Victim and Suspect Race and the Police Clearance of Sexual Assault

Abstract

Research testing the sexual stratification hypothesis has reached mixed conclusions about the relationship between victim and offender race and sexual assault case outcomes. Using data from the National Incident-Based Reporting System, we examine the relationship between the victim–suspect racial dyad and the odds of sexual assault case clearance, and whether the case was cleared via arrest versus an exceptional means clearance for reasons of declined prosecution or victim non-cooperation. The findings indicate little support for the sexual stratification hypothesis. Under certain circumstances, the victim–suspect racial dyad has a modest relationship with the odds of clearance and the manner in which cases are cleared. When the victim and offender are strangers, an arrest is less likely in White-on-Black assaults versus Black-on-White assaults. Further, an arrest is less likely in Black-on-Black family assaults because victims in these cases are less likely to cooperate with police, implying that Black victims’ reluctance or inability to cooperate with police plays an important and overlooked role in producing racial disparities in clearance and arrest outcomes in certain types of cases.

Keywords

race and policing sexual assault victims police clearance of crime exceptional clearance victim noncooperation

Historians and criminologists have long suspected that justice system officials view the rape of White women by Black males as a particularly serious offense; one that warrants swift and harsh repercussions in order to quell public fear and outrage (Brownmiller, 1975; Collins, 1975). Rape remains a politically charged crime in today’s legal culture. But much like the broader literature on racial disparities in the justice system (e.g., Crutchfield, Bridges, & Pitchford, 1994; Myers & Talarico, 1987; Peterson & Hagan, 1984; Spohn, 2000), studies of sexual assault case processing provide few definitive conclusions as to whether victim and suspect race influence the criminal justice system response to sexual assault (Kingsnorth, Lopez, Wentworth, & Cummings, 1998; Spohn, 1994; Spohn & Spears, 1996; Tellis & Spohn, 2008; Walsh, 1987).

Following the publication of the first comprehensive study of racial stratification and the treatment of rape cases by legal officials (LaFree, 1980), analyses of data from several jurisdictions indicated that legal officials appear to downplay the seriousness of rape when it involves Black victims and that officials punish Black males more harshly when they are suspected of assaulting White women (Spohn, 1994; Walsh, 1987). At the same time, findings from other studies suggest that race has little or no impact on the processing of rape cases, except in the context of stranger sexual assault (Spohn & Holleran, 2001). This research has raised crucial questions about the role that race may play in the processing of sexual assault cases that remain unsettled. Existing studies on sexual assault case processing are limited in two key respects. Prior research focuses on prosecution and sentencing outcomes and remains inattentive to the policing of rape despite evidence that law enforcement’s response to crime is sensitive to the social status of victims and suspects (Black, 1980; Howerton, 2006; Smith & Visher, 1981; Smith, Visher, & Davidson, 1984; Stolzenberg, D’Alessio, & Eitle, 2004) as well as evidence that sexual assault cases have significantly lower odds of being cleared by police compared to other violent crimes (Taylor, Holleran, & Topalli, 2009). More recent data show that the majority of rape complaints that come to the attention of police never proceed past the investigation and clearance stages (Spohn & Tellis, 2011; Tasca, Rodriquez, Spohn, & Koss, 2012), suggesting that the police play a crucial gate-keeping role in deciding which cases reach the court system (Alderden & Ullman, 2012), which may establish the context in which racial disparities initially develop and then accumulate as cases move downstream to prosecutors, judges, and juries (Poe-Yamagata & Jones, 2000; Stolzenberg, D’Alessio, & Eitle, 2013; Zatz, 2000).

The literature on race and sexual assault case processing has also been criticized for its near-exclusive focus on conflict-oriented theoretical explanations that are viewed as overly simplistic and static (Kingsnorth et al., 1998; Spohn, 1994). The most prominent of these, the sexual stratification hypothesis, proposes that the severity of the justice system response to rape reflects norms of selective enforcement that stem from underlying prejudicial attitudes on the part of law enforcement and prosecutors. This perspective assumes that any observed racial disparities that function to disadvantage Black suspects or Black victims reflect institutional bias on the part of police that is grounded in a broader system of social stratification (see, e.g., Smith & Visher, 1981). We draw upon the sexual stratification hypothesis to explore whether sexual assault complaints produce harsher outcomes for Black suspects accused of assaulting White women and more lenient outcomes for cases involving Black victims. This study departs from prior studies of sexual assault case processing in three ways. We broaden the generalizability of previous studies by analyzing the outcomes of a large sample of more recent sexual assault complaints that span jurisdictions throughout the United States. Second, we examine the influence of victim and suspect race on law enforcement’s handling of sexual assault rather than focusing on the prosecution and sentencing stages of case processing. We also build on prior studies that traditionally conceive of police discretion as a single dichotomous outcome—the probability of arrest. Police clearance of crime is often assumed to be synonymous with arrest, but the reality is that a nontrivial number of cases are cleared without an arrest due to circumstances outside of police control. These are referred to as exceptional clearances. The most frequent reason they occur is that prosecutors decline to press charges for reasons other than a lack of probable cause or because victims refuse or are unable to cooperate with investigators (Spohn & Tellis, 2012b, 2012a). The notion that victims and prosecutors may constrain police discretion to arrest has been overlooked in studies of the police response to sexual assault (though see Spohn & Tellis, 2011) and is rarely acknowledged as a potential source of the racial disparities observed in national arrest statistics (e.g., Snyder, 2011). We use data from the National Incident-Based Reporting System (NIBRS) to examine the role of victim and suspect race at these two pivotal disposition points, each of which determines on their own whether cases move past the victim reporting stage and toward prosecution.

The Sexual Stratification Hypothesis

The historical disparities in punishment for Black men accused of raping White women have been well documented (Banner, 2002; Spohn & Spears, 1996; Wriggens, 1995). During the 1800s, for example, the state of Georgia singled out such rapes as punishable by death, while White males convicted of raping a Black woman were subject only to fines or imprisonment (Wriggens, 1995; see also Spohn & Spears, 1996). Between 1930 and 1972, 405 of the 453 men executed for rape in the United States were Black men who had been convicted of raping White women (Tellis & Spohn, 2008). This legacy prompted contemporary research on the connections between victim and suspect race and legal reactions to rape. Most of this research interprets racial disparities through the lens of conflict theory and the sexual stratification hypothesis in particular. This perspective argues that women are viewed as sexual property and that access to women is subject to male competition and conflict (Collins, 1971, 1975). A culture which views sexual access as a form of resource competition is expected to foster a legal system in which officials use their discretion to mete out harsher sanctions to men from less powerful social groups who are accused of sexually assaulting women from elite social groups.

LaFree (1980) extended this logic to argue that conflict over access to women is grounded in a system of racial inequality that privileges White males and devalues Black women. According to this view, White women are viewed as a sexual commodity to which White men, by virtue of their higher social status, are presumably entitled exclusive access. Against this cultural backdrop, inter-racial sexual assaults involving Black males and White females are viewed as particularly threatening to the social position of White males. The sexual stratification hypothesis contends that legal officials will manage the perceived threat to White males through the use of sanctions that are expected to vary according to the racial composition of the victim–offender dyad. According to the sexual stratification hypothesis,¹ the expectation is that the criminal justice system will act on behalf of White males to punish Black suspects accused of assaulting White women more harshly than Black-on-Black, White-on-Black, or White-on-White sexual assaults. The theory also assumes that this system of sexual and race-based stratification that views the assault of White women as more serious intrinsically disadvantages Black female rape victims, and thus predicts that the rape of Black women by offenders of any race will be given lower priority by the justice system. In sum, Black men who assault White women are expected to receive the harshest sanctions for rape, followed by White men who assault White women, and Black men who assault Black women (Walsh, 1987). The justice system is expected to afford the greatest leniency to White men who sexually assault Black women.

