Understanding Case Outcomes for Male Victims of Forcible Sexual Assaults

Abstract

While rape and sexual violence have long been a widespread social problem, and one that has garnered significant attention, research that specifically examines the phenomenon of male victimization of sexual violence remains lacking. Addressing the gaps in the research, the current study uses 10 years of law enforcement data from the United States’ National Incident-based Reporting System (2007–2016) on sexual victimization of males 14 years of age or older. The study sought to assess the impact of victim, offender, and incident characteristics associated with the outcome of the case (i.e., open, arrest, and exceptional clearance due to the victim declining or the prosecution refusing to pursue the case) for 20,701 male victims who reported a forcible sex offense to law enforcement as well as agency variation for cleared crimes. Using multilevel multinomial logistic regression, exceptionally cleared cases are more likely to resemble open cases than those resulting in arrest with incident characteristics having a larger influence than victim and offender characteristics. Cases involving concomitant offenses, committed by a stranger, resulting in injury, increase the likelihood of arrest—all of which support the “real rape” hypothesis. Exceptionally cleared cases represent more than one-third of cleared cases and there is significant department variation in the usage of exceptional clearance, as a number of agencies are exceptionally clearing more than half of their cleared cases, artificially increasing their clearance rate. These results, in conjunction with research on female victims, suggests that the handling of sexual assault cases reported to law enforcement remains problematic.

Keywords

crime clearance male victims sex offenses National Incident-based Reporting System arrest exceptional clearance administrative closure

Introduction

Due to the low reporting rates of sexual assault, understanding the factors associated with cases that are cleared is of the utmost importance. For example, factors related to crime clearance may be within control of the police (Wellford & Cronin, 1999), and thusly clearance rates may be improved through appropriate resource allocation. Related, individual characteristics of those involved in sexual assaults may be related to the likelihood of clearance. This victim-devaluing perspective suggests investigative efforts may be less for victims of lower status or deemed less worthy of an official police response (Black, 1976). While there have been significant developments within the past several years, the research in this area is lacking in two important areas. The first is an overwhelming focus on the male offender-female victim paradigm, with most studies excluding male victims (Addington & Rennison, 2008; Alderden & Ullman, 2012; Pattavina et al., 2016; Stacey et al., 2017; Tasca et al., 2013; Walfield, 2016; c.f., Richards et al., 2019). The second is a methodological issue due to studies focusing only are cases that either result in arrest or remain open (e.g., Addington & Rennison, 2008). The failure to include cases that are administratively closed by law enforcement without arresting an offender, referred to as an exceptional clearance in the United States as well as agencies in Europe, the Caribbean, and elsewhere, has important ramifications (e.g., Liem et al., 2019; Maguire et al., 2010). These issues will be discussed in turn.

First, crime clearance research on sexual assault has focused nearly exclusively on offenses perpetrated by males against female victims. This is noteworthy as while it was previously believed that male sexual victimization is rare, recent nationally representative studies have found that male victimization is much more prevalent than previously believed (Basile, Chen, Black, & Saltzman, 2007; Black et al., 2011; Tjaden & Thoennes, 2006). Results from the Centers for Disease Control and Prevention’s 2010 National Intimate Partner and Sexual Violence Survey indicate nearly a quarter of males (23.6%) experience some form of sexual violence over the course of their lifetime. Forms of sexual violence include rape, being made to penetrate an individual, sexual coercion, and/or unwanted sexual contact (Black et al., 2011). Supporting this, a recent study of males enrolled at a university found more than half had experienced sexual violence since the age of 16 and, based on the most severe incident, one in five reported unwanted sexual contacts, with one in six reported being a victim of rape (Turchik, 2012). Yet despite this, research on the experiences of male victims remains lacking.

The second glaring omission among studies of clearance for sexual assault has been the focus of the decision to arrest which omits exceptionally cleared cases (e.g., Addington & Rennison, 2008; Roberts & Lyons, 2009; Tasca et al., 2013) or only contrasting cases resulting in arrest with exceptional clearance (e.g., Pattavina et al., 2016; Walfield, 2016). Aside from arresting an individual, law enforcement may exceptionally clear a case for reasons outside of their control. Although the FBI allows for five types of exceptional clearances, the vast majority are a result of either the victim refusing to cooperate with law enforcement after an offender has been identified or the prosecutor declining the case for reasons other than a lack of probable cause (Hirschel & Faggiani, 2012; Walfield, 2016). Examples of the latter include issues concerning consent, lack of evidence, and victim credibility, particularly when considering how juries will react to the case and considerations of trial sufficiency when assessing to bring a case to court considering limited resources, referred to as downstream orientation (Spohn & Tellis, 2011). The other three types, which are used far less often, are the death of the offender (e.g., suicide or justifiably killed by law enforcement or citizen), denial of extradition due to prosecuting the offender in another jurisdiction, or for minor cases involving juvenile offenders that are handled informally without taking them into custody (Federal Bureau of Investigation [FBI], 2004; Hirschel & Faggiani, 2012; Walfield, 2016).

