Research suggests there has been a rise in the number of hate crimes since 1985. At the same time, legislatures at the local, state, and national level have enacted policies that both track and regulate hate crime. This article is an effort to determine the factors influencing hate crime policy and implementation efforts. The project is divided into three sections: In the first section, the characteristics and extent of hate crime are discussed. Section two describes hate crime policy as social regulatory policy and uses this theoretical framework to explain state variation in laws concerning hate crimes. In section three, I present a model of policy implementation to predict state implementation efforts of federal hate crime policy. Based on the variables suggested by these theoretical frameworks, I present hypotheses and conduct a multiple regression analysis using a fifty-state data set. The results indicate hate crime policies and implementation efforts are largely attempts by politicians to satisfy organized interests in competitive political systems. I discuss the implications of these findings and suggest avenues for future research.
Everyday in the United States someone is attacked on the basis of his or her race, religious affiliation, ethnicity, gender, or sexual orientation. These attacks often take the form of verbal harassment but some end in violent assault or death. Recent studies indicate a rise in the number of "bias" or "hate motivated" crimes since 1985 (Comstock 1991; Jost 1993b; McDevitt and Levin 1993; Lutz 1987; NGLTF Policy Institute 1993,1994) and an apparent resurgence of hate groups (Anti-Defamation League 1988; Hamm 1993).
At the same time that hate crimes appear to be on the rise, a majority of the American states have passed laws that regulate bias motivated criminal behavior. The passage of these statutes is largely viewed as a response to increasing levels of hate crime, but an empirical relationship has not been established. The federal government has taken action on hate crimes as well. In 1990 President Bush signed the Hate Crimes Statistics Act into law. The Act instructed the FBI to collect statistics on hate crimes in the United States and asked for the voluntary cooperation of state law enforcement agencies in collecting hate crime statistics.
This discussion raises two important empirical questions: first, what factors explain the high level of state legislative activity concerning bias motivated crime and second, what are the determinants of state voluntary participation with the federal Hate Crimes Statistics Act of 1990? The research presented here uses multiple regression analysis on a fifty-state data set to examine these questions. In section one, I define and describe the characteristics and politics of hate crime. In section two, I characterize hate crime policy as a social regulatory policy to model the pattern of politics involved in the adoption of hate crime laws. The social regulatory policy framework suggests that hate crime policy is likely to result from the level of hate crimes in a state, potential interest group strength, bureaucratic power, party competition, and issue salience. Finally, in section three, the discussion and results of the first two sections are used to guide an examination of state law enforcement agency participation in the collection of statistics for the Hate Crimes Statistics Act of 1990. 1 conclude with a discussion of the findings and their implications for theories of policy formulation and implementation.
SECTION I: THE CHARACTERISTICS AND POLITICS OF HATE CRIME
Hate crimes are often defined as crimes that are committed, wholly or in part, because of the victim's race, ethnicity, religion, or sexual orientation (U.S. Department of Justice 1993: 1). As with most crime, less violent hate crimes are committed more often than violent crimes but no matter the level of violence, all hate crimes are thought to negatively impact both the victim and society Perpetrators of hate crimes are often characterized as young, white, lower-class males who commit the crimes for excitement or because of resentment of a minority group (Comstock 1991: 60-62; McDevitt and Levin 1993).
In 1992 there were 8,106 hate crime offenses reported to the FBI with crimes against persons accounting for 74 percent of the offenses (U.S. Department of Justice 1993: 7). Table 1 indicates that over 80 percent of hate crimes are directed at whites, blacks, Jews, and homosexuals, with offenses against blacks constituting the largest percentage of hate crimes. Not surprisingly, because minority groups are the main victims of hate crimes, they should have a vested interest in the passage of hate crime legislation. Minority groups may push for hate crime legislation simply as a reaction to the threat but they may also use the issue as a means to expand their political agenda (see Jenness 1995).
SECTION 2: REGULATING HATE: EXPLAINING THE LAWS
Hate Crime Laws as Social Regulatory Policy
Working within a school of thought that suggests policy type will determine factors important for policy adoption, Tatalovich and Daynes (1988: 24) argue that some public policies, including those involving crime, can be classified as social regulatory policy Accepting the premise of Theodore Lowi's policy typology, but also recognizing its limitations in describing "new" types of policy, Tatalovich and Daynes (1988: 2) argue that social regulatory policy regulates a social relationship rather than an economic transaction. Social regulatory policies seek to change behavior that is linked to a normative debate concerning the morality of individual actions and the subsequent consequences of those actions to the rest of society Social regulatory policies are, therefore, often attempts both to regulate behavior and to redistribute values in society (1988: 3). As such, social regulatory policy intersects the traditional goals of regulatory policy and redistributive policy, resulting in a hybrid pattern of politics combining elements of both policy areas. Similar to other types of social regulatory policy such as drug policy (Meier 1992), alcohol policy (Meier and Johnson 1990), capital punishment (Nice 1992), gun control (Spitzer 1988), and abortion policy (Tatalovich 1988), hate crime policies are expected to regulate a harm done to society and place a normative value on that harm, as well as to compensate victims and punish offenders. The normative moral argument made in support of hate crime laws is that hate crimes hurt not only the immediate victims, but also damage the community as a whole by giving value to bias motivated behavior (McDevitt and Levin 1993: 137; Ward 1986: 46).