The sexual stratification hypothesis was refined following the publication of research showing that the public and justice system actors often distinguish between “real” versus “simple” rapes (Estrich, 1987). More recent assessments of the hypothesis have considered whether the importance of race hinges on the nature of the relationship between suspects and victims. Officials may view acquaintance rapes involving Black suspects and victims as an extension of a Black subculture of crime, and hence, less serious than stranger Black-on-Black sexual assaults (Kingsnorth et al., 1998; Spohn, 1994; Spohn & Spears, 1996; Walsh, 1987). The victim–offender racial dyad may become particularly salient in the investigation of stranger sexual assaults due to political messaging and media depictions of high profile incidents that stoke anti-Black stereotypes and Whites’ fear of Black-on-White stranger violence (Beckett, 1997; Covington, 2010). Perhaps the most notable of these types of highly publicized incidents include the gang rape of a White female jogger in central park and the rape of a White woman by Willie Horton, a Black man, during a weekend furlough from a Massachusetts prison, where he was serving a sentence for murder in the 1980s. If evidence exists to suggest that police treat the stranger rapes of White women by Black males more seriously, this challenges the sexual stratification hypothesis’s narrow assertion that official reactions to rape are rooted in a system that views White women as the sexual property of White men and instead suggests that police reactions to such incidents reflect a broader fear of Black stranger violence among Whites.

Prior Research on the Role of Race in Sexual Assault Case Processing

With one exception (LaFree, 1980), studies testing the sexual stratification hypothesis have not focused on police outcomes. Empirical assessments of the theory rest on analyses of sexual assault case outcomes from the 1970s and 1980s during the post-arrest stages of prosecutorial charging (Kingsnorth et al., 1998; LaFree, 1980; Spohn & Spears, 1996; Tellis & Spohn, 2008), conviction (Kingsnorth et al., 1998; LaFree, 1980; Spohn & Spears, 1996), and incarceration (Kingsnorth et al., 1998; LaFree, 1980; Spohn, 1994; Spohn & Spears, 1996).² These studies provide mixed conclusions as to whether victim and suspect race exert a significant influence on these post-arrest outcomes, the stages of processing at which racial disparities emerge, and whether any observed disparities disadvantage Blacks. Two streams of research suggest that law enforcement’s handling of sexual assault complaints merits greater scrutiny: studies of police–citizen contacts and racial bias in arrest and recent studies showing that the attrition of rape cases from the justice system occurs disproportionately at the law enforcement stage (Spohn & Tellis, 2011).

The intensity of the police response to violent crime may be shaped by anti-Black stereotypes about “legitimate” victims and the perceived threat posed by Black males (DuMont, Miller, & Myhr, 2003; Rose & Randall, 1982). Research on traffic and pedestrian stops and searches (Engel & Calnon, 2004; Lundman & Kaufman, 2003) as well as drug arrests (Beckett, Nyrop, & Pfingst, 2006; Beckett, Nyrop, Pfingst, & Bowen, 2005; Rojek, Rosenfeld, & Decker, 2012) has produced a fairly consistent narrative on the import of race during police contact that is relevant to the current study. On one hand, Black residents voice concern that law enforcement officials selectively over enforce the law when it involves Black suspects, while at the same time, Black residents express dissatisfaction with police response times and perceive law enforcement as less concerned with the plight of Black residents who witness or are victims of crime (Brunson, 2007; Schafer, Huebner, & Bynum, 2003; Stewart, Baumer, Brunson, & Simons, 2009). If Black victims question the legitimacy of the police, or fear they will not be taken seriously when reporting a crime, they may be less willing to cooperate with police investigations, particularly for sensitive crimes such as sexual assault. The reluctance to cooperate with police can negatively impact the probability of clearance and arrest, which in turn may reinforce the perception of racially biased policing.

It is within this broader landscape of community relations that police respond to sexual assault complaints. However, few studies of police clearance and arrest outcomes in sexual assault cases have examined the influence of victim or offender race, and instead focus on police perceptions of victim credibility, the effects of victim–offender relationship, and the use of weapons and resultant injuries as influencing stereotypes about what constitutes “real rape” (e.g., Addington & Rennison, 2008; Alderden & Ullman, 2012; Bouffard, 2000; Pattavina, Morabito, & Williams, 2015; Tasca et al., 2012; Walfield, 2016). One exception is LaFree’s (1980) influential study of rape case processing in a Midwest City in the 1970s. The study examined whether Black sexual assault suspects receive harsher penalties than White suspects at multiple stages of the criminal justice system. The findings showed that while Black suspects receive longer sentences and are more likely to be incarcerated, they are no more likely than White suspects to be arrested or convicted. Since the publication of LaFree’s study, more recent police data generally show that sexual assaults suffer from significantly lower odds of clearance compared to other violent crimes, but that race does not play a significant role in predicting the police clearance of sexual assault (Taylor et al., 2009; see also Bouffard, 2000). Analyses of prosecutorial decisions immediately after an arrest yield similar conclusions—the strongest influence on sexual assault case outcomes at earlier stages of processing is not the race of the victim or suspect but rather suspect–victim relationship and victim or suspect use of alcohol or drugs (Tellis & Spohn, 2008). In sum, the results from several recent studies on the police clearance and prosecution of sexual assault are consistent with those from the 1970s—that rape cases suffer from comparatively high attrition rates during the early stages of processing but that victim and suspect race do not consistently influence whether complaints proceed to formal arrest and prosecution. These mixed findings, as well as the broader literature on racial bias and citizens’ perceptions of police legitimacy, suggest that victim and offender race warrant closer examination as a means to understand why only a fraction of rape cases are cleared by arrest, despite the fact that the majority of victims in sexual assault know and can identify their attacker.

The Current Study

We address several gaps in the literature on police clearance of sexual assault. We analyze data from a geographically broader sample of sexual assault cases from the NIBRS, which is one of the few data collections containing sufficiently large numbers of inter- and intra-racial sexual assaults. We also depart from previous studies of clearance by considering how race impacts both the probability and type of police clearance. The clearance process is typically operationalized as a series of binary outcomes: whether the case was cleared and whether the clearance results in an arrest (though see Bouffard, 2000). One study of sexual assault case processing in Los Angeles (Spohn & Tellis, 2011, 2012a), however, found extensive use of exceptional clearance rather than arrest as a means to clear sexual assault cases. Exceptional clearance is used when police have identified a suspect and established probable cause to support an arrest but are unable to arrest the suspect. The most frequent reason for exceptional clearance is victims’ reluctance to cooperate with authorities or because prosecutors decline to press charges (Spohn & Tellis, 2011, 2012b). The potential for prosecutors to contribute to racial disparities in sexual assault case processing through their ability to prevent arrests has not been examined in prior studies. The current analyses focus on whether the racial composition of the victim–suspect dyad is associated with the probability that a sexual assault complaint is cleared, as well as the method of clearance (arrest vs. exceptional clearance), and whether victim and offender race is particularly relevant for certain victim–offender relationships, such as stranger sexual assaults.

Method

Data Source and Sample

The data used in this study are drawn from the 2002 to 2009 NIBRS. NIBRS provides detailed information on the characteristics of crime incidents in participating jurisdictions, including the demographic characteristics of the victim(s) and suspect(s) as well as incident characteristics such as weapon use and injury. The initial sample consisted of all sexual assault incidents known to 3,239 participating law enforcement jurisdictions that reported to NIBRS from 2002 to 2009. These agencies span 35 states and each region of the United States.³

NIBRS is the only official data source in the United States that tracks the incident, victim, and suspect characteristics of sexual assaults known to law enforcement as well as whether each incident was cleared and the manner of clearance. The Federal Bureau of Investigation (FBI) considers a case cleared when four criteria are met: (a) at least one person is arrested, (b) that person is charged with the commission of an offense (c) turned over to the court for prosecution, or (d) the case is cleared by exceptional means (FBI, 2000). The FBI defines an exceptional means clearance as one in which the police have identified a suspect, know his or her whereabouts, and sufficient probable cause exists to support arrest and prosecution but forces outside of police control prohibit the arrest. These forces include but are not limited to: the death of the suspect, a denial of extradition by another jurisdiction in which the suspect has committed a crime and is being prosecuted, the victim’s refusal to cooperate with the prosecution after the suspect has been identified, or prosecution is declined (FBI, 2013b).

