Gender differences on crime and punishment.

Since the 1980 election, considerable political science attention has been devoted to the gap between women's and men's political preferences and voting behavior. Although the differences between the sexes have tended to be fairly small in empirical terms,(FN1) they have appeared across a consistent range of issues. At the same time, feminist theory has been exploring the implications of Carol Gilligan's In a Different Voice (1982). Gilligan's suggestion that women's psychology is premised on an "ethic of care" rather than on an "ethic of justice" has been applied in numerous contexts because it seems to fit with the differences found in women's and men's approaches to a range of topics.(FN2).

Gender differences have been well-established over a wide array of public policy attitudes and behaviors, including policies promoting government caretaking of vulnerable citizens (Conover 1994; Thomas 1994; Rhine et al. 1994; Mueller 1988; Stoper 1988), military commitments (Conover and Sapira 1993; Shapiro and Mahajan 1986; Smith 1984), and voting (Bedyna and Lake 1994; Gilens 1988; Pomper 1975). Less is known, however, about male-female differences on one of the most salient "hot button" issues--crime. This gap is surprising for two reasons: first, there are important theoretical reasons to presuppose gender-based differences in the ways men and women respond to crime; and second, as we shall argue in the conclusions, crime is a strategically important issue. The failure to appreciate gender-related differences pertaining to crime, therefore, undercuts the efforts of candidates to turn the issue to partisan advantage.

In this article we systematically examine how, and why, women and men approach issues of crime and punishment. Our first purpose is to explore the specific areas in which the genders differ. Next, we consider two competing explanations of these differences. The first of these hypotheses relates crime and punishment attitudes to gender differences in perceptions of vulnerability. The second links these attitudes with the "different voices" approach espoused by Gilligan. Our findings support the idea that the gender gap in crime and punishment attitudes owes a great deal to differences in men and women's caretaking orientations rather than emerging solely from differences in their fears of victimization.

PERCEPTIONS OF VULNERABILITYAttitudes toward crime and punishment are often thought to flow from feelings of vulnerability. It would come as no surprise to find that people who are more afraid of crime are also more supportive of efforts to prevent it. For example, individuals who feel particularly vulnerable might be more prone to favor increasing the power of the police or passing laws to make guns harder to obtain. Those who are less afraid may be less likely to see the need for such measures.

Perceptions of vulnerability may explain orientations toward punishment as well as prevention. If harsher penalties are thought to keep the public safer by deterring potential offenders,(FN3) we would expect to find people's support for aggressive responses to crime to increase along with their level of fear. Indeed, studies of attitudes toward crime almost all proceed from the notion that fear of victimization increases harshness (for examples, see Miller, Rossi, and Simpson 1986; Research and Forecasts 1983; Stinchcombe et al. 1980). On this basis, those who see themselves as potential victims are expected to support policies such as longer prison terms and capital punishment more than those who feel more secure.

If perceived vulnerability is the key to these attitudes, we would expect to find a gender gap in public responses to crime and punishment issues. Women are made to feel more vulnerable than men in a number of areas of life. Women have traditionally controlled a smaller share of political and economic resources than men (see for example Goldberg and Kremen 1994). Large numbers of working women are concentrated into lower paying, low prestige jobs, resulting in widespread feelings of economic vulnerability (see, for example, Sidel 1986). Women are also more likely than men to need public assistance, especially if they are raising children as single parents (see, for example Norris 1985).

Women's lives are also profoundly influenced by the threat of physical violence, particularly domestic abuse and rape, in ways that men's are not (Ferraro 1995; Walklate 1995; Gordon and Riger 1989). And, women are much more likely than men to modify their behavior because they fear victimization (Ferraro 1995; Walklate 1995; Gordon and Riger 1989). Stanko (1990: 85) explains this as a response to women's heightened sense of vulnerability:.

Women's lives rest upon a continuum of unsafety. This does not mean that all women occupy the same position in relation to safety and violence.... Somehow, though, as all women reach adulthood, they share a common awareness of their particular vulnerability. Learning the strategies for survival is a continuous lesson about what it means to be female.

Conversely, men tend to have greater confidence in their ability to protect themselves (Stanko 1990). This may explain why surveys invariably report greater fear of crime among women than men (Walklate 1995; Ferraro 1995; Karmen 1991; Stanko 1990; Gordon and Riger 1989; Research and Forecasts Inc. 1983; Stinchcombe et al. 1980).

Women's greater sense of vulnerability therefore seems likely to produce a gender gap in crime and punishment attitudes. If fear is the primary determinant of such attitudes, women should be particularly supportive of crime prevention policies. Women's greater fear may also lead them to be more punitive than men, especially in response to crimes where woman are particularly vulnerable, such as rape or domestic abuse. This would be consistent with Gordon and Riger's (1989: 131) finding that "many women think rape might be deterred by meting out harsher penalties to convicted rapists," and Pierce and Harris's (1993) findings that women were more likely than men to advocate conviction for wife batterers. Given women's particular role in caring for children, we might also expect women to react with greater punitiveness when confronting crimes of violence against children.

A few studies have found women to be more supportive than men of some crime prevention policies, although there is not a great deal of evidence on this point. At least two surveys have found that women are less supportive of civil liberties which inhibit the ability of the police to catch criminals (Rhine et al. 1994; Shapiro and Mahajan 1986). Women also support gun control measures at higher rates than men do (Shapiro and Mahajan 1986; Smith 1984). It may be that men are less supportive of these measures because they feel less need for protection from crime. This relationship is unclear, however, because the link between fear and support for prevention policies has generally been assumed rather than tested.

The empirical relationship between fear and punitiveness is even less clear. Though many studies have concluded that punitiveness varies with a person's fear of crime, their data reveal a connection for men but not for women (see, for example, Ferraro 1995; Mills and Bohannon 1992; Stinchcombe et al. 1980). In fact, men, with their lower levels of fear, are often more supportive of harsher penalties. For example, men are more likely than women to support the use of capital punishment for murderers (Conover 1994; Shapiro and Mahajan 1986; Smith 1984; Stinchcombe et al. 1980). On the other hand, women do favor more punitive responses to rape and domestic abuse, and female jurors are more likely to convict those charged with violent crimes (Mills and Bohannon 1992). It is clear that further exploration of the relationship between gender, perceived vulnerability, and the desire to punish offenders is required.

CRIME AND THE ETHIC OF CAREIt is likely that factors other than mere fear, particularly orientations toward violence and its prevention, play a key role in distinguishing men's attitudes from women's toward crime and punishment. Considerations of crime bring up issues of vulnerability, but they also raise the specter of violence. Women tend to view violence of all sorts more negatively than men do. For example, women are less accepting of violence on television, corporal punishment in schools, and casualties in wartime (Conover and Sapiro 1993; Smith 1984). Women are also more, likely than men to rate violent behavior as seriously criminal (Mills and Bohannon 1992).(FN4) Men are more likely to support policies, such as military spending and military intervention, which involve the use of force. Men are also more likely to choose force in response to public disturbances (Blumenthal et al. 1972) and to condone use of force by police to subdue suspects (Smith 1984). It may be that women's greater discomfort with violence provides a common theme in their greater fear of crime, their greater support for crime prevention measures, and their resistance to harsher punishments.