Because social regulatory policy involves the distribution of values as a key component, the scope of the problem to be regulated need not be severe or even understood for government action to occur. Instead, social regulatory policies may largely arise out of the political demands of citizens and interest groups, the motivations of politicians, and bureaucratic structure and behavior (Meier 1992; Meier and Johnson 1990; Nice 1992; Wirt 1983; for examples of crime policy see Gray and Williams 1980: 155-57; Skogan 1990; Tatalovich and Daynes 1988).
The enactment of social regulatory policy is also dependent on the level of issue salience. If the salience of the issue is high, information costs will be low, allowing for high citizen participation and the mobilization of interest groups, and attract the attention of politicians in a competitive political environment (Meier 1994: 4-7; see also Nice 1992; Polsby 1984: 393). With the cost of information low, bureaucracies may find it difficult to bring their main resource (information) to bear on the process. They may, nonetheless, have preferences on legislation that seeks to expand current law and increase their responsibilities (Gray and Williams 1980: 5, 25; Meier 1994: 13-15).
The pattern of politics surrounding social regulatory policy differs from other types of policy in a number of ways and can be clarified with some simple comparisons. First, whereas competitive regulatory policies tend to involve few actors and have a low profile (Ripley and Franklin 1991: 21), social regulatory policies involve multiple political actors and tend to be highly salient. Second, while social regulatory policy does resemble protective regulatory policy in the influence of interest groups and bureaucratic actors, protective regulatory policy tends to be less salient and is, therefore, usually less attractive to politicians (Ripley and Franklin 1991: 104-105). Furthermore, interest groups are more likely to be influential in protective regulatory policy, largely because their influence is less likely to be countered by opposing interest groups. Interest groups involved in social regulatory policy are also more likely to be citizen groups or single issue groups and not the economic interest groups involved in regulatory policy.
Third, like redistributive policies, social regulatory policies tend to be highly salient and involve citizens, interest groups, and politicians (Ripley and Franklin 1991: 18-19). Unlike redistributive policies however, social regulatory policy does not necessarily involve class conflict. In the case of hate crime policy, this fact may make partisanship less likely In the policy adoption stage, social regulatory policy is also more likely to be influenced by bureaucratic actors than is redistributive policy Fourth, although not a formal component of my model, the unstable relationships among actors in social regulatory policy make the formation of temporary coalitions likely, which is similar to other types of regulatory policy but unlike redistributive policy (see Ripley and Franklin 1991: 18-19). Finally, in economic regulatory policy, the states are not afforded a policymaking role (Tatalovich and Daynes 1988: 224). In social regulatory policy, however, states not only make policy, they also influence the implementation of federal policy, which is illustrated in section three.
Variable Discussion and Measurement
Dependent Variable-Hate Crime Laws: Public policies concerning hate crimes fall under three broad categories: laws prohibiting intimidation or interference with civil rights, laws that create separate bias-motivation crimes, and laws that provide penalty enhancement provisions. Some states have further expanded the scope of their hate crime laws by mandating the training of law enforcement personnel in the identification of hate crimes and the collection of hate crime statistics (Freeman and Kaminer 1994: 8-10). By September of 1995, thirty-nine states had passed hate crime laws in at least one of the five categories. Many of the laws have been based on sample legislation proposed by the Anti-Defamation League and have withstood the scrutiny of the courts (Freeman and Kaminer 1994).
To measure state hate crime policy, a ten-point additive index was created by assigning points to each state based on the scope of hate crimes laws and the specific groups mentioned. States were assigned points on a base index where one point was added for each of the following five categories: a law concerning bias motivated violence and intimidation, a law allowing for civil action by victims, a law providing for increased criminal penalties, a law requiring law enforcement agencies to collect data on bias motivated crimes, and a law requiring training of law enforcement personnel in the identification of hate crimes.l State scores on the base index range from zero to five. There is a fair amount of correlation between each of the five categories in the base index. A factor analysis of the five categories indicates that a single factor accounts for over 50 percent of the total variation with each factor loading at .56 or higher.