The initial sample consisted of 448,380 violent sexual assault incidents reported to law enforcement (forcible rape, forcible sodomy, sexual assault with an object, or forcible fondling⁴) that occurred between 2002 and 2009. We combined violent sexual assaults across multiple years to obtain a sufficient number of inter-racial sexual assaults, which occur less frequently than intra-racial sexual assaults.⁵ Given the sexual stratification argument’s focus on male assaults of females, the analyses were limited to those cases involving only male suspects and female victims resulting in 342,626 (76.4% of our original sample) male-on-female sexual assault incidents.⁶ Next, we restricted the sample to include only cases that were not cleared, that were cleared through an arrest, exceptionally cleared due to the prosecutor declining the case, or exceptionally cleared because of the victim’s refusal to cooperate with the investigation (n = 340,333). We removed a small number of cases (n = 2,293) that were cleared exceptionally due to the death of the suspect, a failure to get the suspect extradited from another jurisdiction, or because a juvenile suspect was handled informally, such as being remanded to the custody of a parent or guardian, without being taken into formal custody.⁷

We include in the analyses all cases involving male offenders’ ages 17 and older due to the fact that some states including South Carolina and Michigan, where all crime data are submitted via NIBRS, classify 17-year-olds as adults in the criminal justice system. Cases involving suspects under the age of 16 were removed from the analysis, since the majority of these cases are handled by juvenile courts, and prosecutors may play a limited role in deciding whether to decline or proceed with charges against minors (Griffin, Addie, Adams, & Firestine, 2011). Because some states that participate in NIBRS still classify 17-year-olds as juveniles, we include a dummy variable for 17-year-old suspects in all models to control for the potential that these cases are processed differently in those jurisdictions.⁸ We restricted the sample of victims to ages 12 and older since one dependent variable in our analyses, victim non-cooperation, assumes victim agency in the investigation process. Specifically, it assumes that victims understand their choice to not cooperate and that victims exercise that right accordingly, which may not be an accurate assumption for very young victims of crime. We included in the analyses a control for victims aged 12–17 since adolescents may react differently to victimization than adult victims and their reluctance to cooperate may not be treated by prosecutors and police as a barrier to proceeding with charges. The removal of these cases due to age reduced the sample to 215,603 sexual assault incidents. Finally, one observation had missing data on two important measures: victim–offender relationship and victim injury.⁹ This case was dropped, resulting in a final sample size of 215,602 male-on-female sexual assault cases that occurred between 2002 and 2009. Forty percent of these 215,602 sexual assault incidents (n = 86,253) were cleared and were included in subsequent analyses examining racial differences in clearance type.

Dependent Variables

Our analysis explores whether victim and suspect race are associated with two outcomes: the probability that police clear a sexual assault complaint and the method in which the case was cleared (by an arrest vs. an exceptional clearance). The initial clearance outcome, clearance, is a dichotomous variable that contrasts the log odds of the case being cleared (clearance = 1) with the odds of the case not being cleared (clearance = 0). Among the sample of cases that were cleared, we estimate the log odds of arrest versus each of the two exceptional clearance categories with a second dependent variable we refer to as clearance type. Clearance type is a trichotomous variable that contrasts the log odds that cases are cleared by arrest (clearance type = 1), versus cleared by exceptional means due to declined prosecution (clearance type = 2), and versus cleared by exceptional means due to victim non-cooperation (clearance type = 3). We distinguish between the two types of exceptional clearance because they represent outcomes that stem from decisions by two distinct actors, the victim and the prosecutor, and thus two potential sources of racial disparities in which solved cases result in more punitive outcomes (i.e., an arrest) versus those that result in attrition from the system and no arrest of the suspect.

Administrative data on reported crimes and police clearance are not without limitations. One study of sexual assault investigations in Los Angeles revealed that the categories of arrest and exceptional clearance are not always mutually exclusive (Spohn & Tellis, 2011). Approximately 20% of sexual assaults reported to the LA County Sheriff’s office and 32% of those reported to the Los Angeles Police Department in 2008 were originally arrests that were reclassified as exceptional clearances after the prosecutor declined charges. Neither of these two agencies submit data to NIBRS and are not part of the current study. However, these findings raise a question as to whether a small portion of cases in NIBRS that are classified as exceptional clearances may in fact have been arrests that were later dropped. This would represent a potential abuse of exceptional clearance policy by police. However, it would not fundamentally alter the fact that an officially recorded exceptional clearance, even if it was once an arrest, is a valid measure of attrition from the system as the result of case closure without an arrest leading to prosecution or conviction. At this time, there is no evidence to suggest that this practice is widespread throughout the United States, and as the authors of the Los Angeles study note, reclassification of arrest clearances is likely the product of LAPD policy on clearance that differ from definitions used by the FBI and that the practice also stems from a “historical context” of policing specific to Los Angeles (Spohn & Tellis, 2011, p. 1400).

Racial Composition of the Suspect–Victim Dyad

We used indicators of victim and suspect race¹⁰ provided in NIBRS to create the following racial dyads: Black suspect–Black victim, Black suspect–White victim,¹¹ White suspect–White victim, White suspect–Black victim, and one additional dyad variable that measures other racial pairings.¹² The sexual stratification hypothesis suggests that all other racial dyads will be afforded greater leniency (i.e., lower odds of clearance, lower odds of an arrest, and higher odds of declined prosecution) than Black-on-White sexual assaults. Thus, the Black suspect–White victim dyad is the reference category in the regression models.

Controls

A number of incident characteristics are included in the analysis as controls. We control for aggravating factors that prior research (Estrich, 1987; Spohn & Tellis, 2012b) has shown impacts the odds of clearing violent crimes, prosecutor charging decisions, and victim cooperation. These include a series of indicators for weapon use, injury, and victim–suspect relationship. We measure weapon use with a series of dichotomous variables that indicate whether a foreign object (gun, knife, blunt object, etc.), a personal weapon (hands, feet, fist, etc.), a weapon of unknown type, or no weapon was used. We chose to combine weapon types into the foreign object category because the majority of cases (94%) did not involve a traditional weapon. Only 1.5% of cases involved a gun and only 1.7% of incidents involved a knife, requiring that we combine them into a single category. Collateral victim injury is measured using three binary variables that indicate the presence of a minor injury, a major injury, or no additional injury (the reference category). Additionally, a series of four dummy variables indicate the type of sexual assault offense: forcible rape, forcible sodomy, sexual assault with an object, or forcible fondling. Incident location is measured with a series of dichotomous variables that include: in a residence (the reference category); at a school or college; in a bar, restaurant or hotel; in some other indoor public location such as stores, airports, and government buildings; in an outdoor public location (park, river, woods); or at some other location that the victim or police could not identify. We capture the time of day in which the incident occurred with a series of variables that indicate whether the incident occurred between midnight and 2:59 am, between 3 am and 5:59 am, between 6 am and 8:59 am, between 9 and 11:59 am, between noon and 2:59 pm, between 3 and 5:59 pm, between 6 and 8:59 pm, and between 9 and 11:59 pm (the reference category).¹³ We control for whether the offense was attempted versus completed and whether police cleared the incident within 24 hours of the assault (for cleared cases only).¹⁴ Controls for incident year were included to account for any variation in police processing due to unobserved factors such as changes in policies or laws guiding the processing of sexual assault. We also include controls for whether the incident included aggravating factors such as multiple victims, multiple suspects, and additional offenses such as a robbery, theft, or homicide.¹⁵

We also include a number of suspect and victim characteristics as controls for the perceived seriousness of sexual assaults (Horney & Spohn, 1996; Spohn & Tellis, 2012a). We measure victim and suspect age with the following age categories: victims and suspects who are aged 17 or younger,¹⁶ aged 18–24 (the reference category), aged 25–34, aged 35–49, and aged 50 and older. We also include a control that indicates whether the incident involved a juvenile victim and adult offender.¹⁷ The relationship between the victim and suspect is measured by a series of dichotomous variables that indicate whether the sexual assault occurred between strangers, current or former¹⁸ intimate partners (spouse, ex-spouse, boyfriend/ girlfriend), friends or acquaintances, or non-intimate family members.