Gender differences in tolerance for violence relate directly to Carol Gilligan's theory of "different voices" in moral reasoning. Based on her dissatisfaction with earlier studies of moral development,(FN5) Gilligan argues that there are two basic moral perspectives, or "voices." Men's responses to moral quandaries tend to conform to an "ethic of justice," which emphasizes individual rights and relies on "absolute rules of truth and fairness" (Gilligan 1982: 166). Conversely, women's responses tend to suggest an "ethic of care," in which the focus is on "society as an interdependent and interconnected web of personal relationships; from this perspective moral or just behavior entails actively attempting to nurture and protect others and relationships among them" (Worden 1993: 206).

At the base of the ethic of care is concern for protecting the vulnerable. From this perspective, the morally responsible person considers the impact of her actions on those around her, beginning always with the "premise that no one should be hurt" (Bender 1988: 31). In this respect, Gilligan's theory of different voices suggests that there may be more to men's and women's attitudes toward crime than personal vulnerability because preventing crime is an important way to protect others from harm.

Gilligan and others who argue for the ethic of care attribute men and women's different voices to socialization experiences (see, for example, Gilligan 1993; Brody 1990).(FN6) They view gender differences as socially constructed, rather than as essential sex traits. Crawford (1995: 16) explains the process of social construction this way: "Each of us believes in gendered ways because we are placed in gendered social contexts. Women encounter different social contexts from men. Women and men face different expectations and norms even for what look like identical situations.".

Female socialization experiences stress connection and concern for others from early childhood on (Belenky et al. 1986; Chodorow 1978). Brody (1990: 80), for example, argues: "girls are encouraged to identify with, and be dependent on, their mothers. As the ones who are taught to be the nurturers, nurses, homemakers, and 'kinkeepers in the family they constantly receive signals that they should be like their mothers. Boys, on the other hand, are encouraged to be instrumental and active like their fathers." As a result, women tend to see strength in connection and nurturance, rather than in assertion or aggression (Gilligan 1982: 167-68).

In contrast, boys' socialization stresses separation, independence, and autonomy (Chodorow 1978; 167). Schools encourage boys to be more assertive and achievement-oriented than girls (Bailey et al. 1992). Boys are also socialized to be more competitive; rather than concern for attachment, there is concern for "who's better than who, who's bigger than who, who's stronger than who" (Stanko 19901: 110). In addition, boys tend to be disciplined with more "power-oriented techniques" (Mussen and Eisenberg-Berg 1977: 67) and are more routinely exposed to more aggressive behavior, even violence on playgrounds and in sports. They are expected to cope with it by becoming stronger, more self-reliant. In the process, they learn the lesson that "safety comes from strength, backed up by physical ability" (Stanko 1990: 110).

Adult experiences also contribute to gender differences. Men are expected to be self-reliant, and they are generally slower to recognize social interdependence (Doyle and Paludi 1991; McClelland 1975). Women's worldviews are shaped by their greater caregiving responsibilities, for children (Ruddick 1989; Elshtain 1981; Chodorow 1978) and for elderly and disabled relatives (Hooyman and Gonyea 1995; Brody 1990). As Hooyman and Gonyea (1995: 22-23) note, "male-female differences exist in caregiving within the home and the paid workforce, not simply because men and women are treated differently as children, but because they also confront different gender-based socially structured opportunities as adults." These experiences have a significant impact that continues to influence men and women in different ways.

Differences in orientation are apparent by gender in attitudes toward a variety of public policy issues. Women are consistently more supportive of "compassion" policies in ways that conform to the ethic of care. Women express greater sympathy for the disadvantaged (Conover 1994), leading greater numbers of women to favor public expenditures for social welfare, education, and health care (Rhine et al. 1994; Stoper 1988; Burt 1986; Shapiro and Mahajan 1986). Related to this concern for the disadvantaged is women's heightened sensitivity to inequality based on race and ethnicity, which leads women to be more supportive of government efforts to protect racial and ethnic minorities (Rhine et al. 1994). Women are also more likely than men to support environmental regulations, the 55 mile an hour speed limit, mandatory seat belt use, and consumer protection laws (Shapiro and Mahajan 1986). All these programs are designed to afford protection and avoid harm, which links them thematically to the caretaking that underlies women's greater support for compassion policies.(FN7).

We might expect gender differences toward crime and punishment policies to reflect similar themes. Efforts to prevent crime tap into the same constellation of values that underlie the ethic of care. Controlling crime is one of the best ways to protect the vulnerable and prevent victimization. Crime is also a serious disturbance in the web of relational connections; it threatens the safety of neighborhoods that Shapiro and Mahajan (1986) find to be so important to women's policy preferences. This makes it likely that women, with their greater concern for preventing harm, will put the most emphasis on prevention as the answer to crime. Stopping crime before it starts is the best way to keep people from getting hurt.

For many men, on the other hand, the ethic of justice is premised on individual rights and conformity to rules. Efforts to prevent crimes, such as gun control or expanding police powers, may conflict with the ethic of justice's emphasis on individual autonomy by allowing the state to interfere before a law has been broken. Thus, the different voices perspective suggests that women should be more supportive of crime prevention efforts than men.

Yet, while we have good reason to expect women to be more prevention-oriented than men, gender differences in the area of punishment are less easy to predict. On the one hand, given socialization differences, we might expect men to be more punitive than women in response to crime. Men are socialized to be more comfortable with aggression. Boys are more likely to experience power-oriented punishment than girls, and they learn to prove themselves through shows of physical force. And, the ethic of justice emphasizes the need for punishment when the rules are broken: criminals violate the rules, so they must pay a penalty. We also know that men are more comfortable than women with public policy options, such as the commitment of troops, which are premised on the use of violence. Thus, men's greater comfort with aggression, in combination with women's greater distaste for violence, might lead men to be more punitive than women.

There are, on the other hand, sound theoretical reasons to presuppose that gender differences toward punishment may vary contextually. For example, we might expect women to react more punitively when confronting crimes of violence against children, for whom they have particular caretaking responsibilities. They may also advocate responding with greater force to crimes where women are particularly vulnerable, such as rape or domestic abuse. This would be consistent with the findings of Gordon and Riger (1989) and Pierce and Harris (1993) cited above. There are, then, reasons to expect a certain amount of punitiveness from both men and women, according to Gilligan's perspective.

As a caveat, it is important to remember, in exploring gender differences, that the attitudinal variations between men and women are matters of emphasis and degree. There are significant differences between men's and women's attitudes, but these differences are often not particularly large. As Conover and Sapiro (1993: 1095) note, these differences are not large enough to divide women and men into different camps. And, they certainly are not large enough to warrant making the kinds of sweeping statements differentiating women and men that have long been part of the stereotype.

It is nevertheless true that there are important differences between the sexes that need to be understood. The majority of both sexes may support a particular candidate or policy, but they may do so for different reasons or with varying degrees of intensity. For example, Klein (1984: 155) found that majorities of both men and women voted for Carter in the 1976 presidential election, but that they did so for different reasons. She also reports that majorities of both sexes were opposed to government provision of jobs and aid to minorities, but that women were less likely to emphasize self-reliance than men. Similarly, Conover and Sapiro (1993: 1092) found that majorities of both men and women were disgusted by the killing involved in the Persian Gulf but men were less intense in their opposition. We expect to find similar differences of degree in men's and women's approaches to crime and punishment.