To capture the coverage of the laws, points were added to the base index in accordance with the number of groups and activities listed by the laws. One point was added for the mention of each of the following group categories in the statute: race, religion, and ethnicity (as one), sexual orientation, gender, and mental or physical disability or handicap.2 Factor analysis also reveals that these categories are highly correlated; one factor explains 60 per cent of the variance between the categories with each category loading at .65 or higher. With these scores added to the base index, state scores range from 0 to 9. Because the index includes both the scope and coverage of hate crime laws, it has face validity as a comprehensive measure of state hate crime policy3 A factor analysis of the full index also indicates a fair degree of unidemsionality, with one significant factor accounting for 55 percent of the variation and all categories loading at .56 or higher. Both the sixpoint base index and the ten-point full index will be examined as separate dependent variables.
Independent Variables: The discussion of hate crime policy as a social regulatory policy indicated that certain factors should be important for explaining state policy variation. I therefore examine the influence of issue salience, interest groups, competition between political parties, the hate crime rate, and the strength of law enforcement bureaucracy.
Because issues must reach the political agenda to be acted on, social regulatory policies are argued to be enacted when the salience of the particular issue is high (Gormley 1986: 599-600; Meier 1994: 10; Meier and Johnson 1990; Nice 1992: 1038). High issue salience may, in part, be due to the severity of the problem and should serve to activate attentive publics. To measure the salience of hate crime issues I counted the number of newspaper articles published in each state on the issue of hate crimes between 1985 and July 1995 using the Newsbank Electronic Information System.4 The number of articles for each state was simply summed, being careful not to double-count articles. The measure was not standardized for state population to ensure that the symbolic value of each article was captured by the measure. Hate crime salience is hypothesized to be positively related to hate crime policy
In the formation of any public policy, interest groups often have the opportunity to influence policy choices. Groups may work to mobilize the public at the grassroots level; however, interest groups are usually most effective when they are able to limit the involvement of citizens and opposing groups (see Lowi 1969; Schattschneider 1960). As an attentive public, interest groups may help to "create" the issue, increase issue salience, or be mobilized by the severity of the problem. Because the earlier analysis of hate crime rates indicated some minorities are more likely to be targets of hate crime, we might expect that groups representing the interests of these minorities are likely to attempt to influence policy concerning hate crimes.
Indeed, the literature indicates that there is a fair amount of interest group activity concerning hate crimes at the local, state, and national level. Groups representing Jews, blacks, Asians, and gays and lesbians have kept statistics on the number of hate crimes since the early 1960s, drafted sample hate crime legislation, lobbied for hate crime laws, launched education campaigns, set up community hate crime centers, and pushed for special training of law enforcement personnel (Czajkoski 1992: 36; Freeman and Kaminer 1994: 1-4; Herek and Berrill 1992: 6; Jenness and Broad 1994: 408).
Unfortunately, none of the groups representing Jews or blacks was willing to provide state-level data on its activity, membership, or financial contributions.5 use two measures of potential interest group strength here. The first is the average number of members in the National Gay and Lesbian Task Force (NGLTF) and the Human Rights Campaign (HRC) per 100,000 state population.6 Both groups lobby for hate crime laws and form coalitions with other minority groups (Berrill 1992). Averaging membership across two groups should provide a potential interest group strength measure that is more diverse than membership in a single group. The measure should also be superior to any surrogate measure of the gay and lesbian population because it captures those gays and lesbians that are already politically mobilized and may be called into action by an interest group (see Haider-Markel and Meier 1996: 336).7
Second, because Jewish organizations, such as the Anti-Defamation League, appear to be the most active in tracking hate crimes (Czajkoski 1992: 36), I include a measure of the percent of the state population that is Jewish as a surrogate indicator of potential Jewish interest group strength.8 Average membership in NGLTF and HRC, and the percent Jewish in each state are hypothesized to be positively related to hate crimes policy
As many researchers have noted, politicians in competitive party systems are often forced to take progressive action or lose their seats in the next election (see Haider-Markel and Meier 1996: 338; Nice 1992: 1040; Skogan 1990: 393; Tatalovich and Daynes 1988: 218). Party competition, therefore, is expected to be positively related to hate crimes policy The measure of party competition used here is the district-level measure designed by Holbrook and Van Dunk (1993).9
While bureaucratic agencies often do not become involved in the policy formulation of highly salient issues, they may have preferences on policy outcomes that spur them to action (Meier 1994: 108; Tatalovich and Daynes 1988: 220-21). In addition, larger bureaucracies are more likely to intervene in the policymaking process (Meier 1994: 14-15). States with the largest law enforcement bureaucracies, therefore, are more likely to make their preferences on hate crime policy known. But what, if any, are the preferences of law enforcement bureaucracy on hate crimes?
