Week 8 Article 2

Predictors of Hate Crime Prosecutions: An Analysis of Data From the National Prosecutors Survey and State-Level Bias Crime Laws

Bryan D. Byers 1 ,

Kiesha Warren-Gordon 1

and James A. Jones

1

Abstract Research on hate crime has focused primarily on law making, law enforcement, and victimization aspects. Few researchers have studied hate crime prosecutions even though this is an important element in such cases. This study uses data from the National Prosecutors Survey of 2001 to predict the likelihood of hate crime prose- cutions. Given the data set is a census of prosecutors, it was necessary to add 10 new variables to the data based on the presence and absence of state hate crime laws and their characteristics. The data were subjected to logistic regression, and it was determined that the three strongest predictors of whether prosecutors pursue hate crime are the presence of (a) race, ethnicity, and religion as protected groups in state hate crime law, (b) sexual orientation as a protected status within state law, (c) the presence of an institutional vandalism provision in state-level hate crime law, and (d) if the prosecutor’s office assigned staff to handle community-related activities. The find- ings are discussed along with suggestions for future research.

Keywords hate/bias crimes, victimization, hate crime prosecution, National Prosecutors Survey, hate crime law, antidefamation league

1 Ball State University, Muncie, IN, USA

Corresponding Author:

Bryan D. Byers, Ball State University, NQ 278, 2000 W. University, Muncie, IN 47306, USA

Email: bbyers@bsu.edu

Race and Justice 2(3) 203-219

ª The Author(s) 2012 Reprints and permission:

sagepub.com/journalsPermissions.nav DOI: 10.1177/2153368712446868

http://raj.sagepub.com

Introduction

The topic of hate or bias crime 1

has been addressed considerably by scholars. Some of

this treatment has involved the study of the definitions of hate crime, hate crime

legislation, policing hate crime, legal and First Amendment issues, Supreme Court

cases, reporting practices by citizens in national data, a few offender studies, and some

victimization studies. Although there have been a number of studies and works

devoted to the topic of hate or bias crime in these arenas, very little attention has been

given to the topic of hate, or bias, crime prosecution. To examine hate crime prose-

cutions, the present study utilized an existing national data set from the National

Archive of Criminal Justice Data’s (NACJD) National Prosecutors Survey of 2001,

which was supplemented with additional independent variables consisting of state-

level hate crime law characteristics obtained from the Antidefamation league (ADL).

The dependent variable for the study was whether the hate crime was prosecuted in

the jurisdiction as reported by prosecutor, or designee, respondents. Therefore, the

purpose of this study is to determine the impact of state-level bias or crime law char-

acteristics on the likelihood of such prosecutions.

Literature Review

The handful of scholars having examined hate crime prosecutions include Finn

(1988), Bell (2002), Franklin (2002), Haider-Markel (2002), Jenness and Grattet

(2004), McPhail and Jenness (2005), McPhail and DiNitto (2005), Read (2005), Levin

et al. (2007), Phillips (2009), and Hodge (2011). In addition to these scholarly treat-

ments of hate crime prosecution topics, there are a few reports and advisory papers

distributed by the National Institute of Justice, the National District Attorneys Asso-

ciation research branch American Prosecutor’s Research Institute (APRI), and private

research firms such as Abt Associates designed to inform on various hate crime topics

and, in some, instances, advise criminal justice practitioners, including prosecutors,

who might handle such cases (Shively, 2005).

It is considered to be difficult to obtain a conviction in hate crime cases for a variety

of reasons, contributing to why so few hate crime cases lead to conviction. Many bias

crimes are ‘‘low-level’’ offenses (Garofalo & Martin, 1993) and the perpetrator is

typically unknown to the victim (Byers & Zeller, 1997, 2001). As a result, and in such

cases, the incident is either minor in the eyes of prosecutors and/or there is no indi-

vidual to identify for prosecution. When the case involves the former, prosecutors

must weigh the seriousness of the offense against the likelihood of conviction, the

relative cost of prosecution, and the efficacy and efficiency of bringing charges

(Jenness & Grattet, 2004). For the most part, prosecutors have little experience with

bias-motivated criminal conduct (Jacobs & Potter, 1998), which challenges some of

the best district attorneys. Even when cases appear to have clear-cut bias motivation,

it has been found that obtaining a conviction based on the standard of ‘‘beyond a rea-

sonable doubt’’ can still be a daunting task (Jacobs & Potter, 1998) along with trying

to prove that the ‘‘bias’’ the alleged offender possessed ‘‘caused’’ the criminal act or

204 Race and Justice 2(3)

led to the ‘‘selection’’ of the victim as a target, depending on how motivation is

addressed in various state law. Thus, prosecuting bias crimes is risky. In spite of these

concerns, Levin et al. (2007), using a newspaper surveillance study of hate crime pro-

secutions in the United States between 2002 and 2005, revealed 110 total cases with a

total of 177 offenders. Of the 177 offenders, only 4 were acquitted. These results

should not suggest that hate crime prosecutions and convictions are easily obtained.

On the contrary, what these results more likely suggest is that the prosecutor had a

strong case.

