepted it since the 1970s at least? There are more than several ways to explain the difference in the findings presented here and those in the Beckett and Western (2001) study. For a variety of reasons, individual cross-policy punitiveness may not match states actual punitiveness (i.e., policy), or apparent punitiveness (i.e., incarceration rates and welfare payouts). We shall focus on three explanations: that public opinion has little or no effect on state policy; that state policy substantially trails public opinion; and that incarceration rates are inaccurate as a measure of state policy. First, public opinion may have very little impact on state policy. Roberts (1992) suggests that public opinion is consistently punitive and as such does not correspond to policy; rather officials respond to overestimated levels of public punitiveness. He compares Canadian and U.S. attitudes that sentences are too lenient and purports to show the general consistency in such beliefs across time and borders (Roberts 1992, 148). In fact, this study shows the percentage of adherents to this belief vary in the United States from the high 40s (in the 1960s) to the high 70s (in the 1980s), which should suggest against constancy in beliefs over time. Additionally, the percentages in the two countries at times vary by twenty points or more. Such data should not be interpreted as representing constancy in beliefs across time and space. Moreover, as we noted above, recent research suggests that there is at least a correlation between public opinion and 27

28 policy, or possibly even an effect of the former on the latter (Lappi-Seppälä 2008, Barker 2006). However, as we are not here rigorously testing the correlation between public opinion and policy, we shall leave this possibility as an explanation for the difference in individual-level and state-level differences. Second, public opinion may have an effect on state policy, but one that takes a while to become manifest on the state level. Public opinion is far more malleable than is state policy. To the extent that policy is influenced by public opinion, there is often a time delay before public opinion is capable of influencing law and policy. That is, public opinion, while trailing political rhetoric, often precedes policy change. This may be especially true on the state level when public opinion more closely resembles national-level trends. For one thing, by the time policy is made, there is also the potential for push-back by state-level penal officials that may have an effect on incarceration rates (Sutton 1987). This would be especially plausible in the 1970s and early 1980s when prison administrators were undergoing a transformation from the penal-welfarism of the 1940s to the 1960s to the new penology of the 1980s and 1990s (Feeley and Simon 1992). Additionally, there is the possibility that the public, or at least some portion thereof, has little effect on state-level policy when there are other checks in place. As Barker (2006, 26) notes, Black political participation in state governance provides a crucial buffer against the use of imprisonment as a blunt instrument of racial social control. Thus, even if there was widespread public support for a penal-welfare regime since the 1970s, it may be that different mechanisms (time lag between desire and policy, administrative/official pushback, political blocks) were in place to prevent penal or welfare reforms from taking place in a number of states. As time went on, these mechanisms would eventually be overcome, more and more states would change their 28

29 policies, and by the 1990s a state-level correlation between penal and welfare policies was apparent. Third, incarceration rates, which were Beckett and Western s (2001) measure of state penal punitiveness, are somewhat problematic as a measure of penal policy punitiveness. While policy change itself often takes time, penal policy takes time to manifest itself in incarceration rates. Thus, Beckett and Western (2001) lag their incarceration rates by two years (e.g., the 1997 rates as a measure of penal punitiveness in 1995). However, incarceration rates are the result of years of development rather than, in most cases, a single political season. For example, the current incarceration rate of California is the result of several decades worth of legislation and does not necessarily correspond to the level of punitiveness of its current population (see, e.g., Zimring, et al., 2001). Indeed, though a standard in measuring state/national punitiveness, incarceration rates are nevertheless the subject of some debate in the literature. Frost (2008) criticizes the operationalization of punitiveness in the literature as simply punishing more (and typically measured as imprisoning more ) and calls for scholars to pay greater attention to penal intensity (e.g., length of sentences) rather than the propensity to incarcerate (i.e., the incarceration rate) (Frost 2008, 278). 7 Moreover, Greenberg (2001, 21) suggests that Beckett and Western s (2001) use of incarceration rates do not allow for the prediction of changes on the margin in a state s level of punitiveness. Finally, after reviewing several problems with using incarceration rates as a measure of punitiveness Lappi-Seppälä (2008, 330) concludes that [t]here is no serious alternative. However, he also suggests that [a] study using imprisonment rates as a key policy indicator clearly needs to justify itself (2008, 322). We believe that, in the particular context of U.S. penal trends, incarceration rates face additional challenges as a measure of punitiveness. 29