METHODS, DATA, AND INSTRUMENTATIONIn the analysis which follows, we first examine gender differences in three areas: fear of crime, attitudes toward crime prevention policies, and attitudes toward punishment for offenders. Importantly, the differences which we subsequently document cannot, alone, provide much leverage with respect to the two explanations discussed above--i.e., vulnerability vs. socialization. First, both of these perspectives predict women to be more supportive of crime prevention policies, albeit for somewhat different reasons. Second, while the vulnerability explanation certainly assumes that women would be more fearful of crime, such a finding would, in no way, be inconsistent with a socialization perspective, such as that articulated by Gilligan. And third, the expectations of both perspectives lead us to be essentially agnostic with regard to punishment: the empirical linkage between fear and punitiveness is far from clear, as is the theoretical relationship between caretaking and punishment. The first section of the analysis, therefore, will be more descriptive than analytical. In the second portion, however, we turn to an analysis of the impact of fear on crime policy attitudes, attempting to address the relative power of these two explanations.

For purposes of analysis, we use data from a probability survey conducted by the University of Kentucky Survey Research Center (UKSRC), which interviewed 501 adults in the greater Lexington, Kentucky area.(FN8) Respondents were selected through a variant of random digit dialing procedures between June 22 and July 5, 1994, and were interviewed by the professional staff at UKSRC for approximately 15 to 20 minutes on matters of crime. Because the survey was initially designed to examine whites' attitudes toward race and crime, only white adults participated.(FN9) All survey items were pretested with a smaller nonprobability sample (N = 410) in April 1994, leading to extensive modifications in the survey items appearing on the final instrument.(FN10).

Given the composition of our sample, we can only be confident that our findings apply to Caucasians. However, we would not be surprised to find a similar gender gap among non-whites. Women of all races are socialized to place greater emphasis on caregiving. We might therefore expect women of all races to be particularly concerned with crime and its prevention. This seems to be borne out by studies such as those by Mills and Bohannon (1992), Miller et al. (1986), and Rossi et. al. (1974), which find that black women consider crime to be a more serious problem than black men do. Our data do not permit an analysis of gender differences among non-whites; our findings may nonetheless suggest possible paths of analysis for others to study.

MEASURESCrucial to this analysis are measures which tap feelings of personal safety and policy attitudes toward crime.(FN11) The four Safety items (Appendix A, Items A.1-4) have been recorded so that maximum values reflect feelings of vulnerability and perceptions of increasing crime rates. The Lexington survey also includes a series of Crime Policy questions, measuring support for both preventive and punitive solutions to crime. Items B.5-8 tap respondent support for crime prevention policies, including spending money for job training programs, creating drug rehabilitation facilities, restricting hand gun purchases, and conducting door-to-door searches for illegal drugs and weapons. While these four policies are designed to prevent crimes from occurring, the four criminal punishment policy items (B.9-12) address policies that are intended to punish the perpetrators once crimes have been committed. Thus, we question respondents about their attitudes toward longer prison sentences for violent criminals, the death penalty, building more prisons, and mandatory life sentences for those who sell drugs to children. Lower scale values on each of these eight items reflect stronger support for the policy, regardless of whether it is preventive or punitive in nature. Finally, the last three items B.13-15) provide an explicit choice to respondents between preventive and punitive policies: between reducing poverty and "cracking down on criminals" (B.13), between rehabilitation and punishment of criminals (B.14), and between social programs and more forceful policing to prevent rioting (B.15). For these last three measures, lower values represent stronger support for more punitive preferences.

ANALYSISAs a first step in the investigation of gender differences, we conduct a series of difference-of-means (t) tests between women and men, the results of which are reported in Table 1. Quite clearly, there is much to suggest heightened feelings of vulnerability among female respondents, as indicated in the first block of items in the table. Women are significantly more likely than men to feel unsafe in their neighborhoods, to judge crime as a more important issue, and to believe there has been a large increase in the nation's violent crime rate over the past five years. While men and women do not differ with regard to perceptions of increases in crime within their own neighborhoods, the differences on the other items, particularly the first, are notable.

It is with the policy items that we find the most interesting differences between genders. We have defined the first four policy measures (5-8) as "preventive," inasmuch as they are designed to minimize the likelihood of crimes being committed. In each of these cases, as anticipated, women are significantly more supportive: they are more likely to favor spending money on job and drug rehabilitation programs, restricting hand guns, and conducting door-to-door searches for drugs and weapons. Differences on the last item, quite possibly, are at least partially a byproduct of the tendency for women to be somewhat less civil libertarian relative to men (e.g., Shapiro and Mahajan 1986).

While women are significantly more supportive of preventive policies across the board, the same cannot be said of their responses to punitive policies (items 9-12). Plainly, the gender differences on both increasing prison terms for those who commit violent crimes and on building more prisons are both insignificant and minimal. And while significant gender differences emerge on the other two items, it is impossible to recognize a consistent pattern, inasmuch as men express more punitive responses to the death penalty question, but less punitive responses when asked about mandatory life prison terms for those who sell drugs to children.

Thus, while women have been found to be universally more supportive of preventive anti-crime policies, gender patterns on punitive measures are clearly context specific. The contexts, moreover, appear consistent with what has been learned about the more general predispositions and concerns of males and females. Numerous studies (e.g., Conover 1994; Shapiro and Mahajan 1986; Smith 1984; and Stinchcombe et al. 1980), as noted, found men to be more supportive of capital punishment, a tendency which, according to many, stems from a more general support among men for policies which involve the use of force or violence. On the other hand, Conover (1994), among others, has also found women to be more sympathetic toward the disadvantaged and the vulnerable--groups which would certainly include children. To find that women are more supportive of life terms for those who sell drugs to children, consequently, suggests that women have the capacity to be more punitive than men, at least when the infraction victimizes an individual or group dependent on the care-taking of women.

This tendency, wherein women are more likely than men to advocate preventive (but not punitive) policy measures, is seen most explicitly in two of the final items of Table 1, both of which pose trade-offs between the utility of social programs to prevent crime versus "get tough" policies to punish perpetrators. Compared to males, females are significantly more likely to see poverty reduction (rather than "cracking down on criminals") as the better solution to crime and, as well, are more likely to feel that riots, such as those occurring in Los Angeles prior to the survey, are better prevented by improved social programs than by a greater use of force.

The differences which we initially anticipated, for the most part, have received empirical support, as women have been found to feel more vulnerable to crime and to be more likely to support preventive anti-crime measures. We also find that men and women do not differ consistently with regard to punitive policy items, quite possibly because women are conflicted in their beliefs. On the one hand, they are more concerned with crime and more likely to want to correct the problem. On the other hand, however, they are more likely to resist the use of force, as suggested by their stronger opposition to the death penalty.

MULTIVARIATE ANALYSISFindings, to this point, must be regarded as preliminary, inasmuch as more fully specified models may reveal the gender-based differences to be spurious. We know, for example, that liberals (and, to a lesser extent, Democrats), tend to be more supportive of preventive policies, and less supportive of punitive policies, relative to conservatives (and Republicans). To the degree that gender is correlated with ideology and partisanship, then, it is conceivable that ideology, rather than gender, is the basis for the attitudinal differences found to this point. Older respondents, moreover, tend to be more fearful of crime in our sample; given the slightly older age of female respondents relative to males,(FN12) it is possible that age, rather than gender, motivates the higher levels of fear reported by women.