Even when cases are more serious and clear-cut in nature, there are issues for pro-

secutors generated from Supreme Court cases which challenged the constitutionality

of bias crime laws (Wisconsin v. Mitchell, 1993) and ordnances (R.A.V. v. The City of

St. Paul, MN, 1992). While in R.A.V., the Court ruled in favor of the plaintiff. This was

not the case in Mitchell. R.A.V.-challenged free speech provisions as assured by the

First Amendment due to the St. Paul ordinance being worded in such a way that these

protections could be infringed upon. As noted by Read (2005), the favorable ruling in

Mitchell did not reduce the burdens and difficulties faced by prosecutors in seeking

hate crime charges and penalty enhancements. In addition to these cases, there are

a number of cases from lower courts as well as those that have challenged hate crime

prosecutions for a variety of substantive reasons (Jenness & Grattet, 2004). Thus, pro-

secuting hate or bias crimes can potentially be fraught with numerous legal obstacles

generated from case law. Such obstacles may deter prosecutors in seeking hate crime

convictions even in the strongest of cases. Given these, and other legal and practical

considerations, researchers have attempted to shed light on some of the factors which

might influence prosecutorial decisions in hate crime cases.

Bell, in Policing Hatred: Law Enforcement, Civil Rights, and Hate Crime (2002),

provides an interesting treatment of the process and decision points involved in hate

crime investigation and prosecution. While not national in scope, her examination

suggests that although the skill of investigators in hate crime cases was important to

seeing the matter to fruition in the courts, there was no guarantee that the case would

be prosecuted or a conviction obtained. Selecting solid cases based on evidence along

with the related matter of whether there was a good chance that the case could be won

were important factors in how prosecutors and the courts dealt with bias crimes.

Prosecuting hate crimes is fraught with difficulty. According to Bell (2002), one of the

greatest difficulties is proving motivation that the alleged offender committed the act

‘‘by reason of’’ or ‘‘because of’’ and primarily due to the victim’s identity (p. 4). As

stated by Jacobs and Potter (1998), ‘‘It is . . .difficult to provide that bias motivation

caused the criminal conduct’’ (p. 103). Haider-Markel (2002), in a survey study of

police administrators and prosecutors/District Attorneys (DAs) in some of the most

populous US cities, examined a number of variables to measure hate crime law

enforcement. The D.A. sample only consisted of 37 offices, but it yielded some inter-

esting findings. The two part survey was divided into (1) questions about administra-

tive procedures and practices and (2) questions in response to a hypothetical hate

crime scenario. When these two types of measures were taken together for the D.A.

sample, it was revealed that [hypothetical] hate crime prosecution would more likely

Byers et al. 205

be pursued if there was a state hate crime policy (law), there was a higher ‘‘same-sex’’

household percentage in the jurisdiction, if fundamental Protestantism was lower, and

if the D.A. was supportive of hate crime prosecution (p. 146). In related, yet unpub-

lished, works, it has been found it is more likely that charges will be brought if the

office of the prosecutor is larger and if there is an ADL office in the state (King,

2001; McPhail & Jenness, 2005) and the offender’s use of words, symbols, acts

deemed offensive to particular groups, and offender and witness statements (APRI,

1998; McPhail & Jenness, 2005). Therefore, some evidence has been found to support

the notion that a variety of factors impact the decision to prosecute—some evidentiary

based and some extralegal.

While Haider-Markel dealt with hypothetical situations, Phillips (2009) studied

hate crime prosecutions in a New Jersey county over a 4-year period. Using archival

data, it was found that of the 643 cases investigated as bias crimes during the study

time frame only 30 (4.66%) were referred for prosecution—which is not out of the ordinary for other jurisdictions or for nonbias-motivated crimes (Jacobs & Potter,

1998; Jenness & Grattet, 2004). Of these, bias crimes based on religion and those

involving multiple bias motivations were most successfully prosecuted (Phillips,

2009, p. 12). It was concluded that those cases where bias seemed to be the chief moti-

vator (as opposed to peripheral) were more likely to end in prosecution. The plurality

of the cases were what could be called ‘‘thrill’’ based hate crimes and relatively low

level of offenses. What these findings seem to suggest is the possibility that even in

low-level offenses, if the bias motivation is clear and other factors are in place, suc-

cessful prosecution is presumably more likely.

Prosecutors are not immune from the political pressures inherent in their positions.

While a local prosecutor or district attorney is obligated to bring charges in cases on

behalf of the ‘‘state’’ or ‘‘the people,’’ there is always an amount of discretion

involved in such decisions. As Jacobs and Potter (1998) note, sometimes it is in a

prosecutor’s political best interest to authorize charges in a hate crime case. As they

state (p. 103):

A prosecutor could use such a trial [bearing in mind there is a great chance the case may

not make it to trial—our emphasis] to cement her support with an advocacy organization

or a particular group. Under certain circumstances, for some prosecutors, demonstrating

solidarity with the victim’s group might be more important politically than obtaining a

conviction.

This position is based on the notion that prosecutors make decisions based on self-

interest and efficiency (Jenness & Grattet, 2004). While this is just one of the several

orientations intended to explain prosecutorial discretion, it is one which is aligned

with the reality that prosecutors and D.A.s at the local-county level are elected offi-

cials. Elected officials have many constituent groups which can bring pressure to bear

that can, in turn, potentially influence decision making. It is well documented that the

creation of bias crime legislation in various states has been due to political influences

explained by ‘‘identity politics’’ and special interest group activity (Becker, Byers, &

206 Race and Justice 2(3)

Jipson, 2008; Jacobs & Potter, 1998), so it stands to reason that those who enforce the

laws may also be influenced in similar ways.