Finally, in some incidents, values for the racial dyad, weapon, victim–offender relationship, and incident location were recorded by law enforcement as “unknown.” A value of unknown is distinct from missing data. The latter is a form of error due to no data being recorded for a data element; only one case was removed from the analyses due to “true” missing data. Consistent with prior studies using NIBRS data (Messner, McHugh, & Felson, 2004; Noonan & Vavra, 2007), we retained cases that had one or more unknown values in lieu of list-wise deletion of cases because a value of unknown is a valid response and data element value in NIBRS and is the equivalent of “don’t know” in survey research. For example, victims attacked by strangers may simply not know the race of the offender. Victims who were drugged or were intoxicated may not remember where they were attacked, whether they knew the attacker(s), or whether the offender used a weapon. All models include controls for cases having unknown values on these four variables.¹⁹

Analytic Technique

We begin our analyses by examining whether suspect and victim race are associated with the odds that police clear sexual assault incidents. We present the results of a binary logistic regression model that estimates the log odds that a sexual assault incident was cleared (1) versus not cleared (0), controlling for all covariates with the exception of time to clearance, which can be observed only for cleared cases. For the subset of sexual assault incidents that were cleared, we estimated a series of multinomial logistic regression equations predicting clearance type.²⁰ These models contrast the log odds of an arrest (1), the more punitive outcome, with two outcomes that result in no arrest and no formal charging of the suspect: declined prosecution (2) and victim refusal to cooperate (3).²¹

Finally, prior research reviewed earlier leads to the expectation that arrest may be especially likely for Black-on-White stranger assaults and significantly less likely for Black-on-Black acquaintance rape. We conclude by repeating our multinomial logistic regression models predicting clearance type on four subsets of cleared cases consisting of family, stranger, friend/acquaintance, and intimate partner sexual assaults to investigate whether the effect of race differs across these four subsamples. We estimated all regression models in Stata (version 12) using a robust variance estimator that adjusts for the clustering of sexual assault complaints within counties. Given that prosecutors, prescreening the cases operate at the county level, the potential exists that cases investigated within the same county are more likely to be disposed in similar ways. This robust estimator adjusts for any downward bias in standard errors that results from cluster-correlated observations (Rogers, 1993; Williams, 2000).²²

Results and Discussion

The descriptive statistics presented in Table 1 show considerable variation in the way in which sexual assault cases are handled. Sixty percent of sexual assaults were not cleared. Of the cleared cases, nearly two thirds (64.7%) end in an arrest, 19.2% of cases are cleared without an arrest because prosecutors declined to press charges, and 16.1% of cases were cleared without an arrest because victims refused to cooperate with the police. The majority of sexual assault incidents are intra-racial. Nearly two thirds (63.1%) of the cases involved White offenders and White victims, and a slightly higher percentage of cases that are ultimately cleared are White-on-White assaults (65.6%). Approximately 15% of all sexual assault cases, as well as those that are cleared by police, involved Black offenders and Black victims. Nearly 11% of cases involved Black suspects and White victims. White-on-Black sexual assaults account for 2% of the sexual assaults reported to NIBRS agencies from 2002 to 2009.

Table 1.

Descriptive Statistics for Variables Included in Analyses of Sexual Assault Clearance and Clearance Type.

	Full Sample		Cleared Sample
	(N = 215,602)		(N = 86,253)
Variables	Frequency (n)	(%)	Frequency (n)	(%)
Dependent variables
Case clearance
Cleared	86,253	40.0	—	—
Uncleared	129,349	60.0	—	—
Clearance type (N = 69,363)
Arrest	—	—	55,793	64.7
Prosecution declined	—	—	16,575	19.2
Victim noncooperation	—	—	13,885	16.1
Explanatory variables
Offender–victim racial dyad
White-on-White	135,952	63.1	56,617	65.6
Black-on-Black	33,827	15.7	13,388	15.5
Black-on-White (reference category)	23,731	11.0	9,270	10.7
White-on-Black	4,065	1.9	1,642	1.9
Other racial dyad	7,248	3.4	3,013	3.5
Unknown racial dyad	10,779	5.0	2,323	2.7
Control variables
Weapon/force used
Foreign object (gun, knife, blunt object, etc.)	13,218	6.1	5,599	6.5
Personal weapons (feet, teeth, hands, head)	136,328	63.2	58,174	67.4
Drugs/narcotics used as weapon	758	0.4	226	0.3
Unknown weapon used	8,565	4.0	2,658	3.1
No weapon used (reference category)	56,733	26.3	19,596	22.7
Victim injury
Serious injury	9,191	4.3	3,934	4.6
Minor injury	37,472	17.4	15,868	18.4
No additional injuries (reference category)	168,939	78.4	66,451	77.0
Multiple victims	10,412	4.8	5,622	6.5
Additional offenses occurred during assault	17,840	8.3	9,471	11.0
Multiple offenders	13,952	6.5	5,326	6.2
Victim–offender relationship
Stranger	22,520	10.4	7,997	9.3
Current or former intimate partner	28,870	13.4	13,751	15.9
Friend/acquaintance (reference category)	114,424	53.1	45,566	52.8
Family member	28,093	13.0	11,744	13.6
Relationship to victim unknown	21,695	10.1	7,195	8.3
Offender characteristics
Age
17	10,154	4.7	3,891	4.5
18–24	67,135	31.1	26,564	30.8
25–34	55,658	25.8	22,200	25.7
35–49	59,288	27.5	24,345	28.2
50 or older	23,367	10.8	9,253	10.7
Victim characteristics
Age
17 or younger	95,943	44.5	38,714	44.9
18–24	54,422	25.2	21,196	24.6
25–34	32,302	15.0	13,277	15.4
35–49	25,869	12.0	10,324	12.0
50 or older	7,066	3.3	2,742	3.2
Adult on child victimization	86,803	40.3	35,239	40.9
Most serious sexual assault offense to occur during incident
Forcible rape	123,472	57.3	48,099	55.8
Forcible fondling (reference category)	74,160	34.4	30,600	35.5
Forcible sodomy	9,059	4.2	3,788	4.4
Sexual assault with an object	8,911	4.1	3,766	4.4
Year
2002 (reference category)	20,823	9.7	8,561	9.9
2003	21,682	10.1	8,992	10.4
2004	24,861	11.5	10,225	11.9
2005	27,846	12.9	11,382	13.2
2006	29,773	13.8	11,620	13.5
2007	29,801	13.8	11,408	13.2
2008	29,945	13.9	11,891	13.8
2009	30,871	14.3	12,174	14.1
Location of offense
Home (reference category)	149,477	69.3	60,849	70.5
School	4,776	2.2	1,859	2.2
Bar/restaurant/hotel	11,274	5.2	4,610	5.3
Other indoor public location	11,949	5.5	4,881	5.7
Outdoor public location	23,551	10.9	8,643	10.0
Unknown location	14,575	6.8%	5,411	6.3
Time of offense
12:00–2:59 a.m.	42,533	19.7	17,192	19.9
3:00–5:59 a.m.	17,595	8.2	7,499	8.7
6:00–8:59 a.m.	14,760	6.9	6,136	7.1
9:00–11:59 a.m.	21,172	9.8	8,283	9.6
12:00–2:59 p.m.	26,784	12.4	10,175	11.8
3:00–5:59 p.m.	25,912	12.0	9,913	11.5
6:00–8:59 p.m.	24,851	11.5	9,889	11.5
9:00–11:59 p.m.	29,638	13.8	12,076	14.0
Assault completed	207,821	96.4	83,395	96.7
Cleared w/in one day of incident	—	—	28,858	33.5

Note. N = 172,884.

A majority of cases (57.3% of full sample, 55.8% of cleared sample) were forcible rapes, followed by forcible fondling (34.4% of full sample, 35.5% of cleared sample). The modal victim age category is 17 or younger, and approximately 40% of all cases involve a juvenile victim and an adult suspect. The majority of sexual assault incidents did not involve the use of any object weapon; in nearly two thirds of the incidents, the offender used a personal weapon, such as feet or fists, and approximately one quarter involved no weapon use. When weapons were used, foreign objects were the most common; rarely were drugs or narcotics used as weapons in sexual assault cases. Likewise, only a small percentage of incidents in both the full and cleared samples involved serious injury to the victim, and only a slightly larger percentage of cases involved minor injuries to the victim. The majority of cases involved friends or acquaintances, followed by assaults between intimates and family members. Approximately 10% of the sexual assaults involved strangers.

Table 2 shows the results of a binary logistic regression that estimates the effects of incident characteristics on the odds that a case was cleared. The results presented in Table 2 suggest that the racial composition of the victim–offender dyad of sexual assault incidents is not significantly associated with the odds of clearance, with the exception of incidents involving a White victim of a White suspect. Although statistically significant, the difference in odds of clearance for White-on-White assaults versus Black-on-White assaults is not substantively significant. The predicted probability of clearance for these incidents as shown in Table 2 indicates that each of the racial dyads has approximately equal probabilities of clearance (approximately 50% probability) compared with Black-on-White sexual assault cases. The only exception is when the race of the victim and offender is unknown, which is associated with a 30% probability of clearance compared to Black-on-White assaults.