Of equal concern is the possibility that the gender differences reported in Table 1 are reflective of potential social status distinctions between men and women. Wealthier and better educated individuals tend to feel safer and less concerned with crime (doubtless because they do not often live in the higher crime neighborhoods). Inasmuch as female respondents are both less affluent and less politically knowledgeable than males,(FN13) it is important to control for socioeconomic variables in analyzing gender differences on crime.

To examine the relationship between gender and anti-crime attitudes more rigorously, we constructed regression equations for all the items for which there were significant differences by gender. In addition to gender, we included measures of ideology, party identification, age, education, political sophistication, and income,(FN14) as independent variables--variables which, for reasons noted above could potentially alter the zero-order relationships found in Table 1. In Table 2, column 1, we report the unstandardized regression coefficients for the gender dummy variables, which are interpretable as the mean differences between males and females for each item, while holding constant levels of political predisposition (ideology and party), age, education, political knowledge, and affluence.

Despite the introduction of other controls, gender continues to be a significant predictor of most attitudes toward crime and punishment. The gender differences reported in Table 1 remain, with only two exceptions: we no longer find significant differences between men and women on the life-prisonterm item, and on the importance-of-crime-as-a-problem item. Otherwise, all differences revealed in the initial t-tests remain statistically significant when controlling for a host of political and demographic variables.

The full regression models (see Appendix B) demonstrate the true importance of gender in shaping crime policy attitudes. It can be seen from the standardized coefficients that, not only does gender remain statistically significant in nine of the eleven equations, it is also the single strongest determinant in five of the models and among the most important predictors in most of the others. This is not to suggest that other factors play no role: clearly older age serves to make respondents feel more vulnerable to crime (see equations 1, 2, and 4), and liberals (and, to a lesser extent, Democrats) consistently endorse preventive policies while opposing punitive policies. Other variables, moreover, are selectively important: affluence, for example, is the strongest determinant of support for the death penalty. It is clearly the case, however, that no other variable plays as consistently an important role in shaping crime policy attitudes (and feelings of vulnerabilityto crime) as does gender.

THE ROLE OF VULNERABILITYBut how do we know that women's responses are not merely an artifact of their own greater fear? It was our initial contention that women should be relatively more vigilant about preventing crime because of the caretaking role into which many women have been socialized in our society, not just because of their heightened feelings of vulnerability. While we cannot directly test the effect of socialization, we can address the impact of fear. Specifically, if a sense of vulnerability underlies the consistent support among women to endorse crime prevention policies, then we should find that male-female attitudinal differences disappear at comparable levels of security. Put differently, if the gender coefficients found in Table 2 remain significant after introducing controls for feelings of safety, we could reasonably conclude that this sense of vulnerability is not the key factor in encouraging the crime prevention sentiments of females.

In the second column of Table 2 (and in Appendix C), we reestimate the same regression equations found in column 1, differing only in the sense that we now control for respondent's fear of crime by adding Item A.1 (the "Safe" variable) as an independent variable. The most relevant finding for our purposes is that neither the significance nor the magnitude of the gender coefficient declines once we control for feelings of vulnerability, providing strong evidence that women are relatively more supportive of crime prevention policies for reasons that are independent of their perceptions of personal threat. The explanation for women's stronger support for crime prevention efforts, in other words, seems to have less to do with their collective sense of vulnerability than with their commitment to protecting the vulnerable and preventing harm.

Before dismissing the possibility that fear of crime motivates crime policy attitudes, however, it is necessary to recall a persistent finding in the literature--viz, that fear of crime and punitiveness are related for men, but not for women (e.g., Feraro 1995; Mills and Bohannon 1992; Stinchcombe et al. 1980). Given this pattern, it is possible that the impact of safety on policy attitudes has been "canceled out" by pooling both genders in our analysis, as in the second column of Table 2. In Table 3, therefore, we replicate the equations from Table 2 (second column), but separately for male and female respondents. To examine the precise impact of perceived safety on policy attitudes, we report coefficients representing the Safe variable on the various policy attitudes. The first column includes all respondents, while the second and third columns include only males and females, respectively. Entries are unstandardized coefficients, which are useful for comparing across the various regression models.

The findings of Table 3 bolster our conclusion that fear of crime does not appear to be an important antecedent of respondent crime policy attitudes, particularly in the case of women: among females, not a single policy attitude is significantly predicted by perceived safety. With men, the threat of crime is somewhat more instrumental, as reported in the extant literature. In each equation but one (mandatory life prison terms to those who sell drugs to children), the impact of safety is stronger for males than for females. In two instances, moreover, men who are more afraid of crime are significantly more punitive than those who are less fearful--they are more likely to reject the argument that reducing poverty is preferable to "cracking down on criminals" for reducing crime, and more likely to advocate forceful police measures (rather than improved social programs) for the purpose of preventing riots.

Our basic contention, however, remains unchanged. While men may be somewhat influenced in their crime policy attitudes by their level of fear, women are not. We remain convinced that the best explanation for women's policy positions is not associated with their heightened fears of crime but, instead, that the manner in which they have been socialized has led them, in many instances, to stronger support for measures which protect themselves, their families, and their neighbors from the menace of crime.

DISCUSSION AND CONCLUSIONSAs with any argument, we are more certain of some parts than of others. We are, for the most part, fully convinced of the attitudinal differences between men and women pertaining to both feelings of vulnerability to, and remedies for, crime. As shown, women are more likely to be concerned with the threats posed by crime, and are more likely to emphasize prevention in their responses to them. We want to emphasize that men and women are not radically divided in their attitudes toward crime. As in Conover and Sapiro's (1993) study of attitudes toward war, we do not find women and men in wholly different camps. Majorities of both sexes agree on most of the items we explored. What we find are differences of degree: women are more inclined toward protection than men, even though both desire to control the crime problem. Men and women agree that crime is a serious problem, but women are more afraid and more upset by the aggression that is employed, both in committing and in punishing it.

Quite clearly, gender is not the sole explanation of crime-related attitudes. The full regression models (see Appendix B) demonstrate the importance of ideology, income, age, and other considerations. The proportion of variance which we are unable to explain, moreover, underscores the likelihood that these models have been underspecified and, consequently, that important determinants of crime policy attitudes have been ignored. Nonetheless, the consistently important role which gender has been found to play in these models provides us with a great deal of confidence in concluding that men and women approach the issue quite differently.

We can be less sure about the reasons behind these gender differences. We earlier suggested two possible scenarios: first, that women are more vulnerable to, and fearful of crime and, consequently, more adamant about preventing it; and second, that women have been socialized into a care-taking role. The analyses in Table 2 (second column) and 3 were designed to test the vulnerability thesis and demonstrated to our satisfaction that fear of crime does not drive crime policy attitudes. To the contrary, when controlling for perceptions of fear of crime, gender differences were equally strong as in models without the fear variable--i.e., at constant levels of vulnerability, men and women still differ in their responses to crime.