However, there are those who suggest that some commonly accepted extralegal

factor influences may not impact decision making at all. McPhail and Jenness (2005)

conducted a qualitative study of prosecutors in the state of Texas using a nonprob-

ability sample of 17 prosecutors, county attorneys, and their assistants. The authors

used a semistructured interview approach and, as they state, ‘‘. . . asked interviewees

how they understand the parameters of hate crime law, how they evaluate the law, how

they think about the feasibility of implementing the law, how they determine whether

to file a hate crime charge and/or pursue a penalty enhancement, and how much

experience they have had with hate crime prosecutions’’ (p. 95). As they conclude:

When considering whether to attach a charge of ‘‘hate crime’’ to a case, prosecutors

engage in a decision-making process that can be stated in a single sentence: Prosecutors

are seeking justice and enforcing the law by employing a standard decision-making pro-

cess while pursuing a strategic advantage when deciding whether to charge a hate crime

enhancement and, at the same time, denying or minimizing the influence of extralegal

factors. (p. 96)

In short, McPhail and Jenness (2005) conclude that prosecutors essentially consider

factors, which are often considered in other types of cases and try to suppress pres-

sures which are not evidence based or germane to the case at hand.

There are several overarching issues one finds in the literature on hate crime

prosecution. First, the nature of the research is quite varied or diverse. This leads to a

lack of clear direction with regard to the overall agenda. The second issue, which is

related to the first, is that the literature on hate crime prosecution is relatively new and,

therefore, limited. If one examines the years associated with the publication of lit-

erature on this topic, it is evident that most of the literature is dated after 2000. Given

the diversity of literature and the newness of this area for purposes of research, there

are opportunities for additional exploration. Even though this area appears to be open

to additional exploration, the previous research on hate crime prosecution does

indicate some common themes.

The first theme suggests that it is difficult to obtain convictions in hate crime cases.

A prosecutor must prove beyond a reasonable doubt that not only was the crime

motivated or the victim selected 2

due to bias. Thus, not only must a prosecutor carry

the burden of proof for the crime but she or he must also show motivation or selection

bias. A second theme is the controversial nature of hate crimes. Few types of crime

have been heard by the United States Supreme Court based on offense motivation. The

contentious nature of hate crime cases exists because of the conflict between criminal

motivation and the First Amendment. 3

This has been a central feature of hate crime

cases handled on appeal. A third, and final, theme suggests that hate crime prosecution

will most likely be pursued if a case has been investigated well, there is strong evi-

dence, the bias motivation is clear, and there is a strategic and political advantage in

charging the case. While these are important themes, no research has examined the

Byers et al. 207

characteristics of state-level hate crime laws to determine the impact these traits may

have on the practice of hate crime prosecution.

Methodology

The methodology used in this study was a combination of existing national data

coupled with an examination of state-level hate crime laws within the United States.

Therefore, both elements of the final data set were national in scope. The primary data

set, National Survey of Prosecutors of 2001 was obtained from the NACJD. In contrast

to previous years, the 2001 data collection, within the National Prosecutors Survey

Series, was a census of the 2,341 chief prosecuting attorneys in the United States han-

dling felony cases in general jurisdiction state-level courts rather than a survey sample

of prosecutors in the United States. Of the 2,341 respondents, multiple respondents

could be from the same state yet different jurisdictions. Previous and subsequent

surveys, but not censuses, occurred in 1990, 1992, 1994, 1996, and 2005. Respondents

were only asked about hate crime prosecutions in the 1994 and 2001 data collections.

Therefore, the 2001 census is the most recent information available on hate crime pro-

secutions from this series. The census was carried out by the National Opinion

Research Center.

The original data set consisted of 229 variables dealing with office characteristics

and prosecutorial practices. In order to study the impact of state-level hate crime law

characteristics, nine additional variables were added to the data set.

Using the state code for each responding prosecutor’s office, information regarding

state bias crime statutes was obtained from the ADL. The ADL maintains a census of

state bias or hate crime laws and releases information regularly/annually about state-

level legislative additions and changes. Being sure to include the characteristics of

state-level hate crime laws for the same year as the census data, the nine additional

variables were added based on the ADL source. 4

The variables were:

1. Did the state have a bias crime statute?

2. Was civil action available as a remedy within the state?

3. If the state had a bias crime statute, did it include a sentencing enhancement

provision?

4. In the state’s bias crime statute, was race, ethnicity, and religious bias included? 5

5. In the state’s bias crime statute, was sexual orientation bias included?

6. In the state’s bias crime statute, was gender based bias included?

7. Did the state have a legal provision prohibiting Institutional Vandalism?

8. Did the state have a law which included bias crime data collection?

9. Did the state authorize training by statute on bias crimes for law enforcement

personnel?

These variables were coded as dichotomous, nominal-level measures (1 ¼ yes; 0 ¼ no), and logistic regression was used to determine what might best predict hate or bias

crime prosecutions. These variables are important to the present exploratory study

208 Race and Justice 2(3)

because our purpose is to determine the predictive value of hate crime law character-

istics on the prosecution of hate crimes (McVeigh, Welch, & Bjarnason, 2003). More

specifically, the following research questions 6

were examined:

1. What are the prosecution patterns for states which do and do not have bias crime

statutes?

2. How does the presence of civil remedies in a state influence bias crime

prosecutions?

3. How does the available tool of sentencing enhancements influence bias crime

prosecutions?

4. For states which specifically include race, ethnicity, and religious based bias in

their statutes, how does this impact bias crime prosecutions?

5. How does the presence of protections based on sexual orientation in bias crime

statutes influence hate crime prosecutions?

6. How does the presence of protections based on gender in bias crime statutes influ-

ence hate crime prosecutions?

7. How does the presence of legal provisions prohibiting institutional vandalism

impact bias crime prosecutions?

8. Does the presence or absence of a state-level data collection provision for bias

crimes influence prosecutions?

9. Does the presence or absence of a state-level law enforcement training provision

for bias crimes influence prosecutions?

The number of research questions presented above reflects the amount of information

on hate crime law characteristics from the ADL. To be comprehensive, we wished to

examine each hate crime law characteristic as it might impact prosecutorial practices

in bias crime cases.