Expanding the law will expand the power and perhaps jurisdiction of justice system agencies, which law enforcement bureaucrats may prefer. If the new laws do not also provide new funding, however, agencies may oppose an expansion of the law. Given the relatively conservative nature of law enforcement, and the fact that the Justice Department was, for a time, the main opposition to the Federal Hate Crimes Statistics Act (ost 1993b: 13), the preferences of state law enforcement bureaucracy are likely to be against the passage of state hate crime laws (see also Berrill 1992). Large bureaucracies, furthermore, should be the most successful in reducing the scope and coverage of hate crime laws. Bureaucratic strength is measured here as the total state justice system expenditures per capita. 10
The scope and coverage of hate crime policy may also be influenced by the occurrence of hate crime. Political scientists have suggested that public policy is often a reaction to some problem (Polsby 1984: 168-69) and at least one study has demonstrated that crime policy is, at least in part, a reaction to crime rates (see Nice 1992; 1042). As hate crime increases, citizens and interest groups may demand that politicians take action. Citizen demand, however, is not a necessary condition for action. Political entrepreneurs, looking for new issues in a competitive party system, may see a policy response to rising hate crime as a way to establish themselves. As these arguments suggest, the hate crime rate is likely to be positively related to hate crime policy, but its impact may be indirect.
Although states are not forced to comply with the federal collection of hate crime statistics, the FBI has made a concerted effort to train local law enforcement personnel in the identification and tracking of hate crimes, even in those states that did not have state hate crime laws (Jost 1993a; 6). To compensate for the "softness" of hate crime data, I measure hate crimes as the average number of police reported hate crime incidents in 1992 and 1993 per 100,000 state population.l2
Results and Analysis of State Hate Crime Policy
The results of Ordinary Least Squares regression analyses of the base index and full index of hate crime policy in the states are displayed in Table 2.13 In each model, most of the variables were found to be significantly related to hate crime policy and each model explains over 60 percent of the variation in hate crimes laws. The model for the full index, the scope and coverage of hate crime laws, is predicted slightly better by the independent variables. The standardized estimates suggest that interest groups have the most influence over the scope of hate crime policy, while the level of party competition has the strongest influence over the scope and coverage of hate crime laws. While this result may be because of the limited nature of the interest group strength measures, it does make intuitive sense that high levels of party competition would lead to a greater number of groups and activities being covered under hate crime laws.
Each model confirms most of the hypotheses and suggests that hate crime policy arises largely out of party competition, the salience of hate crimes issues, and the strength of interest groups. The strength of law enforcement agencies appears to have a negative influence on the scope of hate crime laws, but the influence is not significant when the coverage of the laws is taken into account. A unit effect was also discovered-the state of Oklahoma scores higher on the hate crime policy index than would be expected based on the levels of the independent variables in the state.l4 One option in a case such as this is to eliminate the influence of the (state) observation (see Hamilton 1992: 12933). To control for Oklahoma's inordinate influence on the regression line, a dummy variable was included in the model for the state.
To confirm that Oklahoma's adoption of a hate crime law fits the social regulatory policy framework, I interviewed local activists involved with the legislation. The principal author of the hate crimes bid was Senator Maxine Homer (D-Tulsa), one of a handful of African-American members in the legislature. Senator Homer was able to create a coalition of African-Americans, influential business persons from the relatively small Jewish community, and groups such as the National Conference of Christians and Jews. Much of the motivation for the legislation was driven by a couple of well-publicized vandalisms against Jewish congregations. The legislation originally included sexual orientation, but as a necessity to getting it passed, the category was deleted. While the pattern of politics in Oklahoma largely fits the social regulatory policy framework, the case of Oklahoma suggests that political entrepreneurs play an important role in the policy process, a fact not accounted for in the model. Senator Homer was able to bring together groups that, if acting alone, may not have been able to gather support for the hate crime law ls
The models suggest that the hate crime rate is negatively related to the scope and coverage of hate crime policy This appears to be mostly the result of its moderate collinearity with other variables in the model.l6 The fact that the hate crime rate is not positively related to hate crime policy suggests that hate crime policies may not be enacted as a direct response to the level of hate crime. As mentioned, however, hate crime incidents may indirectly influence policy by activating citizens, interest groups, and politicians. Even though the influence of hate crime is uncertain, it seems likely that hate crime policy is mostly a reaction to the salience of the issue and the demands of interest groups in a competitive political system. While my measure of issue salience is likely to capture some aspect of the occurrence of hate crime, the fact that the two variables are not related (see note 15) makes it appear likely that salience may build on a few select incidents of hate crime rather than on the overall hate crime rate.
These findings lend support to the arguments of Tatalovich and Daynes (1988). Recall that they argue when interest groups that do not seek immediate instrumental benefits from policy changes, but instead seek policy that regulates values, policy will most likely result from interest group strength and factors in the political environment. Hate crime policy, therefore, appears to fit the pattern of politics implicit within the social regulatory policy framework.