Almost by default, then, we have settled on a socialization argument to explain gender differences--an argument which outstrips the data which are available. Minimally, then, we must wrestle with several questions before putting too much weight on socialization as a source of these differences.

First, it is worth asking whether the tests of the vulnerability hypothesis were sufficiently strong. As revealed in both Tables 1 and 2 (first column), there is little doubt that women feel more susceptible to, and threatened by, crime. We did not find, however, that this heightened fear translates into stronger preferences for preventive anti-crime policies for, as noted above, controlling for fear does not eliminate differences between men and women. It is important to note that, in selecting the control variable, we deliberately selected the item ("How safe do you feel ... ?") most closely correlated with gender, thereby providing the most stringent test possible of the impact of holding fear constant on the elimination of gender differences.(FN15).

Still, we acknowledge that vulnerability to crime (and the resultant fear) can be manifest in numerous other ways that will not be evident from items on public opinion surveys. Marital status (i.e., being single), having children to protect, or living in urban (rather than suburban or rural) areas may all serve to heighten feelings of vulnerability and, consequently, to engender preferences for crime prevention policies. To the degree that these (or other) life status factors are associated with gender, it is possible both that we have underestimated the impact of fear and that we have overestimated the role of gender. While marital status and urban residential rates are virtually identical for men and women,(FN16) rates of single parenthood are not: the most recent decennial census found that single mothers outnumber single fathers more than five to one (U.S. Bureau of the Census 1996). Most generally, then, we recommend caution in ruling out the possibility that fear of crime motivates some of the attitudinal differences which we have attributed to gender, and express some regret at not having information pertaining to children (and other forms of vulnerability) on the Lexington survey. Quite conceivably, women are more preemptive with crime because they feel more vulnerable--in ways which our instruments have not been able to determine.

If our data are to be believed, however, the more likely explanation is that women and men have been socialized differently. The gender gap in attitudes toward crime which we have found are consistent with the ethics of care and justice suggested by Gilligan. For women, preventing crime appears to be part of a larger concern for protecting the vulnerable and making sure that no one is hurt. Preventing crime is important to men as well, but they put relatively less emphasis on prevention than punishment when asked to choose between the two. From the female perspective, punishment has its place, especially in the protective context of harm to children, but it is normally more important to avoid pain in the first place, rather than attempting to compensate victims by causing suffering for those who have victimized them. In this respect, women's attitudes toward crime resemble their greater support for social welfare spending, consumer protection, and environmental regulation. Men's greater support for the death penalty, the ultimate punishment, ties in well with their support for other "force and violence" policies, such as defense spending and military intervention, that appeal considerably less to women.

Our data do not allow us to measure the effects of respondents' socialization or moral voice directly. We can merely infer that these forces are at work here. We do, however, have some uncertainties. In the first place, we might expect to find an interaction between age and gender--i.e., that women would differ attitudinally from men, but mainly among respondents who are older and presumably socialized under more rigidly defined sex roles. Working with a series of interactive models, however, we were unable to uncover such an effect: gender differences were found to be relatively uniform regardless of age.(FN17) While there are many ways to "explain away" such results (e.g., gender-based socialization patterns may persist in the South) the results still warrant caution.

There are, moreover, numerous norms, values, beliefs, and doctrines which are inherent to the socialization process that have little to do with gender and which, potentially, may have a profound impact on crime policy attitudes. Different religions, for example, may preach more punitive, or more forgiving, responses to law-breaking. We unfortunately do not have the data necessary to rule out the possibility that male-female attitudinal differences stem from different religious orientations or, for that matter, from other group or doctrinal affiliations.(FN18) For these reasons, attributions of gender differences to the socialization process must remain speculative.

Thus, while we remain quite confident of the gender differences which we find, the precise reasons for these differences must remain hypothetical. But this does not minimize the importance of the findings, which extends beyond the debate on gender. To the contrary, our results have implications for students of both legislative and electoral politics. In the legislative arena, they provide grounds for questioning the often-stated assumption that crime policy in the United States will abandon preventive efforts and embrace increasingly punitive anti-crime measures. At the same time that Democrats and Republicans attempt to "out tough" one another by proposing ever-more-harsh policies in response to crime, the public is electing a steadily increasing number of female legislators (see, for example, Mezey 1994). If the attitudinal differences which we find translate into behavioral differences when women become policymakers,(FN19) we should see more support for preventive policies in state and national legislatures, assuming, of course, that the electorate continues to elect more females to make these choices.

The electoral implications are equally intriguing. To the degree that concerns about crime filter into the public's voting decisions, both political parties may find themselves in something of a dilemma. Since 1980, Republicans have been at the short end of the so-called "gender gap"--a phenomenon in which women (relative to men) have been notably less supportive of both Republican candidates and their issue positions. At the same time, the party has benefited from citizen perceptions that their party is the one more capable of dealing with the crime problem. In an attempt to exploit this issue, Republicans have decried preventive anti-crime policies as "pork," employing tough anti-crime rhetoric and advocating only those policies that emphasize severe punishment for lawbreakers. The paradox, of course, is that the very policies which the Republicans exploit for electoral gain (among white males) may also turn out to be the same policies that widen the gender gap by alienating female voters. For Democrats, the dilemma is that they must somehow appear "tough on crime" without losing the support of women in the electorate--women who may find the position unattractive. In an effort to court women voters, therefore, candidates of both parties may find it useful to champion more crime prevention policies.

Added material.

JON HURWITZ AND SHANNON SMITHEY, UNIVERSITY OF PITTSBURGHThe authors wish to acknowledge the assistance provided by Bert Rockman and the University of Pittsburgh Center for American Politics and Society, as well as that offered by the Editors and anonymous referees. We are particularly indebted to Mark Peffley for his involvement in all stages of this project. The order of the authors was determined by a coin flip.

TABLE 1 MALE-FEMALE COMPARISONS ON ATTITUDES TOWARD SAFETY AND CRIME POLICY.

(TABLE) VARIABLE MEANS T VALUESSAFETY1. How Safe in Neighborhood? Male 1.68 7.327(FN**) Female 2.272. How Important is Crime? Male 2.06 1.881(FN*) Female 2.183. Crime in Neighborhood Male 2.44 0.623 Female 2.484. Crime in Nation Male 3.42 4.674(FN**) Female 3.75CRIME POLICY5. Spend More for Jobs Male 2.13 4.271(FN**) Female 1.736. Drug Rehabilitation Male 1.84 2.851(FN**) Female 1.627. Restrict Hand Guns Male 2.31 7.375(FN**) Female 1.578. Searches for Drugs Male 3.11 3.477(FN**) Female 2.779. Increase Prison Terms Male 1.56 1.433 Female 1.4510. Death Penalty Male 1.69 3.276(FN**) Female 1.9911. Build More Prisons Male 2.88 0.244 Female 2.8612. Life Prison Terms Male 1.66 2.236(FN*) Female 1.4813. Reduce Poverty Male 3.03 3.559(FN**) Female 3.3314. Rehabilitate or Punish? Male 2.49 0.820 Female 2.5715. Prevent Riots Male 1.38 3.915(FN**) Female 1.56.

FOOTNOTES* p < .05.