Results

The results of this study are both descriptive and inferential. Descriptively, we

examined the frequency of hate crime prosecutions compared to other types of

selected crimes which prosecutor’s offices address at the local level and also exam-

ined the cross-tabulation of the presence of a hate crime law in states by hate crime

prosecutions. The descriptive data are presented in order to show the basic patterns

of hate crime prosecutions. Inferentially, we conducted a number of nonlinear logistic

regression analyses to determine the model, which best explained hate crime prosecu-

tions given the available variables from the original data set and those which were

added based on information from the ADL.

Table 1 shows the frequency of hate crime prosecutions as reported by the pro-

secutor’s offices compared to other types of special crimes typically addressed at a

local or county level. The question within the survey read, ‘‘During the past 12

months, did your office prosecute any of the following kinds of felony offenses?’’

What followed was a listing of various offenses and some of these were selected for

Byers et al. 209

Table 1. 7

In 2001, nearly 80% of the offices responding stated that they had not prosecuted a hate crime in the past 12 months. Such low level of prosecution patterns

did not exist for domestic violence, stalking, or child abuse cases. The only other

offense type which also had a higher percentage of ‘‘no’’ responses was elder abuse.

Even in this instance, the number of ‘‘yes’’ responses for elder abuse prosecutions was

more than double the number of ‘‘yes’’ responses for hate crime prosecutions.

We also descriptively examined the cross-tabulation of whether the state had a hate

crime law and if the jurisdiction prosecuted a hate crime in the past 12 months. As

presented in Table 2, if a state had a law the jurisdiction was more likely to report

Table 1. Frequency and Percent of Reported Prosecutions

f %

Hate crime Yes 427 20.2 No 1,690 79.8

Domestic violence Yes 2,025 95.7 No 92 4.3

Elder abuse Yes 881 41.6 No 1,236 58.4

Stalking Yes 1,293 61.1 No 824 38.9

Child abuse Yes 1,974 93.2 No 143 6.8

Note. Missing cases were excluded and before exclusion did not exceed 9.6% within any category.

Table 2. Prosecution of Hate Crime in the Previous 12 Months, by Bias-Motivated Violence and Intimidation Statute in the State According to the Antidefamation league (ADL)

Bias-Motivated Violence and Intimidation

Statute in the State According to the ADL

Yes No

Prosecuted a hate crime Yes 21.8% 7.9% (407) (20)

No 78.2 92.1 (1,457) (233)

Total 100% 100% (1,864) (253)

Note. Pearson w2 ¼ 26.84; df ¼ 1; p < .001. There were 224 missing cases accounting for 9.6% of the total sample of 2,341 with a final sample of 2,117.

210 Race and Justice 2(3)

having prosecuted a hate crime in the past 12 months (21.8% vs. 7.9%). However, it is noteworthy that the vast majority of respondents reported that they had not prosecuted

such cases in the past 12 months even if the state had a law (78.2%) or if the state had no law at all (92.1%). What is equally interesting is the fact that the 21.8% of respon- dents reporting that they had prosecuted a hate crime law represents the 88% of responding offices with hate crime laws in their jurisdictions. It is possible for a pro-

secutor in a jurisdiction without a hate crime statute to prosecute a hate crime as a pre-

dicate offense (murder, rape, robbery, assault, arson, etc.). If prosecuted in a state with

a hate crime statute, the offense is more likely to be charged as a substantive offense

which adds the label of ‘‘hate crime’’ or ‘‘bias crime’’ (or other labels) to the charge.

Our final step was to conduct a logistic regression 8

to determine which variables,

originally within the data set, along with those which were added, best predict hate

crime prosecutions—our dependent variable. Several models were constructed to

make this determination and the model which best explained hate crime prosecutions

included eight variables. All cases examined in the logistic regression phase of this

study were from states/jurisdictions with hate crime laws as described by the ADL.

The rationale for selecting jurisdictions with hate crime laws is based on the fact that

these cases represent the majority of responding prosecutors and prosecutors in states

without a hate crime law are not likely to prosecute such cases (see Table 2).

Before discussing the logistic regression analyses and models, we wish to present

the percentages and frequencies for the independent variables within the regression

equations. The first independent variable in Table 3, ‘‘Prosecutor’s Office Assigned

Staff to Community Activity,’’ is the result of respondent self-report. The remaining

variables in Table 3 are based on the ADLs census of state hate crime laws as entered

into the data set based on the state code for each respondent.

Two logistic regression models were tested to determine the best fitting model to

explain hate crime prosecutions. The variables in each analysis included all variables

Table 3. Percentages and Frequencies for Offices on Selected Independent Variables

Attributes

Variable Yes No

Prosecutor’s office assigned staff to community activity 23.5% (485) 76.5% (1,577) State hate crime law protecting sexual orientation 31.0 (640) 69.0 (1,422) Civil action allowed by statute for bias crime 62.5 (1,288) 37.5 (774) Sentencing enhancement allowed by statute for bias crime 67.2 (1,386) 32.8 (676) Race, religion, and ethnicity as protected groups in hate crime

statute 75.7 (1,560) 24.3 (502)

State hate crime law protecting gender 36.9 (761) 63.1 (1,301) Institutional vandalism proscribed for hate crimes by

state statute 73.9 (1,524) 26.1 (538)

Legislated data collection provision for hate crimes 58.5 (1,207) 41.5 (855) Legislated training on bias crime for law enforcement 23.2 (478) 76.8 (1,584)

Byers et al. 211

added by the researchers based on the state-level characteristics of hate crime laws as

provided by the ADL along with the original variables within the data set. Model 1

presented in Table 4 includes eight independent variables, which were regressed on

the dependent measure of whether or not jurisdictions reported a hate crime prosecu-

tion within the past 12 months. This model explained approximately 16.3% of the var- iance in hate crime prosecutions (Nagelkerke R

2 ¼ .163). The three specific variables adding explanatory power to the model included ‘‘community,’’ ‘‘R/R/E,’’ and ‘‘Inst.