Tatalovich and Daynes (1988) further argue that interest groups are not always able to ensure that their policy victories are successfully implemented and enforced by administrators. Anecdotal evidence suggests that the implementation and enforcement of state and local hate crime policy is weak. In fact, few arrests and even fewer prosecutions are made under existing hate crime laws. For example, fewer than 20 hate crime convictions have been made in Wisconsin since 1988 under the states broad hate crime law.l7A study of Boston between 1983 and 1987 found that of 452 hate crimes reported to police, only 60 were cleared by arrest, only 38 had charges filed, only 30 cases resulted in conviction, and only 5 individuals served prison time (McDevitt and Levin 1993: 194-95). In California, the data collection portion of the state's hate crime law was not funded in 1994 (Freeman and Kaminer 1994: 30). At the same time, cities such as Chicago, Portland, and San Francisco appear to be especially vigilant in implementing and enforcing hate policy (Morales 1995: 34-35). While states likely vary in their implementation and enforcement efforts, the empirical and anecdotal evidence does suggest that hate crime policy is a form of social regulatory policy that arises out of a specific array of political forces.
SECTION 3: REGULATING HATE WITH FEDERAL POLICY: STATE VOLUNTARY PARTICIPATION IN IMPLEMENTING FEDERAL HATE CRIME POLICY
The 1990 Hate Crime Statistics Act (HCSA) requires the U.S. Department of Justice to collect statistics on crimes motivated by race, ethnicity, religion, and sexual orientation.ls During the multiple attempts to pass a version of the HCSA, the most contentious aspect of the legislation was including a clause for sexual orientation. A version of the HCSA that did not include sexual orientation was passed in the House by voice vote during the 99th Congress, but was never voted on in the Senate. Introduced in the House during the 100th Congress, the second version of HCSA included sexual orientation and was supported by a coalition of over 60 groups, but again never received a floor vote in the Senate. It was not until 1989, during the 101st Congress, that the full version of the HCSA was passed by both the House and the Senate.
While there was no organized interest group opposition to the bill, it was originally opposed by the Justice Department and representatives in both the House and the Senate made unsuccessful attempts to remove sexual orientation from the bill. Senator Helms (R, NC) attempted to amend the bill by including the following provisions: (1) sodomy laws must be enforced; (2) homosexuality threatens the fabric of the American family; (3) Congress shall pass no law banning discrimination based on sexual orientation; and (4) the public schools cannot teach that homosexuality is normal and healthy. The Helms amendment failed after an intense lobbying effort, but the debate over who would be covered by the HCSA and the fact that states were not required to participate in data collection illustrates that federal hate crimes policy can be considered social regulatory policy that both redistributes values and is an attempt to change behavior. Because hate crime policy at both the state and federal level is a social regulatory policy, we might also expect that state implementation of federal hate crime policy will fit the social regulatory policy pattern of politics.
Policy Implementation and Enforcement Theory
Tatalovich and Daynes (1988; 224-25) argue that the implementation of social regulatory policy often depends on the voluntary compliance of state and local officials. Federal hate crime policy clearly fits this description. Given the voluntary nature of implementing federal social regulatory policy, therefore, local political conditions should largely determine implementation effort.
More specifically, Tatalovich and Daynes (1988: 221-22) argue that the bureaucratic implementation of social regulatory policy, like other types of policy, is largely dependent on the orientations of the bureaucracy, pressures exerted by organized interests, and the support of politicians in a competitive party system. State efforts to implement federal hate crime policy should, therefore, be fairly well predicted by the same factors that determine state hate crime policy
Dependent Variable: The FBI is directed by law to collect hate crimes statistics under the national Hate Crimes Statistics Act, but the participation of state and local law enforcement agencies is voluntary. I measure state effort to implement federal hate crime policy, therefore, as the percent of the state's population covered by police agencies submitting hate crime incident reports to the FBI under the Uniform Crime Reporting guidelines.l9 In 1992 nearly 6,200 law enforcement agencies from 39 states submitted hate crime reports to the FBI and these agencies covered about 51 percent of the U.S. population (U.S. Department of Justice 1993: 23). Hate crime incidents do not have to occur for agencies to submit reports, in fact, most small agencies did not report any incidents in their hate crime reports. State police agencies are not included in this measure because their jurisdictions cover an entire state and they only report a few hate crimes, if any (U.S. Department of Justice 1993, 1994). Nonetheless, omitting or including state agencies does not change the results of the model. Because of the variability across the states in agency participation, my measure of policy implementation should be the most robust available measure of state effort to implement federal hate crime policy
Independent Variables: Because I have suggested that state and federal hate crime policy is a type of social regulatory policy, I expect that state efforts to implement federal hate crime policy will be predicted by the same factors used to predict state hate crime policy in section two. Each independent variable from section two, except the measure of bureaucratic strength, is used in the models to predict state implementation effort.20 Because hate crimes are likely to occur in urban areas and this may ensure greater voluntary collection of hate crime statistics, I also included the percent of the state population living in urban areas. This measure, however, was not significant and a joint F-test suggested that the variable added no explanatory power to the models. Urbanism, therefore, is not included in the models presented.