Table 2.

Logistic Regression of Sexual Assault Case Clearance on Offender, Victim, and Incident Characteristics.

Variables	b	SE	Exp(b)	Predicted Probability
Explanatory variables
Offender–victim racial dyad (Black-on-White)
Black-on-Black	−.029	.063	0.971	.493
White-on-White	.062*	.031	1.064	.515
White-on-Black	.060	.040	1.062	.515
Other racial dyad	.050	.044	1.051	.512
Unknown racial dyad	−.815*	.057	0.443	.307
Controls
Weapon/force used (no weapon)
Foreign object (gun, knife, blunt object, etc.)	.269*	.076	1.309	.567
Personal weapons (feet, teeth, hands, head)	.322*	.051	1.380	.580
Drugs/narcotics used as weapon	−.204*	.097	0.815	.449
Unknown weapon used	−.134*	.060	0.875	.467
Victim injury (no injury)
Serious injury	.124*	.042	1.132	.531
Minor injury	.064*	.030	1.066	.516
Multiple victims	.395*	.028	1.484	.597
Additional offenses occurred during assault	.497*	.044	1.644	.622
Multiple offenders	−.086	.068	0.918	.479
Victim–offender relationship (friend/acquaintance)
Current or former intimate partner	.307*	.023	1.359	.576
Stranger	−.241*	.045	0.786	.440
Family member	.025	.024	1.025	.506
Relationship to victim unknown	−.231*	.054	0.794	.443
Offender age (18–24)
17	.057	.065	1.059	.514
25–34	.005	.015	1.005	.501
35–49	.047*	.017	1.048	.512
50 or older	.021	.024	1.021	.505
Victim age (18–24)
17 or younger	−.026	.070	0.974	.494
25–34	.042*	.017	1.043	.510
35–49	−.021	.019	0.979	.495
50 or older	−.023	.029	0.977	.494
Adult on child victimization	.131	.070	1.140	.533
Type of sex offense (forcible fondling)
Forcible rape	−.129*	.024	0.879	.468
Forcible sodomy	−.011	.046	0.989	.497
Sexual assault with an object	−.017	.054	0.983	.496
Location of offense (residence)
School	.021	.047	1.021	.505
Bar/restaurant/hotel	.077*	.026	1.080	.519
Other indoor public location	.138*	.036	1.148	.534
Outdoor public location	−.107*	.028	0.899	.473
Unknown location	−.025	.039	0.975	.494
Time of offense (9:00–11:59 p.m.)
12:00–2:59 a.m.	−.023	.030	0.977	.494
3:00–5:59 a.m.	.091*	.035	1.095	.523
6:00–8:59 a.m.	−.010	.042	0.990	.498
9:00–11:59 a.m.	−.106*	.041	0.899	.474
12:00–2:59 p.m.	−.147*	.033	0.863	.463
3:00–5:59 p.m.	−.130*	.025	0.878	.468
6:00–8:59 p.m.	−.064*	.024	0.938	.484
Assault completed	.039	.055	1.040	.510
Constant	−.632*	.099

Note. N = 172,884. Robust standard errors (SE) shown are adjusted for the clustering of cases within counties. Reference group used to interpret coefficients appears in parentheses. Formula for calculating predicted probability = exp(b)/((1+exp(b)).

*p < .05.

A number of aggravating factors are significantly associated with the likelihood of sexual assault case clearance. Specifically, the odds that a case is cleared increases by approximately 38%,²³ exp(b) = 1.380, when personal weapons are used in an incident than when no weapon is used and increases by 31% when foreign object weapons such as a knife or gun is used, exp(b) = 1.309. When drugs are used as a weapon in an incident; however, the odds of clearance are reduced by 18%, exp(b) = 0.815. Major and minor injury to the victim are associated with a significantly higher, 13%; exp(b) = 1.132 and 7%; exp(b) = 1.066, respectively, odds of clearance. Additionally, the odds of case clearance are 36% higher in cases, where the victim and offender are current or former intimate partners compared to assaults between friends or acquaintances, whereas stranger assaults are approximately 22% less likely to be cleared compared with acquaintance assaults, exp(b) = 0.786. We attribute the latter finding to the challenges associated with identifying suspects that are strangers. Overall, the findings in Table 2 suggest that victim and suspect race do not yield a significant influence on the odds of case clearance. The odds that a case is cleared is more strongly related to victim–offender relationship and the seriousness of the incident as measured by the presence of weapons and co-occurring offenses.

The results shown in Table 2 provide only a partial picture of the factors that shape clearance outcomes. The factors that predict an arrest clearance may be different than those that influence exceptional clearances. Exceptional clearance may be driven by prosecutors’ decisions to decline charges or by victim non-cooperation, each of whom can exercise discretion that impacts law enforcement’s ability to make an arrest. Table 3 presents the results of a multinomial logistic regression that contrasts the odds of exceptional clearance due to victim non-cooperation versus arrest and the odds of an exceptional clearance due to declined prosecution versus arrest.

Table 3.

Multinomial Regression of Clearance Type in Cases of Sexual Assault.

	Arrest Versus Victim Non-cooperation			Arrest Versus Prosecution Declined
Variables	B	SE	Exp(b)	b	SE	Exp(b)
Explanatory variables
Offender–victim racial dyad (Black-on-White)
Black-on-Black	−.063	.102	0.939	0.202	.112	1.224
White-on-White	.161*	.057	1.175	0.075	.056	1.078
White-on-Black	.124	.087	1.132	0.010	.078	1.010
Other racial dyad	.059	.080	1.061	0.037	.075	1.038
Unknown racial dyad	.915*	.148	2.497	0.612*	.179	1.844
Controls
Weapon/force used (no weapon)
Foreign object (gun, knife, blunt object, etc.)	.160	.129	1.174	0.075	.097	1.078
Personal weapons (feet, teeth, hands, head)	−.127	.096	0.881	−0.189*	.074	0.828
Drugs/narcotics used as weapon	−.529*	.185	0.589	−0.302	.192	0.739
Unknown weapon used	−.188	.127	0.829	−0.208*	.100	0.812
Victim injury (no injury)
Serious injury	.546*	.083	1.726	0.479*	.084	1.614
Minor injury	.407*	.042	1.502	0.543*	.045	1.721
Multiple victims	.915*	.069	2.497	0.653*	.063	1.921
Additional offenses occurred during assault	.962*	.071	2.617	1.185*	.059	3.271
Multiple offenders	.107	.110	1.113	0.163	.107	1.177
Victim–offender relationship (friend/acquaintance)
Current or former intimate partner	−.042	.043	0.959	−0.015	.044	0.985
Stranger	.066	.074	1.068	0.245*	.090	1.278
Family member	.340*	.052	1.405	0.028	.041	1.028
Relationship to victim unknown	−.061	.122	0.941	0.095	.081	1.100
Offender age (18–24)
17	.388*	.146	1.474	0.247	.142	1.280
25–34	.206*	.031	1.229	0.107*	.031	1.113
35–49	.345*	.034	1.412	0.173*	.033	1.189
50 or Older	.252*	.047	1.287	0.147*	.044	1.158
Victim age (18–24)
17 or younger	.537*	.157	1.711	0.319*	.149	1.376
25–34	−.051	.033	0.950	−0.056	.035	0.946
35–49	−.191*	.043	0.826	−0.144*	.036	0.866
50 or older	−.071	.062	0.931	−0.147*	.071	0.863
Adult on child victimization	.867*	.152	2.380	0.653*	.154	1.921
Type of sex offense (forcible fondling)
Forcible rape	−.360*	.050	0.698	−0.245*	.055	0.783
Forcible sodomy	.070	.098	1.073	−0.128	.084	0.880
Sexual assault with an object	−.111	.113	0.895	0.160	.101	1.174
Location of offense (residence)
School	−.023	.104	0.977	−0.251*	.086	0.778
Bar/restaurant/hotel	−.166*	.051	0.847	−0.140*	.061	0.869
Other indoor public location	.481*	.064	1.618	0.264*	.088	1.302
Outdoor public location	−.067	.048	0.935	0.086	.050	1.090
Unknown location	.001	.065	1.001	0.009	.069	1.009
Time of offense (9:00–11:59 p.m.)
12:00–2:59 a.m.	−.124*	.050	0.883	−0.033	.055	0.968
3:00–5:59 a.m.	−.066	.078	0.936	0.026	.082	1.026
6:00–8:59 a.m.	.005	.091	1.005	0.032	.100	1.033
9:00–11:59 a.m.	−.034	.082	0.967	−0.031	.079	0.969
12:00–2:59 p.m.	.001	.062	1.001	0.009	.061	1.009
3:00–5:59 p.m.	.055	.048	1.057	0.059	.047	1.061
6:00–8:59 p.m.	−.021	.044	0.979	0.052	.048	1.053
Assault completed	.114	.069	1.121	−0.131	.080	0.877
Cleared w/in one day of incident	1.938*	.103	6.945	2.448*	.101	11.565
Constant	−.039	.172		0.101	.149

Note. N = 69,363. Robust standard errors (SE) adjusted for clustering within counties are reported. Reference group used to interpret coefficients appears in parentheses.