Vandalism.’’ ‘‘Gender’’ was close to being significant in the model but did not quite

reach the acceptable p < .05 threshold (p ¼ .054). Of those states with bias crime laws, as represented by local prosecutor’s offices, prosecutors who assigned a staff member

to community activities were nearly 4 times more likely to prosecute a hate crime

(Exp(B) ¼ 3.848). Prosecutors in states with hate crime law which articulate bias motivation based on race, religion, and ethnicity are almost 3 times more likely to pro-

secute hate crimes, Exp(B) ¼ 2.932. Finally, prosecutors in states with laws prohibit- ing Institutional Vandalism were nearly two and half times more likely to prosecute

hate crimes, Exp(B) ¼ 2.307.

Table 4. Logistic Regression Results for States With Hate Crime Laws (Model 1)

Variable B SE Wald p Exp (B) 95% Confidence Interval

(CI) for Exp (B)

Community 1.348 .123 119.413 .0001 3.848 [3.002, 4.900] Civil action .171 .161 1.118 .290 1.186 [.865, 1.627] Enhancement �.079 .143 .301 .584 .924 [.698, 1.224] R/R/E 1.076 .219 24.029 .0001 2.932 [1.907, 4.507] Gender .383 .199 3.699 .054 1.467 [.993, 2.168] Inst. vandalism .836 .194 18.543 .0001 2.307 [1.577, 3.375] Data collection .081 .146 .312 .576 1.085 [.815, 1.444] L.E. training .043 .194 .050 .824 1.044 [.714, 1.526] Constant �3.598 .243 218.441 .0001 .027

Note. df for all variables equals 1. Variable Descriptions: Community—this variable addresses whether or not a prosecutor’s office assigned staff to community activities, such as working with or meeting with community groups. Civil action—this variable addresses whether or not a state statute allows civil action in hate crime cases. Enhancement—this variable describes whether or not a state has, as part of its bias crime statute, a senten- cing enhancement provision. R/R/E—this variable defines whether or not a state explicitly includes bias based on race, religion, and eth- nicity within its bias crime statute. Gender—this variable defines whether or not a state’s bias crime statute includes gender bias as a bias motivator. Inst. Vandalism—this variable defines whether or not a state has a provision for institutional vandalism (e.g., vandalism against houses of worship, cemeteries, etc.) Data Collection—this variable defines whether or not a state has a legal provision of data collection of bias or hate crime cases. L.E. Training—this variable defines whether or not a state mandates law enforcement training for bias or hate crime cases.

212 Race and Justice 2(3)

Model 2, presented in Table 5, also included eight independent variables. This

model was constructed by adding ‘‘sexual orientation.’’ However, when this variable

was added, the Hosmer and Lemeshow goodness-of-fit statistics indicated there was

an assumption violation indicating a potential confounding effect of two or more vari-

ables in the model (p < .05). Each variable was added and subtracted from the model to

ascertain this violation, and it was discovered that ‘‘sexual orientation’’ was interact-

ing with ‘‘Law Enforcement Training.’’ When the latter variable was dropped from the

model, the assumption violation disappeared. This final best fitting model explained

an estimated 17% of the variance in hate crime prosecutions (Nagelkerke R2 ¼ .170). In Model 1, three variables were significantly correlated with hate crime prosecutions.

In Model 2, four variables were significant predictors of hate crime prosecutions,

while in Model 2, four variables were significant predictors. As was the case in Model

1, the variables ‘‘community,’’ ‘‘R/R/E,’’ and ‘‘Inst. Vandalism’’ were all significant

for Model 2. In addition to these, ‘‘Sex Orientation’’ was also found to be significant.

Model 2 reveals that having staff assigned to community-related activities increases

the likelihood or hate crime prosecution nearly 4 times, Exp(B)¼3.894, a state hate crime law protecting those from bias based on sexual orientation increases the

Table 5. Logistic Regression Results for States With Hate Crime Laws (Model 2)

Variable B SE Wald p Exp (B) 95% Confidence Interval

(CI) for Exp (B)

Community 1.359 .124 120.204 .0001 3.894 [3.054, 4.965] Civil action �.024 .170 .019 .889 .977 [.700, 1.627] Enhancement �.170 .145 1.359 .244 .844 [.635, 1.122] R/R/E 1.101 .219 25.276 .0001 3.006 [1.957, 4.617] Sex. orientation .593 .183 10.522 .001 1.809 [1.264, 2.589] Gender .056 .190 .087 .767 1.058 [.729, 1.536] Inst. vandalism .688 .179 14.732 .0001 1.990 [1.400, 2.827] Data collection .024 .145 .028 .868 1.024 [.771, 1.361] Constant �3.598 .243 218.441 .0001 .027

Note. df for all variables equals 1. Variable Descriptions: Community—this variable addresses whether or not a prosecutor’s office assigned staff to community activities such as working with or meeting with community groups. Civil action—this variable addresses whether or not a state statute allows civil action in hate crime cases. Enhancement—this variable describes whether or not a state has, as part of its bias crime statute, a senten- cing enhancement provision. R/R/E—this variable defines whether or not a state explicitly includes bias based on race, religion, and ethnicity within its bias crime statute. Sex. Orientation—this variable defines whether or not a state’s bias crime statute includes sexual orientation as a bias motivator. Gender—this variable defines whether or not a state’s bias crime statute includes gender bias as a bias motivator. Inst. vandalism—this variable defines whether or not a state has a provision for institutional vandalism (e.g., vandalism against houses of worship, cemeteries, etc.) Data collection—this variable defines whether or not a state has a legal provision of data collection of bias or hate crime cases.