** p < .01.

Tests are based on samples of 465 to 501 respondents. For the Safety items (1-4), larger values reflect feelings of vulnerability and perceptions of increasing crime. Crime Policy items 5-12 have been recorded so that lower values correspond to stronger support for the policy measure. Items 13-15 present trade-offs to respondents and are coded such that lower values represent stronger support for more punitive preferences.

TABLE 2 REGRESSION MODELS.

(TABLE)VARIABLE GENDER b GENDER b (w/out Safe) (with Safe)SAFETY 1. How Safe in Neighborhood? .395(FN**) 2. How Important is Crime? .027 4. Crime in Nation .309(FN**)CRIME POLICY 5. Spend More for Jobs -.289(FN**) -.318(FN**) 6. Drug Rehabilitation -.200(FN**) -.182(FN*) 7. Restrict Had Guns -.627(FN**) -.661(FN**) 8. Searches for Drugs -.209(FN*) -.216(FN*) 10. Death Penalty .282(FN**) .315(FN**) 12. Life Prison Terms -.134 -.110 13. Reduce Poverty .214(FN**) .252(FN**) 15. Prevent Riots .184(FN**) .200(FN**).

FOOTNOTES* p <.05.

** p <.01.

All models are based on 439 cases. In addition to gender, independent variables include ideology, party identification, age, education, political information, income and, in the last column, safety.

TABLE 3 THE IMPACT OF FEAR OF CRIME ON POLICY ATTITUDES.

(TABLE)VARIABLE ALL MALES FEMALESSpend More for Jobs .035 .121 .052Drug Rehabilitation .080 .155(FN*) .044Restrict Hand Guns .007 .121 .021Searches for Drugs .006 .053 .017Death Penalty .045 .119 .064Life Prison Terms .082 .006 .102Reduce Poverty .070 .148(FN*) .069Prevent Riots .008 .191(FN**) .039.

FOOTNOTES* p < .05.

** p < .01.

All models are based on 439 cases. Table entries are unstandardized regression coefficients representing the impact of the safety variable on policy attitudes. Other independent variables include ideology, party identification, age, education, political information, and income.

FOOTNOTES1 The difference in attitudes is often much smaller than the difference by race. As an example of this point, see Bedyna and Lake 1994.

2 For example, interpersonal violence (Walklate 1995), responsibility for children and aging parents (Brody 1990), and legislative priorities (Mezey 1994).

3 This appears to be a common assumption. For example, Stinchcombe et al. (1980: 66) note, "We punish criminals partly because we think criminals deserve to suffer for their wrongdoings but primarily because we think that punishment will somehow reduce crime.".

4 Women disapprove of the use of force or violence in a number of arenas (see Smith 1984; Shapiro and Mahajan 1986). For example, in response to reading stories describing instances of wife battering, male college students were much more likely to consider a husband's use of force justified, and female college students were much more likely to assign culpability and advocate convictions for batterers (Pierce and Harris 1993).

5 Prior to Gilligan, women's responses to moral dilemmas tended to rate lower on scales of moral maturity because they were more contextualized than the "rights and rules" approach which characterized mens answers.

6 Gilligan has been accused of advocating a theory of biological determinism. For this and other critiques of Gilligan's work, see, for example, Larrabee 1993; Friedman 1987.

7 Millbrath (1984) and Steger and Witt (1989) pay particular attention to women's greater support of environmentalism, which they attribute to women's greater perception of risks (to health, from pollution, and so forth) and the environmental emphasis on "the interconnectedness of all things (Steger and Witt 1989: 629).

8 According to the Uniform Crime Reports data published annually by the Federal Bureau of Investigation (United States Federal Bureau of Investigation 1995), Lexington is a relatively typical American city in terms of its crime rates. For the 1994 year (when our data were collected), the rate of total crimes, violent crimes (murder, forcible rape, robbery, and aggravated assault), and property crimes (burglary, larceny-theft, and motor vehicle theft) per 100,000 inhabitants in the Lexington Metropolitan Statistical Area was 5,770.1, 921.5, and 4,848.6, respectively. These compare to national rates of 5,374.4, 716.0, and 4,658.3, respectively; rates for metropolitan areas for the same year were: 5,894.2, 812.0, and 5,082.2, respectively. Thus, compared to other urban areas of the United States, Lexington has a slightly lower overall crime rate (97.9 percent of the national urban rate) and property crime rate (95.4 percent), but a somewhat higher rate of violent crime (13.5 percent above the national urban rate). Most of the discrepancy in violent crime can be explained in terms of Lexington's high rate of aggravated assault crimes; its property crime statistics are suppressed by the very low rates of motor vehicle theft. While the FBI data do not include measures of dispersion necessary for the calculation of significance tests, an examination of the state-by-state statistics indicates that Lexington crime rates are quite close to both national and urban averages, considering the wide inter-state ranges. Overall crime per 100,000, for example, ranged from 8,250.0 (Florida) to 2,528.4 (West Virginia). High (and low) rates for violent crime: 1,146.8 in Florida (81.8 in North Dakota); for property crimes: 7,221.4 in Arizona (2,312.6 in West Virginia).

9 The study was originally designed to explore the effect of racial stereotypes on attitudes toward crime (see Hurwitz and Peffley 1997). As such, only Caucasians were included in the sample.

10 Although the representation of women in the sample essentially mirrors the proportion in the Lexington population, the sample overrepresents more educated and affluent individuals. The median income in the 1990 Census is $17,941, while the median income category in the sample is $20,001 to $30,000. The population (and sample) percentages of educational groups are as follows: did not complete high school--19.8 percent in the population (7.6 percent in the sample); completed high school--23.4 percent (24.9 percent); attended college without graduating--26.2 percent (24.7 percent); graduated college--18.6 percent (23.9 percent); graduate work--12.0 percent (18.1 percent). Accordingly, the sample was weighted to reflect education levels in the Lexington population. While the weighted sample should provide results which are more generalizable to the Lexington population, our findings do not differ significantly depending on whether the sample is weighted or unweighted.

11 Survey items (and means) are included in Appendix A.

12 The average age for female and male respondents is 47.1 and 43.2 years, respectively (p < .01).

13 Among our respondents, gender is modestly correlated with education (r = .10, p < .05), and more substantially correlated with income (r = .21, p < .01) and our measure of political information (r = .27, p, .01). Females, in other words, have lower levels of education, income, and political knowledge, relative to males.

14 Ideology and party identification are measured with standard 7-point self-placement scales. Educated is coded as: 1 = 0 to 4 years; 2 = 5 to 8 years; 3 = 9 to 11 years; 4 = (high school graduate; 5 = some college; 6 = college graduate; 7 = graduate work. Political sophistication is assessed with a 5-item knowledge index, based on the respondent's knowledge of: (1) Al Gore's present office; (2) the branch of government responsible for determining the constitutionality of a law; (3) the size of the majority necessary for the Senate and House of Representatives to override a presidential veto; (4) the party which controlled the House of Representatives prior to the 1992 election; and (5) the more conservative of the political parties (Cronbach's alpha = .739). Income is coded as: 1 = under $5,000; 2 = $5,000 to $10,000; 3 = $10,001 to $20,000; 4 = $20,001 to $30,000; 5 = $30,001 to $50,000; 6 = $50,001 to $100,000; 7 = over $100,000.