Existing state hate crime policy may also influence state implementation of federal policy If a state does not have a hate crime law, but voluntarily collects hate crime statistics, it should indicate that the political forces in the state are willing to support federal hate crime policies, but there is not enough support to pass a state hate crime law. Most likely, however, states with broader hate crime laws should simply be more likely to have higher levels of implementation effort. To control for the possible bias of existing state hate crime policy I include a measure of the scope of each state's hate crime law in the implementation model (see section two for measurement).21
Results of Implementation Analysis
Preliminary analysis of the independent variables indicated nonlinear relationships with the dependent variable, so each variable used in this analysis, except the policy measure (because it is a simple count), was subjected to a log transformation.22 The results of multiple regression analysis using implementation effort in 1992 and average implementation effort between 1992 and 1993 are shown in Table 3.23 Because the variables are in natural log form, the coefficients can be interpreted as a percent increase/decrease in the population covered by participating law enforcement agencies for each percent increase/decrease in the independent variable.
The results of the model explaining state implementation effort show a pattern similar to that found in the hate crime policy models. Potential interest group strength, party competition, and high issue salience are all positively related to the implementation of federal hate crime policy The potential strength of Jewish interest groups is not significant in either model, which may result from the weakness of the measure or may actually indicate the Jewish groups focus more on passing legislation than on bureaucratic implementation. The potential interest group strength of lesbian and gay groups has the greatest influence on state implementation effort, while party competition has less impact in the model than it did in the state policy model. This result makes sense-interest groups often have the most incentive to ensure strong policy implementation because the have they most to gain, while politicians, even in competitive party systems, are less motivated to ensure implementation because they are unlikely to receive much credit for their actions.
State hate crime policy has a positive influence on efforts to implement federal policy, suggesting existing policy at the state level creates inertia that will increase efforts to implement polices which depend on federalism to be successful. Two unit effects were identified and dummy variables were added to control for the inordinate effect of these cases. Arkansas has a greater implementation effort and California has less implementation than would be suggested by the levels of the independent variables in those states.24 The hate crime rate does not appear to have a positive or significant influence on state implementation effort. This may suggest that, like state hate crime policy, state efforts to implement federal hate crime policy are likely not a reaction to the level of hate crime. Instead, hate crime may influence implementation indirectly by increasing issue salience, activating interest groups, or simply by increasing public knowledge of the problem.
DISCUSSION AND CONCLUSIONS
Given the lack of empirical investigations of hate crime, this study argues that hate crime policy is a social regulatory policy with an identifiable pattern of politics. Social regulatory policy is thought to arise out of factors in the political environment including the strength of interest groups, party competition, issue salience, and bureaucratic strength. Based on this perspective, this study proposed models useful for explaining the scope and coverage of hate crimes policy and state effort to implement federal hate crime policy.
The analysis of state and federal hate crime policy suggests that, indeed, hate crime policy is largely determined by political pressures, including party competition, interest group strength, issue salience, and the strength of law enforcement bureaucracy The empirical analysis of policy implementation confirmed the social regulatory policy pattern of politics-state efforts to implement federal policy were predicted best by interest group strength, party competition, issue salience, and the scope of state hate crime policy. Neither the policy nor implementation analysis, however, was able to determine that the hate crime rate had a positive influence on state hate crime policy or the implementation of federal hate crime policy While hate crime may have an indirect influence on policy, the precise impact of such crimes may well be best determined at the state or local level when time-series data on hate crimes becomes available.
This study, therefore, found that the social regulatory policy theory offered by Tatalovich and Daynes (1988) is a useful framework for understanding hate crime policy and implementation. The pattern of politics implicit within the social regulatory policy framework continues to illustrate how and when public officials respond to the demands of their constituencies and organized interests. Researchers can fruitfully apply this framework to understand the politics involved in the formulation and implementation of other similar, types of public policy.
Received: June 12, 1996
Accepted: December 1,1996
NOTE: An earlier version of this article was presented at the 1996 annual meeting of the Midwest Political Science Association. I would like to thank David Pritchard, Kenneth J. Meier, and the editors of PRQ for their comments on earlier drafts of this article.
Each category specified here was assigned one point since I have no theoretical reason for believing that one category is more important than another, only that the scope of the law is of greatest importance. Data are from Freeman and Kaminer (1994) and a September 1995 update obtained from the Anti-Defamation League by the author.
2 Less than five states include political affiliation and age in their hate crime laws and some states have separate laws for institutional vandalism and interference with religious worship. Laws concerning these activities and groups are not included in the analysis because factor analysis demonstrates that they are significantly different from the major groups protected in hate crime laws. Including these groups, however, does not significantly alter the results of the models. 3 Similar indices have been used in explaining crime policy (see Nice 1992: 1041). 4 This system is an electronic index of major U.S. newspapers. It contains over 900,000 articles which can be referenced by subject and listed by state (the keywords used here were "hate crime" and "bias crime). At least one newspaper from each state is indexed in Newsbank.