Byers et al. 213

likelihood of hate crime prosecution by nearly 2 times, Exp(B) ¼ 1.809, the presence of race, religion, and ethnicity as protected classes increases the prosecution likeli-

hood by 3 times, Exp(B) ¼ 3.006, and the presence of an institutional vandalism provision in state hate crime law increases the likelihood of prosecution nearly 2

times, Exp(B) ¼ 1.990.

Discussion

These results, then, provide us with some discussion points regarding our original

research questions. Our first question regarding the prosecution patterns for states

which do and do not have bias crime statutes can be addressed generally from the

analyses found in the first two tables. It seems evident that hate crimes are not pro-

secuted at a high level as reported by prosecutors. This speaks to the literature on the

difficulties and controversies one might encounter when attempting to prosecute hate

crimes. This seems to also be the case even if the state where the prosecutor is

practicing has a hate crime statute. Our second through ninth research questions can

best be addressed from the analyses found in Tables 3 and 4. While the variable

dealing with prosecutors assigning staff to community-related activities was not orig-

inally contemplated within the research questions, it did surface as an important vari-

able within both models. It appears, then, that prosecutor’s offices with staff involved

in community-related activities are more likely to prosecute hate crimes. We might

account for this finding by concluding that those offices involved in community activ-

ities may be more sensitive to bias crimes due to their community-prosecution orien-

tation. This finding is supported in other research as well (King, 2008).

It was not surprising to us to find support for the research question dealing with the

presence of race, ethnicity, and religion provisions in state hate crime laws, which was

also supported in both models. The majority of hate crime cases, based on previous

research, are motivated by racial, religious, and ethnic bias. As a result, and when a

prosecutor’s office addresses a bias crime case, it is more than likely going to involve

one of these three types of bias motivation. Perhaps, given this familiarity, states with

hate crime provisions which articulate these forms of prohibited biases are more likely

to prosecute such cases. We also found that the addition of sexual orientation in state

hate crime laws was a predictor of hate crime prosecutions as found in Model 2. Pre-

vious research has demonstrated that not only is there strong advocacy in the United

States based on sexual orientation but this also emerges in the research literature (Byers,

Becker, & Opiola, 2003). Strong advocacy might impact hate crime prosecutions based

on antisexual orientation bias when states include this provision in their laws.

The final research question with empirical support in both models is that prose-

cutors in states with provisions prohibiting Institutional Vandalism, whether or not

these states have a bias crime statute, are more likely to prosecute bias crimes.

Institutional Vandalism, as advocated by the ADL, is essentially a second category of

bias crime which can accompany a state’s bias crime statute. The ADL advocated for

this second provision in hate crime law and many states followed this recommendation

(Jacobs & Potter, 1998, p. 34). Institutional Vandalism implies some level of ‘‘greater

214 Race and Justice 2(3)

harm,’’ given that this type of predicate offense takes into account its direct effects on

institutions (churches, synagogues, cemeteries, mortuary, school, educational facility,

community center, grounds adjacent to and owned, or rented by a group outlined

above), which include the individual and collective effects on many different people. As

a result, one might conclude that those states with Institutional Vandalism provisions are

those which (a) may be more deeply committed legislatively to the prosecution of hate

crimes and (b) may be more likely to accept the ‘‘greater harm hypotheses’’ (Jacobs &

Potter, 1998, p. 84) argument that the deleterious effects of hate crimes can, and often

are, be felt beyond those directly victimized (Johnson & Byers, 2003).

Study Findings and Policy Implications

Several research questions were examined in this study. While there were many

research questions, nine in total, the findings (whether positive or negative) revealed

interesting and useful insights regarding hate crime prosecutions as discussed below.

Regarding the first research question, we found that states with hate crime laws were

more likely to have prosecutors who reported having charged hate crime cases.

However, and even though there was some evidence in support of Research Question

1, this finding is tempered by the fact that the vast majority of jurisdictions reported

they had not prosecuted a hate crime in the past 12 months. This finding speaks to the

policy implication of how difficult it is to prosecute hate crimes based on evidentiary

issues and the legal requirement in criminal cases to prove a case beyond a reasonable

doubt as well as the controversial nature of hate crime prosecution. Having state-level

legal provisions for civil remedies, sentencing enhancements, gender bias protections,

data collection efforts, and state level-law enforcement training for hate crime had no

influence on the likelihood of hate crime prosecutions. Even though these findings did

not assist in the prediction of hate crime prosecutions, it is important to discuss the

policy implications herein. When civil remedies are available, this may dissuade crim-

inal prosecution. The burden of proof in civil matters is lower than in criminal cases.

The standard in civil cases is the ‘‘preponderance of the evidence’’ rather than

‘‘beyond a reasonable doubt.’’ As a result, hate crime cases may result in legal remedy

more often as civil cases rather than criminal cases. This has certainly been true for

cases taken on by the Southern Poverty Law Center. Pertaining to the presence of

state-level sentencing enhancements, this provision may also present problems for

prosecutors. Attempting to prosecute a hate crime and prove the case beyond a reason-

able doubt is but one hurdle a prosecutor must surpass. Determining that an alleged

offender deserves additional penalty for having committed a bias crime is an addi-

tional hurdle. This twofold issue may become more complicated for prosecutors when

the media and community groups learn of a case. This places a tremendous burden on

prosecutors and some may not wish to either face the question of why a sentencing

enhancement was not pursued or why a sentencing enhancement was not granted

by the court. Controversy could also surround ‘‘gender bias’’ prosecutions. Gender

bias as a protected category is one of the most contentious issues found within the field

of hate studies. Within prosecutorial practice, distinguishing gender bias from, for

Byers et al. 215

example, domestic violence can be a daunting task. The last two variables, which did

not predict bias crime prosecutions, were data collection and law enforcement train-

ing. Data collection provisions were not surprising to us. However, law enforcement

training was perplexing. Since bias or hate crime cases are first handled by police

departments, it would seem that having bias crime investigatory training would increase

the likelihood of prosecution. Given the data year, law enforcement training, as well as

prosecution, was in its infancy. Therefore, it is possible that thorough and complete

understanding of the elements of bias crime had not adequately been articulated and

augmented with bias crime case experience, thus reducing the likelihood of prosecution.