*p < .05.

With one exception, the victim–suspect racial dyad is unrelated to the probability of arrest versus each type of exceptional clearance. White-on-White assaults, however, are significantly more likely, exp(b) = 1.175, to result in an arrest clearance versus victim non-cooperation compared with Black-on-White sexual assaults. It is possible that White victims in such cases perceive that they will be taken more seriously by law enforcement and are less likely to withhold cooperation. Another possibility that we explore in subsequent analyses is whether racial disparities arise in the context of select victim–offender relationships (Spohn & Spears, 1996; Walsh, 1987).

In general, the results in Table 3 suggest that victim and offender age and several indicators of aggravated sexual assault are the most consistently and strongly associated with the odds of a case being cleared by arrest versus through exceptional means rather than the victim–suspect race dyad. Cases where victims experience additional injuries and additional co-occurring offenses are more likely to be cleared by arrest. Consistent with insights from previous studies of sexual assault processing, incidents that involve the use of drugs as a weapon are less likely to end in arrest and more likely to end in victim non-cooperation than are cases that involve no weapon. Such a finding is not surprising as victims who have been drugged may be less likely to cooperate with the police because they are concerned about their perceived credibility. One of the strongest influences on clearance type is victim and suspect age. The odds of an arrest versus victim non-cooperation is estimated to increase by 138% when the incident involves a juvenile victim and adult offender. In a similar pattern, the results in Table 3 show that the odds of an arrest versus declined prosecution is expected to increase by 191% in such cases.

Notably, another one of the strongest predictors of clearance type is a quick clearance. The odds of an arrest is estimated to be 6 times larger than victim non-cooperation (b = 1.938) and 11 times larger (b = 2.45) than the odds of declined prosecution when police clear a case within 24 hrs of the incident. It remains unclear, however, whether this relationship captures the vigor with which the police investigate sexual assault cases or whether it is capturing time elapsed between an incident and when the police were notified by the victim or other third party, which may impact the availability of physical evidence as well as impact law enforcement’s perception of the victim’s credibility. Furthermore, the way in which some agencies record the incident date varies. A small number of NIBRS agencies record the date the incident became known to law enforcement rather than the incident date. In these cases, it would be unclear the amount of time that lapsed between the incident and the report.

Results from Table 3 also indicate that victim–offender relationship plays a significant role in the likelihood of arrest. Sexual assaults that involve family members are more likely to be cleared by arrest rather than victim non-cooperation when compared to incidents involving friends or acquaintances. Additionally, it is more likely that stranger assaults (vs. acquaintance assaults) will result in an arrest than in declined prosecution.

Next, we explore the possibility of a statistical interaction between racial composition and victim–offender relationship. We disaggregated the cleared sexual assault cases into four subsamples as defined by victim–offender relationship (e.g., stranger, intimate, acquaintance, and family) and estimated the same multinomial logistic regression models shown in Table 3. The results of these four additional models are presented in Table 4.²⁴ This analysis allows us to examine whether the salience of race emerges only within the context of certain victim–offender relationship categories. Panel 1 presents the racial composition effects for stranger assaults. Notably, and consistent with the arguments of the sexual stratification hypothesis, cases involving White suspects accused of assaulting Black women are significantly less likely to result in an arrest versus declined prosecution compared with Black-on-White assaults. Controlling for all other covariates, the significant parameter estimate for White-on-Black stranger assaults (b = −.366) indicates that the odds of arrest versus declined prosecution is predicted to decline approximately 30%, exp(b) = 0.694. Put another way, it is significantly more likely that assaults involving White male stranger attacks on Black females will result in declined prosecution and never proceed to an arrest or formal case filing. Among all stranger assaults examined in Panel 1 in Table 4, White-on-Black stranger assaults exhibited the lowest overall predicted probability (approximately 40%; not shown in Table) of arrest versus declined prosecution. Although NIBRS does not document the reasons for declined prosecution, this finding could suggest that prosecutors view allegations of assaults against Black women by White male strangers as less serious or credible.

Table 4.

Multinomial Regression of Race and Sex Dyads on Exceptional Clearance in Cases of Sexual Assault by Victim–Offender Relationship.

	(1) Stranger						(2) Intimate Partner
	Arrest Versus Victim Non-cooperation			Arrest Versus Prosecution Declined			Arrest Versus Victim Non-cooperation			Arrest Versus Prosecution Declined
Variables	b	SE	Exp(b)	b	SE	Exp(b)	b	SE	Exp(b)	b	SE	Exp(b)
Race dyads (Black-on-White)
Black-on-Black	−0.326	.171	0.722	−.175	.212	0.839	−0.242	.154	0.785	.076	.169	1.079
White-on-White	0.100	.122	1.105	−.129	.106	0.879	0.112	.111	1.119	.005	.107	1.005
White-on-Black	−0.207	.205	0.813	−.366*	.181	0.694	0.279	.252	1.322	−.007	.180	0.993
Other Racial Dyad	0.160	.224	1.174	.056	.214	1.058	0.108	.204	1.114	−.100	.195	0.905
Constant	0.860*	.336		.825*	.355		0.310	.253		.345	.300
N	7,997						13,751
	(3) Acquaintance						(4) Family
	Arrest Versus Victim Non-cooperation			Arrest Versus Prosecution Declined			Arrest Versus Victim Non-cooperation			Arrest Versus Prosecution Declined
Variables	b	SE	Exp(b)	b	SE	Exp(b)	b	SE	Exp(b)	b	SE	Exp(b)
Race dyads (Black-on-White)
Black-on-Black	0.041	.104	1.042	.266*	.116	1.305	−0.540*	.270	0.583	.207	.214	1.230
White-on-White	0.132*	.113	1.141	.134*	.066	1.143	−0.229	.281	0.795	.068	.184	1.070
White-on-Black	0.132	.000	1.141	.086	.098	1.090	0.120	.584	1.127	−.081	.337	0.922
Other Racial Dyad	0.016	.094	1.016	.047	.083	1.048	−0.412	.376	0.662	−.032	.241	0.969
Constant	−0.143	.183		−.107	.170		0.056	.475		.072	.323
N	45,560						11,744

Note. Coefficients for control variables are not shown. Robust standard errors (SE) adjusted for clustering within counties are reported. Reference group used to interpret coefficients appears in parentheses.

*p < .05.

We find no discernable relationship between victim and offender race in the clearance of intimate partner sexual assaults. However, in the results for acquaintance assaults shown in Panel 3, we find that intra-racial incidents (i.e., White-on-White and Black-on-Black assaults) are more likely to end in arrest versus exceptional clearance. In particular, an arrest is significantly more likely in cases where White women are assaulted by White male acquaintances versus Black male acquaintances; the odds of arrest versus any type of exceptional clearance increases by 14%, exp(b) = 1.141 and exp(b) = 1.143, when White females are assaulted by White male acquaintances versus Black male acquaintances. Although the reasons why remain unclear from administrative police data, the results suggest that White victims assaulted by Black acquaintances are significantly less likely to cooperate with police and less likely to have their cases prosecuted than White women assaulted by White male acquaintances even though the case is solved and a suspect identified by police.