5 I contacted the Anti-Defamation League of B'nai B'rith, the National Association for the Advancement of Colored People, and the American Jewish Committee. None of these organizations was willing to provide membership or financial contribution data by state. 6 These two groups are the largest gay and lesbian groups in the U.S. Both groups have been highly active on hate crimes and form coalitions with other minority groups to combat anti-minority violence (Herek and Berrill 1992: 6; Jenness and Broad 1994; 407). Membership data were obtained by the author for each group.
A possible surrogate measure of the gay and lesbian population, such as the same-sex partner measure used in the 1990 census, is correlated with state membership levels in gay and lesbian interest groups. Preliminary bivariate regressions with the hate crime scale, however, suggested that the membership measure is the best predictor of policy "Data are collected by Jewish churches and reported in the U.S. Bureau of the Census (1993). 1 also included measures of the black and Asian populations, but nether significantly increased the explanatory power of the models. 9 Holbrook and Van Dunk (1993) do not provide an estimate for Louisiana. Comparisons with other measures of party competition allow Louisiana's score to be estimated as 17.01.
10 Data are from U.S. Bureau of the Census (1993).
II While hate crime laws may have some deterrent effect on the hate crime rate, deterrent effects are best investigated in a time-series rather than cross-sectional model (Durant and Legge 1993). To investigate the possibility of a deterrent effect for those states that had hate crime policies before 1992, created a dummy variable for these states. A bivariate regression predicting hate crime rates with the dummy variable found that there was no significant difference between the states that had passed hate crime laws before 1992 and those that had passed laws after 1992. 12 Data are from U.S. Department of Justice (1993, 1994). 13 Because the dependent variable only ranges from 0 to 5 in the base index model, I also estimated the model using ordered probit. The results match the OLS results very closely with all variables showing significant relationships. Because the measure was normally distributed and for ease of interpretation, Table 2 presents only the OLS results.
4 Regression diagnostics including studentized residuals, Hat diagonal, and Cook's D indicated that Oklahoma had an inordinate amount of influence on the regression line. 15 Information on Oklahoma's hate crime law was collected in an interview with Tom Neal, the editor of Tulsa News, a local gay newspaper.
The hate crime rate was collinear with membership levels in the gay and lesbian interest groups (regression adjusted R-square of .18), but surprisingly, not with the measure of issue salience. Attempts to purge the variable of collinearity were not successful. A bivariate regression indicates that hate crimes are not significantly related to hate crime policy
ls The discussion of the passage of the HCSA borrows from Berrill (1992: 17-19).
'9 Data are from U.S. Department of Justice (1993, 1994). 20 A joint F-test indicated that bureaucratic strength had no significant impact on the models, therefore, I omitted the variable to improve the parsimony of the models. Bureaucratic strength is probably not significant in the models because those states with bureaucratic agents most opposed to the collection of hate crime statistics did not submit hate crime reports and are therefore counted as missing in the data-set.
21 Because section two demonstrated that state policy is well predicted by the political factors specified, the measure of state hate crime policy used in the implementation model was purged of its collinearity with the political variables by running a regression and using the residuals for the policy measure in the implementation model. 22 When logged variables are used regression coefficients must be interpreted as elastici
ties (see Tufte 1979).
23 I averaged the dependent variable. across 1992 and 1993 to reduce the likelihood of measurement error. The results for 1992 and 1993 are quite similar.
Regression diagnostics revealed that both states had an inordinate impact on the regression line. To eliminate the skewed impact of these states, dummy variables were included in the model for each.
Anti-Defamation League of B'nai B'rith, rev. ed. 1988. Hate Groups in America: A Record of Bigotry and Violence. New York: Civil Rights Division of the ADL.
Berrill, Kevin T.1992. Countering Anti-Gay Violence Through Legislation. Washington, DC: National Gay and Lesbian Task Force Policy Institute. Comstock, Gary D. 1991. Violence Against Lesbians and Gay Men. New York: Columbia University Press.
Czajkoski, Eugene H. 1992. "Criminalizing Hate: An Empirical Assessment." Federal Probation 56: 36-40.
Durant, Robert E, and Jerome S. Legge, Jr. 1993. "Policy Design, Social Regulation, and Theory Building: Lessons from the Traffic Safety Policy Arena." Political Research Quarterly 46: 641-56.
Freeman, Steven M., and Debbie N. Kaminer. 1994. Hate Crimes Laws: A Comprehensive Guide. New York: Anti-Defamation League. Gormley, William T. 1986. "Regulatory Issues in a Federal System." Polity 18: 595-620.
Gray, Virginia, and Bruce Williams. 1980. The Organizational Politics of Crimi
nal Justice. Lexington, VA: Lexington Books. Haider-Markel, Donald P, and Kenneth J. Meier. 1996. "The Politics of Gay and Lesbian Rights: Expanding the Scope of the Conflict."Journal of Politics 58: 352-69.