There was support, however, for Research Questions 4, 5, and 7 in influencing the

likelihood of hate crime prosecutions. As indicated in Model 2, having prosecutorial

staff assigned to community-related activities increased the likelihood of hate crime

prosecutions by nearly 4 times, states with legal provisions articulating prohibitions

against racial, religious, and ethnic-motivated hate crime increased the likelihood of

bias crime prosecutions over 3 times, the inclusion of sexual orientation bias in hate

crime law increased the likelihood of hate crime prosecution by nearly 2 times, and

having a legal provision, as advocated by the ADL, against Institutional Vandalism

increased the chance of hate crime prosecution by almost 2 times. Therefore, these

factors best explained hate crime prosecutions and would suggest that these have

important policy implications. Involvement in community-related activities appears

to be particularly important. Such activity places the face of the prosecutor’s office

in the community and could encourage more coordination and collaboration between

prosecutor’s offices and citizens. This could lead to a deeper understanding of

community needs and empathy regarding community issues including hate crimes.

Study Critique and Suggestions for Future Researchers

All studies have strengths and weaknesses. There are several strengths worth noting.

First, the study is national in scope and is a census of county-level prosecutors.

Second, the study utilized the only available national data on prosecutor self-

reports on hate crime prosecutions. Third, the study sheds empirical light on what fac-

tors do and do not appear to impact hate crime prosecutions. As with any study, the

present effort has weaknesses as well. We would like to note three. First, using

archival data is always somewhat problematic in the sense that it was not originally

collected as applied in a subsequent study. Second, the dependent variable limited the

research in terms of understanding the process surrounding hate crime prosecutions.

Given it was an outcome variable (did they or did they not prosecute a hate crime) as

opposed to a process variable (how they went about making the decision to prosecute

or not), limits the researchers in more fully understanding the decision-making process.

Third, assumptions had to be made regarding how familiar prosecutors were of the state-

level legal provisions found in hate crime legislation. This may not have been a safe

assumption especially given the small number of cases prosecuted nationally.

Another issue worthy of attention is the age of the data set used in this study. We

believe that prosecutors would respond much the same way as they did in 2001. The

216 Race and Justice 2(3)

issues and difficulties surrounding hate crime prosecution which existed then are in

existence now. At the time of the data collection, state hate crime laws were well

established across the nation. However, future research should continue to examine

hate crime prosecution through subsequent data collection efforts. The National

District Attorney’s Association should consider addressing the following issues in

their future data collection: prosecutor’s attitudes toward their state hate crime laws;

prosecutor’s attitudes regarding current hate crime laws in general; prosecutor per-

spectives on how hate crime laws should change; and, finally, how cyber hate crimes

may have changed the definition, interpretation, and application of hate crime statutes.

Future researchers interested in studying the hate crime prosecutions may find the

following suggestions helpful. First, future research will need to investigate through

original research efforts what factors influence prosecutorial decision making in bias

crime cases. Currently, research (including the present study) has focused on theo-

retical arguments, small sample qualitative studies, and research using existing data.

Second, future researchers may ask a large sample of prosecutors what factors

influence hate crime prosecution decisions. This could be done directly using social

surveys or indirectly using the Factorial Survey Method (Rossi & Nock, 1982). The

latter would allow the researcher more experimental control within a survey research

methodology. Third, future inquiries might examine the differences in hate crime

prosecutions in states with hate crime statutes which do include sexual orientation

versus those states that do not and states without any hate crime legislation. It would

be particularly interesting to examine this in the light of other political factors in the

states such as gay marriage provisions and gay adoption.

Notes

1. The term hate crime and bias crime are often used interchangeably and are therefore both

used within this article.

2. Some state statutes articulate that a hate crime is one which is ‘‘motivated’’ by bias against a

particular group which is then directed at an individual. Other statutes state that a hate crime

is an offense whereby the perpetrator ‘‘selected’’ the victim based on group membership.

3. For further discussion on this issue, see Becker, Byers, and Jipson (2000).

4. The original ADL source presents the presence or absence of the hate crime law character-

istic in a dichotomous format. The variables are presented here in question format so they are

understandable to the reader.

5. According to the ADL census of state hate or bias crime laws, there are states which provide

protections against bias or hate crime but these are not protections for individuals which are

designated in certain state laws through the protections based on race, ethnicity, and/or reli-

gion. Rather, these are institutional protections against ‘‘Institutional Vandalism,’’ which are

directed at protecting groups of people but the institution is the victim. Therefore, this ques-

tion differentiates states with provisions for race, ethnicity, and/or religion to protect individ-

ual victims versus those which only address Institutional vandalism.

6. The authors use research questions rather than hypotheses given the data do not lend them-

selves to hypothesis testing as is customary in experimental design.

Byers et al. 217

7. We excluded health care fraud, bank, or thrift fraud, telemarketing fraud, illegal sale, or pos-

session of a firearm, and police use of excessive force because these offenses might also be

addressed federally while the offenses selected are almost exclusively local-level prosecu-

tions and involve some level of vulnerability based on the victim’s status.