15 Additional analyses (unreported), using the other three safety measures (see Appendix A, items A.2-4) both individually and simultaneously as control variables, did not result in diminished gender differences.

16 Census statistics indicate that 62.7 percent of men, and 59.2 percent of women are married; and that 29.3 percent of men, and 27.7 percent of women live in cities (U.S. Bureau of the Census 1996).

17 We estimated both linear and nonlinear interaction terms, given the possibility of age cohort effects. Neither form yielded significant interaction coefficients. We also searched, without success, for interactions between gender and political sophistication, education, income, ideology, and party identification.

18 We are somewhat skeptical of religious explanations of gender differences, inasmuch as they are generally inconsistent with our findings. Catholicism, for example, has long taken a position against the death penalty. The most recent American National Election Survey found higher levels of catholicism among men (31.7 percent) than among women (24.7), leading to the prediction that men would be more opposed to the death penalty. We, however, find the opposite to be true.

19 Numerous studies (Mezey 1994; Thomas 1994; Reingold 1992; among others) suggest a relationship between gender and policy priorities. Reingold (1992), for example, finds that female legislators are more likely than their male counterparts to introduce and support legislation of particular interest to women, and that female legislators of both parties were more conscious of representing women's interests than were their male colleagues.

REFERENCESBailey, Susan M. 1992. The AAUW Report: How Schools Short Change Girls. Washington, DC: National Education Association.

Baxter, Sandra, and Marjorie Lansing. 1985. Women and Politics: the Visible Majority. Ann Arbor: University of Michigan Press.

Belenky, Mary Field, et al. 1986 Women's Ways of Knowing: The Development of Self, Voice and Mind. New York: Basic Books.

Bender, Leslie. 1988. "A Lawyer's Primer on Feminist Theory and Tort." Journal of Legal Education 38: 3-38.

Bedyna, Mary E., and Celinda C. Lake. 1994. "Gender and Voting in the 1992 Presidential Election." In Elizabeth Adell Cook, Sue Thomas, and Clyde Wilcox, eds., The Year of the Woman: Myths and Realities. Boulder, CO: Westview Press.

Blumenthal, M., et al. 1972. Justifying Violence. Ann Arbor: Institute for Social Research.

Brody, Elaine M. 1990. Women in the Middle: The Parent Care Years. New York: Springer.

Burt, Sandra. 1986. "Different Democracies? A Preliminary Examination of the Political Worlds of Canadian Men and Women." Women and Politics 6: 57-79.

Chodorow, Nancy. 1978. The Reproduction of Mothering. Berkeley: University of California Press.

Conover, Pamela Johnston. 1994. "Feminists and the Gender Gap." In Marianne Githens, Pippa Norris, and Joni Lovenduski, eds., Different Roles, Different Voices: Women and Politics in the United States and Europe. New York: HarperCollins.

Conover, Pamela Johnston, and Virginia Sapiro. 1993. "Gender, Feminist Consciousness and War." American Journal of Political Science 37: 1079-99.

Crawford, Mary. 1995. Talking Difference: On Gender and Language. Thousand Oaks, CA: Sage.

Deitch, Cynthia. 1988. "Sex Differences in Government Spending." In Carol Mueller, ed., The Politics of the Gender Gap: The Social Construction of Political Influence pp. 192-216. Beverly Hills, CA: Sage.

Doyle, James A., and Michele A. Paludi. 1991. Sex and Gender, 2nd ed. Dubuque, IA: William C. Brown Publishers.

Elshtain, Jean Bethke. 1981. Public Man, Private Woman: Women in Social and Political Thought. Princeton, NJ: Princeton University Press.

Ferraro, Kenneth. 1995. Fear of Crime: Interpreting Victim Risk. Albany: SUNY Press.

Friedman, Marilyn. 1987. "Beyond Caring: the Demoralyzing of Gender." Canadian Journal of Philosophy Supplementary Volume 13: 87-110.

Gilens, Martin. 1988. "Gender and Support for Reagan: A Comprehensive Model of Presidential Approval." American Journal of Political Science 32: 19-49.

Gilligan, Carol. 1982. In a Different Voice: Psychological Theory and Women's Development. Cambridge, MA: Harvard University Press.

Gilligan, Carol. 1993. "Reply to Critics." In Mary Jeanne Larrabee, ed., An Ethic of Care: Feminist Interdisciplinary Perspectives, pp. 207-14. New York: Routledge.

Goldberg, Gertrude S., and Eleanor Kremen. 1994. "The Feminization of Poverty: Not Only in America." In Marianne Githens, Pippa Norris, and Joni Lovenduski, eds., Different Roles, Different Voices: Women and Politics in the United States and Europe. New York: HarperCollins.

Gordon, Margaret T., and Stephanie Riger. 1989. The Female Fear. New York: Free Press.

Hooyman, Nancy R., and Judith Gonyea. 1995. Feminist Perspectives on Family Care: Policies for Gender Justice. Thousand Oaks, CA: Sage.

Hurwitz, Jon, and Mark Peffley. 1997. "Public Perceptions of Race and Crime: The Role of Racial Stereotypes." American Journal of Political Science 41: 375-401.

Karmen, Andrew A. 1991. "Victims of Crime." In Joseph F. Sheley, Criminology: A Contemporary Handbook, pp. 121-38. Belmont, CA: Wadsworth.

Klein, Ethel. 1984. Gender Politics. Cambridge, MA: Harvard University Press.

Larrabee, Mary Jeanne, ed. 1993. An Ethic of Care: Feminist Interdisciplinary Perspectives. New York: Routledge.

McClelland, David. 1975. Power: The Inner Experience. New York: Irvington.

Mezey, Susan Gluck. 1994. "Increasing the Number of Women in Office: Does it Matter?" In Elizabeth Adell Cook, Sue Thomas, and Clyde Wilcox, The Year of the Woman: Myths and Realities. Boulder: Westview Press.

Milbrath, Lester. 1984. Environmental Vanguard for a New Society. Albany: State University of New York Press.

Miller, Arthur. 1988. "Gender and the Vote: 1984." In Carol Mueller, ed., The Politics of the Gender Gap: The Social Construction of Political Influence, pp. 258-82. Beverly Hills, CA: Sage.

Miller, J. L., Peter H. Rossi, and Jon E. Simpson. 1986. "Perceptions of Justice: Race and Gender Differences in Judgments of Appropriate Prison Sentences." Law and Society Review 20: 313-34.

Mills, Carol J., and Wayne E. Bohannon. 1992. "Juror Characteristics: To What Extent are They Related to Jury Verdicts?" In Elliot E. Slotnick, ed., Judicial Politics: Readings from Judicature, pp. 261-68. Chicago: America Judicature Society.

Mueller, Carol M. 1988. The Politics of the Gender Gap: The Social Construction of Political Influence. Beverly Hills, CA: Sage.

Mussen, P., and N. Eisenberg-Berg. 1977. Roots of Caring, Sharing and Helping. San Francisco: Freeman.

Norris, Pippa. 1985. "The Gender Gap in Britain and America." Parliamentary Affairs 38: 192-201.

Pierce, Maureen C., and Richard Jackson Harris. 1993. "The Effects of Provocation, Race and Injury Description on Men's and Women's Perceptions of a Wife-Battering Incident." Journal of Applied Social Psychology 23: 767-90.