30 To the extent that incarceration rates may not accurately reflect the level of punitiveness in a given state, this measure of punitiveness may greatly cloak the correlation with the state s welfare generosity. First, the rate of increase in incarceration rates has been roughly constant over the last three decades, despite the fact that there have been substantial policy changes (aimed at severity) since the 1970s and even before. Measuring punitiveness over time by examining incarcerations rates alone will at first underestimate the level of punitiveness in policy and only later reflect more closely the true level of punitiveness. Second, since 1973, support for welfare has varied substantially. Importantly, support for welfare was lowest between 1976 and 1980 and in 1994 (based on GSS aggregate trends). The result of these two time trends is, in 1975, punitiveness would have appeared (relatively) low and welfare support low (positive correlation); in 1985, punitiveness would have appeared higher and welfare support higher (positive correlation); and finally in 1995, punitiveness would have appeared very high and welfare support very low (negative correlation). Thus, of the three years, it is conceivable that the negative relationship could only appear in It is possible that a positive relationship may have been more likely to surface from the data given more observation years, but any correlations found would not likely reflect the true level of correlation, such as the one apparent on the individual level one that is strong, but fluctuates over time. 8 Conclusion The purpose of the current research has been to determine whether welfare and penal preferences are correlated on the individual level, as they appear to be on the state and national levels, and in what way. We have determined that, net of other factors, these preferences are correlated on the individual level and have been since (at least) 1973, though with varying degrees of strength. Moreover, we have also seen that the nature of this correlation portion 30

31 punitive (supports harsher punishments, opposes welfare) and portion welfarist (opposes harsher punishments, supports welfare) changes over time. We suggest that by understanding whether and why these views are correlated on the individual-level is an important aspect that has thus far been mostly ignored by the explanations of the penal-welfare complex. It may help us to determine which (or which combination) of the current theories is a more powerful explanation for the policy regime change political rhetoric, racism, neo-liberal market pressures, etc. is more accurate. Further research should explore the roots of the change in the strength and tone of the welfare-penality correlation on the individual level. Thus, it will be useful to determine if the high levels of correlation (particularly in the punitive direction) in the early 1970s are an artifact of Nixon-era political rhetoric or racial backlash, and if the increase in the level of correlation around the Reagan administration may be a result of Reagan-era political rhetoric. Most importantly, it will be necessary to examine why cross-policy punitiveness began to give way to cross-policy welfarism on the individual level beginning in the late 1990s. BIBLIOGRAPHY Barker, Vanessa (2006) The Politics of Punishing: Building a State Governance Theory American Imprisonment Variation. Punishment and Society, Vol. 8, pp Barker, Vanessa (2007) The Politics of Pain: A Political Institutionalist Analysis of Crime Victims Moral Protests. Law & Society Review, Vol. 41, No. 3, pp Beckett, Katherine (1997) Making Crime Pay: Law and Order in Contemporary American Politics. New York: Oxford University Press. 31

32 Beckett, Katherine, and Bruce Western (2001) Governing Social Marginality: Welfare, Incarceration, and the Transformation of State Policy. Punishment & Society, vol. 3, pp Cavadino, Michael, and James Dignan (2006) Penal Systems: A Comparative Approach. London: Sage Publications. Costelloe, Michael T., Ted Chiricos, and Marc Gertz (2009) Punitive Attitudes Toward Criminals: Exploring the Relevance of Crime Salience and Economic Insecurity. Punishment and Society, Vol. 11, pp Davis, James Allan and Smith, Tom W. (2007) General social surveys, [machinereadable data file]. Principal Investigator, James A. Davis; Director and Co-Principal Investigator, Tom W. Smith; Co-Principal Investigator, Peter V. Marsden; Sponsored by National Science Foundation. --NORC ed.-- Chicago: National Opinion Research Center [producer]; Storrs, CT: The Roper Center for Public Opinion Research, University of Connecticut [distributor]. Downes, D. (2001) The Macho Penal Economy: Mass Incarceration in the United States - A European Perspective. Punishment & Society, Vol. 3, No. 1, Downes, D. & Hansen, K. (2006) "Welfare and Punishment in Comparative Perspective." In Perspectives on Punishment: The Contours of Control, edited by Armstrong and McAra. Oxford: Oxford University Press: Feeley, Malcolm, and Jonathan Simon (1992) The New Penology: Notes on the Emerging Strategy of Corrections and Its Implications. Criminology Vol. 30, No. 4, pp Frost, Natasha A. (2008) The Mismeasure of Punishment. Punishment & Society, Vol. 10, p