The results shown in Panel 4 of Table 4 display the parameter estimates for models examining the effects of offender–victim racial dyad on clearance type for family sexual assaults. The results of this model indicate that cases involving Black victims assaulted by Black family members are significantly less likely to result in an arrest and that this is due to victim non-cooperation. The magnitude of the effect merits mention; the odds of an arrest (vs. victim non-cooperation) decline an estimated 42% in assaults involving Black family members compared to family assaults involving Black suspects and White victims. The overall probability of an arrest, which is not shown in Table 4, for these cases is 0.368, which corresponds to a 37% probability of arrest following case clearance. This specific type of sexual assault (i.e., assaults between Black family members) exhibited the single lowest probability of arrest due to victim non-cooperation than any other racial category examined in the panels shown in Table 4.²⁵ Overall, the final set of analyses summarized in Table 4 suggest that race does play a significant role in the clearance of sexual assault, but the magnitude and nature of that relationship hinge on the relationship between the victim and offender.

Conclusion

The social and legal context in which sexual assault cases are processed has changed dramatically in the past 30 years; but, the perception persists that race enters into legal decision-making as well as the response to crime on the part of police. The mere perception that race enters into police and prosecutor decision-making undermines criminal justice system legitimacy (Rocque, 2011) and has implications for victims’ decisions on whether to report their sexual assault (Spohn & Tellis, 2012a).

In the current study, we sought to replicate prior research testing the sexual stratification hypothesis using a national sample of sexual assault cases and to expand on prior studies by exploring more precisely the source of any observed racial disparities in the likelihood of arrest in sexual assault cases. The overall findings do not lend strong or consistent support to the sexual stratification hypothesis. Instead, the findings suggest that the relationship between the victim–suspect race dyad and clearance may be more nuanced. Although we do not find a relationship between race and overall odds of clearance, our findings suggest that when examining clearance type, and in particular the relationship between the victim and the suspect, that race is associated with whether solved cases ultimately result in arrest. Our initial regression findings suggest that race has no major impact on the probability that the police will successfully clear cases. However, among those cases that are cleared, an arrest is more likely in White intra-racial cases because White victims in these assaults are more likely to cooperate with investigators, thus increasing the likelihood of the case being cleared by arrest.

However, models that examined how race disparities may depend on the relationship between victims and offenders revealed a finding that is consistent with the sexual stratification hypothesis. In stranger sexual assaults, an arrest is significantly less likely when White males are suspected of assaulting Black females because prosecutors are more likely to decline these cases compared with cases where Black male strangers assault White females. Consistent with the sexual stratification hypothesis’s arguments, this suggests that prosecutors view the rape of Black women by White male strangers as less serious than the rape of White women by Black strangers. Another possibility is that prosecutors perceive White-on-Black stranger sexual assaults as less likely to result in conviction, or that jurors are less likely to view Black female victims of White stranger assault as credible or perhaps more blameworthy. Future research should explore the basis of such findings and examine whether institutional bias underlies the harsher treatment of Black-on-White stranger sexual assaults or whether other legally relevant factors not observed in the current study such as the strength of evidence or the presence of witnesses account for this finding. Moreover, future research is needed to reconcile these two seemingly contradictory patterns of disparity; that intra-racial assaults garner more punitive treatment by prosecutors within the context of acquaintance rape, while inter-racial incidents result in disparate outcomes when victims and suspects are strangers.

The second major finding to emerge from the analysis of the interaction between victim–offender relationship and race is that Black-on-Black family assaults are less likely to be cleared by arrest than they are to be cleared exceptionally due to victim non-cooperation. In other words, Black victims of sexual assault committed by Black family members are less likely to cooperate with police, which results in their cases being less likely to result in an arrest even though the suspect has been identified and sufficient probable cause exists to support an arrest. This suggests that victim’ decisions, rather than decisions by justice system personnel, may underlie racial disparities in arrest in certain contexts such as family sexual assault. Data from the National Crime Victimization Survey show Black women are significantly more likely to report their victimization to the police than White women, even controlling for the use of a weapon or additional injury to the victim (Bachman, 1998). In what is seemingly a paradox, African Americans are also more likely to distrust the police than any other racial group (Brunson, 2007; Tyler, 2005). Some of this mistrust is grounded in African Americans’ negative experiences with police (Brunson, 2007). These negative experiences undermine the legitimacy of the police in many communities and may leave Black victims reluctant to cooperate with the police, even if they initially reported the incident (Taylor et al., 2009). Black victims may withdraw their cooperation for another equally troubling reason. Recent research on the police use of exceptional clearance in Los Angeles revealed that in some cases law enforcement officers and prosecutors encourage some victims to drop their charges (Spohn & Tellis, 2012b). Although there is no evidence to suggest that police are more likely to make such suggestions to Black victims, these informal recommendations may foster a more nuanced and hidden form of discrimination in other jurisdictions that is not highlighted in quantitative studies of arrest and prosecution. More research is needed on whether Black female victims have less faith in the ability of police to apprehend their assailant, or if they perceive that the criminal justice system is more likely to discredit their allegations, which carries with it an additional stigma. Such behaviors on the part of legal officials, if they are indeed prevalent, could be an important and overlooked factor underlying racial disparities in arrest reported in prior research.

In conclusion, we do not find support for the sexual stratification hypothesis. This may be due to our inability to control for certain factors that prior research has found to significantly impact clearance. Specifically, a limitation of the NIBRS data is the lack of information regarding the strength of evidence, such as whether the victim submitted to a physical examination, when the sexual assault was reported to police, and whether the victim was intoxicated (which could impact victim believability factors). These types of data are available in individual prosecutor case files but are not reported in large-scale police data, as such we know little about how those types of issues impact police outcomes.

Although there was little support for the sexual stratification hypothesis, the findings did yield broader insights that should inform future research on police decision-making in sexual assault cases. The current study suggests that operationalizing the arrest process as a product of police discretionary authority is overly simplistic and precludes a broader understanding of the sources of case attrition from the system. Prior studies examining the relationship between race and arrest frame the disparity in the use of arrest as the result of selective enforcement on the part of the police. The current study suggests, however, that this is too narrow an interpretation of racial disparities in arrest. The results show that once cases are cleared, race affects the probability of arrest not because race strongly impacts whether cases are solved, but rather that race impacts whether an arrest occurs due to the actions of victims and prosecutors. Future research should explore the ways in which race influences these early and largely unregulated stages of case processing.

Footnotes

Declaration of Conflicting Interests

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

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

Notes

References

Addington

L. A.

Rennison

C. M.

(2008). Rape co-occurrence: Do additional crimes affect victim reporting and police clearance of rape? Journal of Quantitative Criminology, 24, 205–226.

Alderden

M. A.

Ullman

S. E.

(2012). Creating a more complete and current picture: Examining police and prosecutor decision-making when processing sexual assault cases. Violence Against Women, 18, 525–551.

Bachman

(1998). The factors related to rape reporting behavior and arrest: New evidence from the National Crime Victimization Survey. Criminal Justice and Behavior, 25, 8–29.

Banner

(2002). The death penalty: An American history. Cambridge, MA: Harvard University Press.

Beckett

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

Beckett

Nyrop

Pfingst

(2006). Race, drugs, and policing: Understanding disparities in drug delivery arrests. Criminology, 44, 105–137.

Beckett

Nyrop

Pfingst

Bowen

(2005). Drug use, drug possession arrests, and the question of race: Lessons from Seattle. Social Problems, 52, 419–441.

Black

(1980). The manners and customs of the police. New York, NY: Academic Press.

Bouffard

J. A.

(2000). Predicting type of sexual assault case closure from victim, suspect, and case characteristics. Journal of Criminal Justice, 28, 527–542.

10.

Brownmiller

(1975). Against our will: Men, women and rape. New York, NY: Simon & Schuester.

11.

Brunson

R. K.

(2007). “Police don’t like black people”: African-American young men’s accumulated police experiences. Criminology and Public Policy, 6, 71–101.

12.

Collins

(1971). A conflict theory of sexual stratification. Social Problems, 19, 3–21.

13.

Collins

(1975). Conflict sociology: Toward an explanatory science. New York, NY: Academic Press.

14.

Covington

(2010). Crime and racial constructions: Cultural misinformation about African Americans in Media and Academia. Plymouth, England: Lexington Books.

15.

Crutchfield

R. D.

Bridges

G. S.

Pitchford

S. R.

(1994). Analytical and aggregation biases in analyses of imprisonment: Reconciling discrepancies in studies of racial disparity. Journal of Research in Crime and Delinquency, 31, 166–182.