Hamilton, Lawrence C. 1992. Regression With Graphics: A Second Course in Applied Statistics. Pacific Grove, CA: Brooks/Cole. Hamm, Mark S.1993. American Skinheads: The Criminology and Control of Hate Crime. Westport, CT: Praeger.
Herek, Gregory M., and Kevin T. Berrill, eds. 1992. Hate Crimes: Confronting Violence Against Lesbians and Gay Men. Newbury Park, CA: Sage. Holbrook, Thomas M., and Emily Van Dunk. 1993. "Electoral Competition in the American States." American Political Science Review 87: 955-62. Jenness, Valerie. 1995. "Social Movement Growth, Domain Expansion, and Framing Processes: The Gay/Lesbian Movement and Violence Against Gays and Lesbians as a Social Problem." Social Problems 42 (1): 145-70. Jenness, Valerie, and Kendal Broad. 1994. "Antiviolence Activism and the (In)visibility of Gender in the Gay/Lesbian and Women's Movements." Gender & Society 8: 402-23.
Jost, Kenneth. 1993a. "The Issues." CQ Researcher 3 (1): 3-7.
. 1993b. "Background: Violence and Prejudice." CQ Researcher 3 (1): 7-15.
Lowi, Theodore. 1969. The End of Liberalism. New York: Norton. Lutz, Chris. 1987. They Don't All Wear Sheets: A Chronology of Racist and Far Right Violence-1980-1986. Atlanta: Center for Democratic Renewal. McDevitt, Jack, and Jack Levin. 1993. Hate Crimes: The Rising Tide of Bigotry and Bloodshed. New York: Plenum Press.
Meier, Kenneth J. 1992. "The Politics of Drug Abuse: Laws, Implementation, and Consequences." Western Political Quarterly 45: 41-69. . 1994. The Politics of Sin: Drugs, Alcohol, and Public Policy. Armonk, NY: M. E. Sharpe.
Meier, Kenneth J., and Cathy Jolinson. 1990. "The Politics of Demon Rum: Regulating Alcohol and its Deleterious Consequences." American Politics Quarterly 18: 404-29.
Morales, Jorge. 1995. "Cities of Light." The Advocate, July 11: 34-35. National Gay and Lesbian Task Force Policy Institute. 1993. Anti-Gay Violence, Victimization & Defamation in 1992. Washington, DC: National Gay and Lesbian Task Force Policy Institute.
. 1994. Anti-Gay Violence, Victimization Sr Defamation in 1993. Washington, DC: National Gay and Lesbian Task Force Policy Institute. Nice, David C. 1992. "The States and the Death Penalty" Western Political Quarterly 45: 1037-48.
Polsby, Nelson. 1984. Political Innovation in America. New Haven: Yale University Press.
Ripley, Randall B., and Grace A. Franklin. 1991. Congress, the Bureaucracy, and Public Policy, 5th ed. Pacific Grove, CA: Brooks/Cole. Schattschneider, E. E. 1960. The Semi-Sovereign People. New York: Holt, Reinhardt and Winston.
Skogan, Wesley 1990. "Crime and Punishment." In Virginia Gray, Herbert Jacob, and Robert Albritton, eds., Politics in the American States, 5th ed. Glenview, IL: Scott, Foresman/Little, Brown. Spizer, Robert J. 1988. "Gun Control: Constitutional Mandate or Myth?" In Raymond Tatalovich and Byron W Daynes, eds., Social Regulatory Policy. Boulder, CO: Westview Press.
Tatalovich, Raymond.1988. "Abortion: Prochoice Versus Prolife." In Raymond Tatalovich and Byron W Daynes, eds., Social Regulatory Policy. Boulder, CO: Westview Press.
Tatalovich, Raymond, and Byron W Daynes, eds.1988. Social Regulatory Policy: Moral Controversies in American Politics. Boulder, CO: Westview Press. Tufte, Edward R. 1979. Data Analysis for Politics and Policy. Englewood Cliffs, NJ: Prentice-Hall.
U.S. Bureau of the Census. 1993. Statistical Abstract of the United States: 1993, 113th ed. Washington, DC: U.S. Bureau of the Census. U.S. Department of Justice. 1993. Hate Crimes Statistics, 1992. Washington, DC: Federal Bureau of Investigation, Criminal Justice Information Services Division.
. 1994. Hate Crimes Statistics, 1993. Washington, DC: Federal Bureau of Investigation, Criminal Justice Information Services Division. Ward, Benjamin.1986. "Hate Crimes: The Police Response in New York City" The Police Chief 53 (December): 46-47.
Wirt, Fredrick M. 1983. "Institutionalization: Prison and School Policies." In Virginia Gray, Herbert Jacob, and Kenneth N. Vines, eds., Politics in the American States, 4th ed. Glenview, IL: Scott, Foresman/Little, Brown.…