33 Garland, David (1985) Punishment and Welfare: A History of Penal Strategies. Aldershot: Gower. Garland, David (1990) The Culture of Control: Crime and Social Order in Contemporary Society. Chicago: University of Chicago Press. Garland, David (2001) The Culture of Control: Crime and Social Order in Contemporary Society. New York: Oxford University Press. Greenberg, David F. (2001) Novus Ordo Saeclorum?: A Commentary on Downes, and on Beckett and Western. Punishment & Society, vol. 3, p Ignatieff, Michael (1978) A Just Measure of Pain: The Penitentiary in the Industrial Revolution, New York: Pantheon Books. Jackson, P., and L. Carroll (1981) Race and the War on Crime: The Sociopolitical Determinants of Municipal Police Expenditures. American Sociological Review, Vol. 46, pp Lakoff, George (2002) Moral Politics: How Liberals and Conservatives Think. Chicago: University of Chicago Press. Lappi-Seppälä, T. (2008) Trust, Welfare, and Political Culture: Explaining Difference in National Penal Policies." In Crime and Justice: A Review of Research, Vol. 37, edited by Michael Tonry. Chicago: The University of Chicago Press: Parenti, Christian (1999) Lockdown America: Police and Prison in the Age of Crisis. New York: Verso. Putnam, Robert (2007) E Pluribus Unum: Diversity and Community in the Twenty-first Century. Scandinavian Political Studies, Vol. 30, No. 2, pp Roberts, Julian V. (1992) Opinion, Crime, and Criminal Justice. Crime and Justice, Vol. 16, pp

34 Roberts, Julian V., and Anthony N. Doob (1989) Sentencing and Public Opinion: Taking False Shadows for True Substances. Osgoode Hall Law Journal, Vol. 27, pp Rusche, Georg, and Otto Kirchheimer (1939) Punishment and Social Structure. New York: Russell & Russell. Simon, Jonathan (2007) Governing through Crime: How the War on Crime Transformed American Democracy and Created a Culture of Fear. New York: Oxford University Press. Snell, Tracy L. (1998) Capital Punishment U.S. Department of Justice, Bureau of Justice Bulletin NCJ Sutton, John R. (1987) Doing Time: Dynamics of Imprisonment in the Reformist State. American Sociological Review, Vol. 52, No. 5, pp Sutton, John R. (2004) The Political Economy of Imprisonment in Affluent Western Democracies, American Sociological Review, Vol. 69, No. 2, pp Tonry, Michael. (2009) Explanations of American Punishment Policies: A National History. Punishment and Society, Vol. 11, pp Tyler, T.R., and Boeckmann, R. (1997) Three strikes and you are out, but why? The psychology of public support for punishing rule breakers. Law and Society Review, Vol. 31, pp Unnever, James D., Francis T. Cullen, and Cheryl Lero Jonson (2008) Race, Racism, and Support for Capital Punishment. In Crime and Justice: A Review of Research, Vol. 37, edited by Michael Tonry. Chicago: The University of Chicago Press: Wacquant, Loïc (2001) Deadly Symbiosis: When Ghetto and Prison Meet and Mesh. Punishment and Society, Vol. 3, pp

35 Young, Jock (1999) Cannibalism and Bulimia: Patterns of Social Control in Late Modernity. Theoretical Criminology, Vol. 3(4), pp Zimring, Frank, and David Johnson (2006) Public Opinion and the Governance of Punishment in Democratic Political Systems. The ANNALS of the American Academy of Political and Social Science, vol. 605, pp Zimring, Frank, and Gordon Hawkins (1997) Crime is Not the Problem: Lethal Violence in America. New York: Oxford University Press. Zimring, Frank, Sam Kamin, and Gordon Hawkins (2001) Punishment and Democracy: Three Strikes and You re Out in California. New York: Oxford University Press. 35

36 APPENDIX I: TABLES Table 1. GSS Respondents Characteristics, 1972 to 2006 Variable Total N Mean SD Age (in Years) Education (in Years) Dwelling is a House % 48.6% Liberal Party % 50.0% Black % 34.5% Woman % 49.6% 36