16.

DuMont

Miller

Myhr

T. L.

(2003). The role of ‘real rape’ and ‘real victim’ stereotypes in the police reporting practices of sexually assaulted women. Violence Against Women, 9, 466–486.

17.

Engel

R. S.

Calnon

J. M.

(2004). Examining the influence of drivers’ characteristics during traffic stops with police: Results from a national survey. Justice Quarterly, 21, 49–90.

18.

Estrich

(1987). Real rape. Cambridge, MA: Harvard University Press.

19.

Federal Bureau of Investigation. (2000). National incident-based reporting system Volume 1: Data collection guidelines. Washington, DC: U.S. Department of Justice.

20.

Federal Bureau of Investigation. (2013a). National incident-based reporting system, 2012. Washington, DC: U.S. Department of Justice.

21.

Federal Bureau of Investigation. (2013b). National incident-based reporting system user manual: Version 1.0. Washington, DC: U.S. Department of Justice.

22.

Griffin

Addie

Adams

Firestine

(2011). Trying juveniles as adults: An analysis of state transfer laws and reporting (Juvenile Offenders and Victims: National Report Series). Washington, DC: U.S. Department of Justice.

23.

Horney

Spohn

(1996). The influence of blame and believability factors on the processing of simple versus aggravated rape cases. Criminology, 34, 135–162.

24.

Howerton

(2006). Police response to crime: Differences in the application of law by race. Journal of Ethnicity in Criminal Justice, 4, 51–66.

25.

Kingsnorth

Lopez

Wentworth

Cummings

(1998). Adult sexual assault: The role of racial/ethnic composition in prosecution and sentencing. Journal of Criminal Justice, 26, 359–371.

26.

LaFree

G. D.

(1980). The effect of sexual stratification by race on official reactions to rape. American Sociological Review, 45, 842–854.

27.

Long

J. S.

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

28.

Lundman

R. J.

Kaufman

R. L.

(2003). Driving while black: Effects of race, ethnicity, and gender on citizen self-reports of traffic stops and police actions. Criminology, 41, 195–220.

29.

Messner

S. F.

McHugh

Felson

R. B.

(2004). Distinctive characteristics of assaults motivated by bias. Criminology, 42, 585–618.

30.

Myers

Talarico

(1987). The social contexts of criminal sentencing. New York, NY: Springer-Verlag.

31.

Noonan

J. H.

Vavra

M. C.

(2007). Crime in schools and colleges: A study of offenders and arrestees reported via National Incident-Based Reporting System data. Washington, DC: U.S. Department of Justice.

32.

Pattavina

Morabito

Williams

L. M.

(2015). Examining connections between the police and prosecution in sexual assault case processing: Does the use of exceptional clearance facilitate a downstream orientation? Victims & Offenders, 00, 1–20. doi:10.1080/15564886.2015.1046622

33.

Peterson

R. D.

Hagan

(1984). Changing conceptions of race: Towards an account of anomalous findings of sentencing research. American Sociological Review, 49, 56–70.

34.

Poe-Yamagata

Jones

M. A.

(2000). And justice for some: Differential treatment of minority youth in the justice system. Washington, DC: Youth Law Center.

35.

Roberts

(2015). Adjusting rates of homicide clearance by arrest for investigation difficulty: Modeling incident- and jurisdiction-level obstacles. Homicide Studies, 19, 273–300 doi:10.1177/ 1088767914536984

36.

Rocque

(2011). Racial disparities in the criminal justice system and perceptions of legitimacy: A theoretical linkage. Race and Justice: An International Journal, 1, 292–315.

37.

Rogers

(1993). Regression standard errors in clustered samples. Stata Technical Bulletin, 13, 19–23.

38.

Rojek

Rosenfeld

Decker

(2012). Policing race: The racial stratification of search in police traffic stops. Criminology, 50, 993–1024.

39.

Rose

V. M.

Randall

S. C.

(1982). The impact of investigator perceptions of victim legitimacy on the processing of rape/sexual assault cases. Symbolic Interaction, 5, 23–36.

40.

Schafer

Huebner

Bynum

(2003). Citizen perceptions of police services: Race, neighborhood context, and community policing. Police Quarterly, 6, 440–468.

41.

Smith

D. A.

Visher

C. A.

(1981). Street level justice: Situational determinants of police arrest decisions. Social Problems, 29, 167–177.

42.

Smith

D. A.

Visher

C. A.

Davidson

L. A.

(1984). Equity and discretionary justice: The influence of race on police arrest decisions. Journal of Criminal Law and Criminology, 75, 234–249.

43.

Snyder

H. N.

(2011). Arrest in the United States: 1980-2009 (Bureau of Justice Statistics Special Report). Washington, DC: U.S. Department of Justice, Office of Justice Programs.

44.

Spohn

(1994). Crime and the social control of blacks: The effect of suspect/victim race on sentences for violent felonies. In Bridges

Myers

(Eds.), Inequality, crime, and social control (pp. 249–268). Boulder, CO: Westview Press.

45.

Spohn

(2000). Thirty years of sentencing reform: The quest for a racially neutral sentencing process. National Institute of Justice: Criminal Justice 2000. Washington, DC: National Institute of Justice.

46.

Spohn

Holleran

(2001). Prosecuting sexual assault: A comparison of charging decisions in sexual assault cases involving strangers, acquaintances, and intimate partners. Justice Quarterly, 18, 651–688.

47.

Spohn

Spears

J. W.

(1996). The effect of suspect and victim characteristics on sexual assault case processing decisions. Justice Quarterly, 13, 649–679.

48.

Spohn

Tellis

(2011). Justice denied? The exceptional clearance of rape cases in Los Angeles. Albany Law Review, 74, 1379–1421.

49.

Spohn

Tellis

(2012a). Policing and prosecuting sexual assault in Los Angeles City and County (NIJ Publication NO. 237582). Washington, DC: U.S. Department of Justice.

50.

Spohn

Tellis

(2012b). The criminal justice system’s response to sexual violence. Violence Against Women, 18, 169–192.

51.

Stewart

Baumer

Brunson

Simons

(2009). Neighborhood racial context and perceptions of racial discrimination among black youth. Criminology, 47, 847–887.

52.

Stolzenberg

D’Alessio

S. J.

Eitle

(2004). A multilevel test of racial threat theory. Criminology, 42, 673–698.

53.

Stolzenberg

D’Alessio

S. J.

Eitle

(2013). Race and cumulative discrimination in the prosecution of criminal defendants. Race and Justice, 3, 275–299.

54.

Tasca

Rodriguez

Spohn

Koss

M. P.

(2012). Police decision making in sexual assault cases: Predictors of suspect identification and arrest. Journal of Interpersonal Violence, 28, 1157–1177.

55.

Taylor

T. J.

Holleran

Topalli

(2009). Racial bias in case processing: Does victim race affect police clearance of violent crime incidents. Justice Quarterly, 26, 562–591.

56.

Teigen

(2014, 10 1). Juvenile age of jurisdiction and transfer to adult court laws. National Conference of State Legislatures. Retrieved from http://www.ncsl.org

57.

Tellis

Spohn

(2008). The sexual stratification hypothesis revisited: Testing assumptions about simple and aggravated rape. Journal of Criminal Justice, 36, 252–261.

58.

Tyler

T. R.

(2005). Policing in black and white: Ethnic group differences in trust and confidence in the police. Police Quarterly, 8, 322–342.

59.

Walfield

S. M.

(2016). When a cleared rape is not cleared: A multilevel study of arrest and exceptional clearance. Journal of Interpersonal Violence, 31, 1767–1792.

60.

Walsh

(1987). The sexual stratification hypothesis and sexual assault in light of the changing conceptions of race. Criminology, 25, 153–174.

61.

Williams

R. L.

(2000). A note on robust variance estimation for cluster-correlated data. Biometrics, 56, 645–646.

62.

Wriggens

(1995). Rape, racism, and the law. In Searles

Berger

(Eds.), Rape and society (pp. 215–222). Boulder, CO: Westview.

63.

Zatz

(2000). The convergence of race, ethnicity, gender, and class on court decision making: Looking toward the 21st century. Criminal Justice: The National Institute of Justice Journal, 3, 503–552.