8. Logistic regression is the appropriate analytical tool for variables which are nominal or

dichotomous.

9. These percentages and frequencies are based on the effect of the listwise deletion in the

regression analysis. Therefore, these represent the frequencies upon which the variables

were regressed.

References

American Prosecutor’s Research Institute. (1998). Prosecutorial response to hate and bias—

Motivated crime. A report submitted to the National Institute of Justice, U.S. Department

of Justice, Washington, DC.

Becker, P. J., Byers, B., & Jipson, A. (2008). Hate crime legislation in three border states:

Experiences of Indiana, Kentucky and Ohio. Criminal Law Bulletin, 44, 241–255.

Bell, J. (2002). Policing hatred: Law enforcement, civil rights, and hate crime. New York:

New York University Press.

Byers, B., Becker, P., & Opiola, K. (2003). Critical examination of hate crime scholarship and

the underrepresentation of African Americans in victimization research. In M. D. Free, Jr.

(Ed.), Racial issues in criminal justice: The case for African-Americans (pp. 113–136).

Westport, CT: Praeger.

Byers, B., & Zeller, R. A. (2001). Official hate crime statistics: An examination of the

‘epidemic hypothesis’. Journal of Crime & Justice, 24, 73–86.

Byers, B., & Zeller, R. A. (1997). An examination of official hate crime offense and bias

motivation statistics for 1991-1994. Journal for Crime & Justice, 20, 91–106.

Franklin, K. (2002). Good intentions: The enforcement of hate crime penalty-enhancement

status. American Behavioral Scientist, 46, 154–172.

Finn, P. (1988). Bias crime: A special target for Prosecutors. Prosecutor, 21, 9–15.

Garofalo, J., & Martin, S. (1993). Bias motivated crimes: Their characteristic and law enforce-

ment response. National Institute of Justice. Carbondale: Carbondale, Illinois University

Center of the Study of Crime, Delinquency, and Corrections.

Haider-Markel, D. P. (2002). Regulating hate: State and local influences on hate crime law

enforcement. State Politics & Policy Quarterly, 2, 126–160.

Hodge, J. P. (2011). Gendered hate: Exploring gender in hate crime law. Boston, MA:

Northeastern University Press.

Jacobs, J. B., & Potter, K. (1998). Hate crimes: Criminal law & identity politics. New York,

NY: Oxford University Press.

Jenness, V., & Ryken, G. (2004). Policing hate crime in California. Berkeley, CA:California

Policy Research Center.

Johnson, S., & Byers, B. (2003). Attitudes toward hate crime laws. Journal of Criminal Justice,

31, 227–235.

King, R. (2008). Conservatism, institutionalism, and the social control of intergroup conflict.

American Journal of Sociology, 113, 1351–1393.

218 Race and Justice 2(3)

Levin, J., Rabrenovic, G., Ferraro, V., Doran, T., & Methe, D. (2007). When a crime

committed by a teenager becomes a hate crime: Results from two studies. American

Behavioral Scientist, 51, 246–257.

McPhail, B. A., & DiNitto, D. M. (2005). Prosecutorial perspectives on gender-bias hate crimes.

Violence Against Women, 11, 1162–1185.

McPhail, B., & Jenness, V. (2005). To charge or not to charge?- that is the question: The pursuit

of strategic advantage in prosecutorial decision making surrounding hate crime. Journal of

Hate Studies, 4, 89–119.

McVeigh, R., Welch, M., & Bjarnason, T. (2003). Hate crime reporting as a successful social

movement outcome. American Sociological Review, 68, 843–867.

Phillips, N. (2009). The prosecution of hate crimes: The limitations of the hate crime typology.

Journal of Interpersonal Violence, 24, 883–905.

R.A.V. v. The City of St. Paul, MN, 505 U.S. 377 (1992).

Read, E. (2005). Put to the proof: Evidentiary consideration in Wisconsin hate crime prosecu-

tions. Marquette Law Review, 89, 453–474.

Rossi, P. H., & Nock, S. L. (Eds.). (1982). Measuring social judgments: The factorial survey

approach. Beverly Hills, CA: Sage.

Shively, M. (2005). Study of literature and legislation on hate crime in America. Final report

submitted to the National Institute of Justice, Washington, DC (NCJ 210300). Retrieved

from www.ncjrs.gov/pdffiles1/nij/grants/210300.pdf

Wisconsin v. Mitchell, 508 U.S. 476 (1993).

Bios

Bryan D. Byers is a professor of criminal justice and criminology at the Ball State

University. His field experiences include positions as a prosecutor’s adult protective

services investigator, deputy coroner, mental health liaison, and criminal justice trai-

ner. He has published over thirty articles and book chapters on such topics as hate

crime, crisis intervention, elder abuse, and death and dying. He has seven books to his

credit in social psychology, statistics, elder abuse, and crisis intervention.

Kiesha Warren-Gordon is an assistant professor in the Department of Criminal Jus-

tice and Criminology at the Ball State University. She received her PhD in 2003 from

the Department of Sociology, Western Michigan University. Her primary interests

involve issues of race, ethnicity, and justice. Her work has appeared in various outlets

including, Journal of Forensic Science, Race, Class and Gender, Qualitative

Research Reports, and Race and Justice: An International Journal.

James A. Jones is the director of Research and Academic Effectiveness at the Ball

State University. He has been providing consultation and data analysis services to

faculty, graduate students, and staff at Ball State University for over 15 years. He has

coauthored with Bryan Byers on The Impact of Terrorist Attacks of 9/11 on Anti-

Islamic Hate Crime.

Byers et al. 219