37 Table 2. Variables Explained Variable GSS Question Text Variable Values Welfare q. 69. We are faced with many problems in this country, none of which can be solved easily or inexpensively. I'm going to name some of these problems, and for each one I'd like you to tell me whether you think we're spending too much money on it, too little money, or about the right amount. k. Welfare. Courts Death Penalty q. 90. In general, do you think the courts in this area deal too harshly or not harshly enough with criminals? [Excludes response about right ] q. 91. In general, do you think the courts in this area deal too harshly, or not harshly enough with criminals, or don't you have enough information about the courts to say? [Excludes response about right ] q. 82a. Are you in favor of the death penalty for persons convicted of murder? (asked in Surveys) q. 82b. Do you favor or oppose the death penalty for persons convicted of murder? (asked in Surveys from 1974 and after) If Welfare = 1, too little If Welfare = 0, too much If Courts = 1, not harshly enough If Courts = 0, too harshly If Death penalty = 1, favor If death penalty = 0, oppose (or no ) 37

38 Table 3. Summary of Relevant Opinions and Punitive Index among Respondents Opinion Total N Responding N with Response Percent with Response Issues courts not harsh enough 43,432 35, support the death penalty 43,537 31, not enough spent on welfare 19,998 5, Punitiveness Index 0.0 (Neither) 40,897 3, (One of Two) 40,897 11, (Both) 40,897 26,

39 Table 4. Correlation Among Variables Welfare Courts Death Penalty Welfare *** *** *** Courts *** *** Death Penalty *** *All values are statistically significant. 39

40 Table 5. Differential Support for Welfare Between Punitive and Non-Punitive Members of Homogenous Groups. Courts White Male Republicans Death Penalty Punitiveness Index Courts White Male Democrats Death Penalty Punitiveness Index Too Too Too Too Support Oppose High Low Lenient Harsh Lenient Harsh Support Oppose High Low Mean SD N z-score

41 Revision Table 6. Coefficients on Measures of Punitiveness in Welfare Regression Analyses Model I Model II Model III Model IV Model V Model VI Courts (0.010) (0.010) (0.009) Death Penalty (0.008) (0.008) (0.007) Punitiveness (0.011) (0.011) Controls No Yes Yes Yes No Yes Pseudo R- Squared N 17,125 16,742 17,561 17,700 17,125 16,742 Controls, where included, are not shown. Standard errors are shown in parentheses. All values shown are significant at the p < level. 41

42 Revision Table 7. OLS Regression Coefficients of Support for Welfare on Support for Harsher Courts (Model III) and the Death Penalty (Model IV), and the Punitiveness Index (Model VI) Model III (by Year) Model IV (by Year) Model VI (by year) Year B courts Adj. R 2 N B death Adj. R 2 N B Index Adj. R 2 N * * * * * * * * * penalty * *

43 Revision * Control variables not shown. Lack of statistical significance is denoted by *

44 Revision Table 8. Coefficients for Measure of Punitiveness in Regressions of Welfare Support on Penal Punitiveness Coef. SE Nixon Ford Carter Reagan (1) Reagan G.H.W. (2) Bush Death Penalty Clinton (1) Clinton (2) G.W. Bush (1) G.W. Bush (2) Adj. R N Courts Coef. SE Adj. R N Punitive Index Coef. SE Adj. R N All values statistically significant at the p <0.05 level. 44

45 Revision Table 9. Coefficients on Presidencies in Regression of Cross-Policy Punitiveness on Controls and Presidencies. Using Death Penalty Support Using Harsh Courts Using Both Nixon Ford Carter Reagan Reagan GHW Bush Clinton Clinton GW Bush * N Adj. r Controls not shown. Lack of statistical significance is denoted by *. All remaining values are significant at the p <.01 level. Reference presidency is second term of the G.W. Bush administration. 45

46 Revision Graph 1. Annual Aggregate Support for Welfare, the Death Penalty, and Harsher Courts, in Standard Units Note: The levels of support reported here are the aggregate support found in the GSS, not the levels of support found when controlling for our standard controls. 46

47 Revision Graph 2. Portion that Supports Punitive Punishments and Opposes Welfare (Using Various Measures of Punitive Punishments). Note: The levels of support reported here are the aggregate support found in the GSS, not the levels of support found when controlling for our standard controls. 47

48 Revision Graph 3. Cross-Policy Inverse Correlations and Proportion Supporting Cross-Policy Punitiveness (Using Various Measures of Penal Punitiveness) Note: The levels of support reported here are the aggregate support found in the GSS, not the levels of support found when controlling for our standard controls. 48

49 Revision Graph 4. Presidency Coefficients in Controlled Regression of Cross-Policy Punitiveness on Presidency (Using Various Measures of Penal Punitiveness) Comparison group is the second George W. Bush Administration. 49