Pomper, Gerald. 1975. Voter's Choice. New York: Dodd, Mead.

Reingold, Beth. 1992. "Concepts of Representation Among Female and Male State Legislators." Legislative Studies Quarterly 17: 509-37.

Research and Forecasts Inc. with Andy Friedburg. 1983. America Afraid: How Fear of Crime Changes the Way We Live. New York: American Library.

Rhine, Staci, Shannon Ishiyama Smithey, and Audrey Haynes. 1994. "Cross-National Differences in Political Attitudes: The Effects of Gender in Canada." Paper presented at the Annual Meeting of the Midwest Political Science Association, Chicago, IL.

Rossi, Peter H., Emily Waite, Christine E. Bose, and Richard E. Berk. 1974. "The Seriousness of Crimes: Normative Structure and Individual Differences." American Sociological Review 39: 224.

Ruddick, Sara. 1989. Maternal Thinking: Towards a Politics of Peace. Boston: Beacon Press.

Shapiro, Robert Y., and Harpreet Mahajan. 1986. "Gender Differences in Policy Preferences: A Summary of Trends from the 1960s to the 1980s." Public Opinion Quarterly 50: 42-61.

Sidel, Ruth. 1986. Women and Children Last. New York: Penguin Books.

Smith, Tom W. 1984. "The Polls: Gender and Attitudes Toward Violence." Public Opinion Quarterly 48: 384-96.

Stanko, Elizabeth. 1990. Everyday Violence. London: Pandora Press.

Steger, Mary Ann E., and Stephanie L. Witt. 1989. "Gender Differences in Environmental Orientations: A Comparison of Publics and Activists in Canada and the U.S." Western Political Quarterly 42: 627-49.

Stinchcombe, Arthur, Rebecca Adams, Carol A. Heimer, Kim Lane Scheppele, Tom W. Smith and D. Garth Taylor. 1980. Crime and Punishment--Changing American Attitudes in America. San Francisco: Jossey-Bass.

Stoper, Emily. 1988. "The Gender Gap Concealed and Revealed: 1936-1984." Paper presented at the annual meeting of the Western Political Science Association, San Francisco.

Thomas, Sue. 1994. How Women Legislate. New York: Oxford University Press.

United States Bureau of the Census. 1996. Statistical Abstract of the United States: 1996, 116th ed. Washington, DC.

United States Federal Bureau of Investigation (U.S. Department of Justice). 1995. Crime in the United States 1994: Uniform Crime Reports. Washington, DC: U.S. Government Printing Office.

Walklate, Sandra. 1995. Gender and Crime: An Introduction. New York: Prentice Hall.

Worden, Alissa Pollitz. 1993. "The Attitudes of Women and Men in Policing: Testing Conventional and Contemporary Wisdom." Criminology 31: 203-41.

Received: December 13, 1996.

Accepted: August 19, 1997.

APPENDIX A: LEXINGTON SURVEY MEASURES (AND SAMPLE MEANS)A. SAFETY1. How safe do you feel, or would you feel, being out alone in your neighborhood at night? Very safe, somewhat safe, somewhat unsafe, or very unsafe? (2.03).

2. Of all the problems facing the country today, such as the deficit, the economy, and troubles overseas, would you say that the problem of crime is no more important than many other problems, is one of the most important problems, or is the single most important problem our nation faces today? (2.13).

3. Over the past five years or so, would you say that crime in your neighborhood has increased a lot, increased a little, decreased, or stayed about the same? (2.47).

4. Over the past five years or so, would you say that violent crime in our nation has increased a lot, increased a little, decreased, or stayed about the same? (3.61).

B. CRIME POLICY ATTITUDES:(FN**)CRIME PREVENTION5. We need to spend more money on programs to help poor people; blacks and minorities find jobs so they can earn a decent living without turning to crime. (1.90).

6. Creating more drug rehabilitation programs for people in these neighborhoods.(FN*) (1.71).

7. One good way to reduce violence from hand guns is to place more restrictions on the purchase of these weapons. (1.88).

8. Conducting door to door searches in these neighborhoods for illegal drugs and weapons.(FN*) (2.91).

CRIMINAL PUNISHMENT9. The best way to deal with violent crime is to dramatically increase prison terms for people who commit violent crimes. (1.50).

10. I strongly favor the death penalty for anyone convicted of murder. (1.86).

11. We need to build more prisons to keep criminals off the streets, even if it means increasing taxes. (2.87).

12. Imposing mandatory life prison terms for anyone caught selling drugs to children. (1.56).

PREVENTION VS. PUNISHMENT13. To really reduce crime in this country we need to focus more on reducing poverty instead of just cracking down on criminals. (3.20).

14. In dealing with criminals, it is more important to rehabilitate them than to punish them. (2.54).

15. People we talk to have different opinions about the riots that occurred in Los Angeles. If you had to choose, which do you think would be the best way to prevent disturbances like the LA riots in the future? Do you think there should be better social programs in poor and minority neighborhoods or do you think police should act more forcefully to put down any disturbances? (1.48).

FOOTNOTES* These items were preceded by the introduction: Here are some things that law enforcement might do to deal with the problems of drugs and crime in mostly black; white neighborhoods in the inner city. For each one, please tell me whether you strongly approve, somewhat approve, somewhat disapprove, or strongly disapprove.

** With the exception of item 15 (which is dichotomous), responses to all items in Part B are measured on four-point scales ranging from Strongly Agree to Strongly Disagree.

APPENDIX B: COMPLETE REGRESSION MODELS.

(TABLE) 1 2 4 5 6 7Gender .204** .205** -.141** -.120** -.272**Ideol -.086* -.179** -.140**Party Id -.101* -.082* Age -.097* -.081* -.101* Educ -.110* -.221* Info -.137** .177**Income -.196** .137*R2 .247 .134 .087 .091 .072 .168R2Adj .233 .118 .071 .075 .055 .153N 439 439 439 439 439 439.

(TABLE) 8 10 12 13 15Gender -.095* .133** .119** .189**Ideol .084* .127** .128** .143** .115*Party Id .085* .144** .111* Age .107* .116** .149** Educ .194** .188** .146* InfoIncome -.211*R2 .104 .100 .125 .108 .117R2Adj .088 .083 .109 .092 .101N 439 439 439 439 439.

Table entries are standardized beta coefficients and include only those variables significant at p < .05* (p < .01**). The column heading numbers correspond to the numbers representing the dependent variables in Table 2. The independent variables have been coded so that lower values correspond to: males; conservatives; Republicans; younger, less educated, less informed, less affluent, more highly religious, and more safe, individuals.

APPENDIX C: COMPLETE REGRESSION MODELS, INCLUDING SAFETY.

(TABLE) 5 6 7 8 10 12 13 15Gender -.156** -.109* -.286** -.098* .149** .140** .200**Ideol -.183** -.136** .085* .122* .124** .137** .112**Party ID -.101* -.082* .085* .143** .111* Age .109* .109* .144** Educ .185** .180** -.109* .140* Info .109* .187**Income .151** -.226** Safe -.102*R2 .095 .074 .173 .105 .105 .128 .116 .119R2 Adj .076 .055 .155 .086 .086 .110 .098 .101N 439 439 439 439 439 439 439 439.