Exceptionally cleared cases are important for two reasons. Despite identifying an offender, the individual is neither arrested nor prosecuted for the crime, thus any deterrent the criminal justice system may impose is negated and may even further embolden the offender due to the lack of an official response. As such, these cases more closely resemble incidents that remain open in that no offender is held accountable for their actions despite a verifiable crime occurring that could be prosecuted. Second, the FBI aggregates both arrests and exceptional clearances in their Crime in the United States annual reports, and delineation is not provided. In the first national study using data representing 238 agencies and nearly 20,000 incidents on cleared cases for female victims of forcible sex offenses, Walfield (2016) found that nearly half had been cleared exceptionally. Further, he found there were numerous differences between arrests and exceptional clearances, with incidents involving a single offense, no injuries, no weapons, and involving known offenders were more likely to be exceptionally cleared. Given that the FBI’s clearance rates are used by researchers to better understand case attrition throughout the criminal justice system, as well as used to assess police performance by the public and government officials, this calls into question the reliability of those studies (refer to Bayley, 1994; Gorby, 2013; Sparrow, 2015, for criticisms of clearance rates as performance indicators).

Utilizing multilevel multinomial logistic regression (MNLM), this study seeks to address these two limitations. We focus solely on male victims of forcible sexual offenses committed by either male or female offenders and assess how victim, offender, and incident variables affect four case outcomes: open (i.e., unsolved), arrest, cleared exceptionally due to the victim refusing, and cleared exceptionally due to the prosecution declining. To do so, we utilize data from the FBI’s National Incident-based Reporting System (NIBRS) which is a voluntary reporting system for crimes known to law enforcement that captures significant detail on each single crime incident, with 34 states certified to report data as of 2016, the most recent dataset (FBI, 2017).

Exceptional Clearance

As stated, there are situations that allow law enforcement to clear a case without arresting an offender; when this occurs, the incident has been closed as the result of an exceptional clearance (FBI, 2004). The FBI specifically identifies criteria by which exceptional clearances can be made by the responding law enforcement agency (LEA):

The LEA investigation must have clearly and definitively established the identity of at least one offender.

The LEA must have sufficient probable cause to support arresting, charging, and prosecuting the offender.

The LEA must know the exact location of the offender so they could make an arrest if circumstances did not prevent it.

There must be a reason outside the control of the LEA preventing arrest. (FBI, 2004, p. 70)

Further, law enforcement agencies are explicitly instructed to consider administratively closing an investigation distinct and separate from exceptional clearances. Thus, it may be assumed that the exceptional clearance of a criminal incident occurs in cases in which an offender has been identified, probable cause was found, and the offender could have been arrested—implicating the exceptional clearance was the result of a decision made by someone other than a law enforcement agent. When the Uniform Crime Report (UCR) was first developed, the designers had expected that this type of clearance would rarely be utilized as indicated by the name exceptional and as “for over 60 years, the FBI continues to label its tables containing clearance data as ‘cleared by arrest.’” (Feeney, 2000, p. 19), though recent research by Walfield (2016) found a large number of agencies had high rates of exceptional clearances relative to arrest for sexual assaults with significant variation in how exceptional clearances were utilized. Some agencies used this clearance type for the vast majority of cases whereas others rarely, if at all, utilized it. This has been supported by investigations into other city practices such as in Baltimore (Yeung et al., 2018), Cleveland (Atassi et al., 2011), and Washington DC (Human Rights Watch, 2013). More recent investigative reporting of 64 of the 100 largest agencies in the United States by ProPublica found that more incidents of rape were cleared by exceptional means than by a traditional arrest in approximately half of the agencies, with significant variation therein (Yeung et al., 2018). For instance, it was found that Baltimore County reported clearing 70% of its rape cases during the 2016 calendar year, yet internal records show that arrests were only made in about 30% of the cases (Yeung et al., 2018). Based on clearance rates alone, ProPublica’s findings would indicate an agency like Wichita is performing better than New Orleans as the former reported 69% of cases were cleared yet only 7% were arrested whereas the latter reported a clearance rate of 18%, of which nearly all (17%) were cleared by arrest (Yeung et al., 2018). Given that agencies over the last few decades have been caught purposefully manipulating data regarding forcible sex offenses, these findings are troubling and have the potential to erode the public’s trust and confidence in law enforcement (National Research Council, 2014).

This is problematic as when the FBI presents clearance rates, they aggregate the two types of clearances despite considerable evidence showing significant differences between arrests and exceptional clearances (Bouffard, 2000; Hirschel & Faggiani, 2012; Pattavina et al., 2016; Spohn & Tellis, 2012; Stacey et al., 2017; Walfield, 2016). This results in a highly misleading statistic since no breakdown is provided (Spohn & Tellis, 2012). In fact, as early as 1985, researchers recommended that these be reported separately, noting that the utilization of exceptional clearances could lead to potential misuse and ultimately an inflated and inaccurate clearance rate (Poggio et al., 1985) which is now supported by extant research (e.g., Human Rights Watch, 2013; Spohn & Tellis, 2012; Walfield, 2016; Yeung et al., 2018). Spohn and Tellis (2012, pp. 424–425) note that “this practice contributes to an organizational reluctance to address the misuse of the exceptional clearance by detectives because police leadership is aware that ultimately the FBI presents only one statistic to the public, which is misleading as to how cases are being solved.”

Clearing Sexual Assault

Despite there being a significant amount of research that has examined the characteristics of incidents of rape and sexual assault that are associated with an official arrest, the body of work examining those characteristics and associations with exceptional clearance is scant and focuses nearly exclusively on the male offender-female victim paradigm (c.f., Richards et al., 2019). Withstanding the relatively small body of work for female victims of sexual assault, research has found a variety of factors are associated with the likelihood an incident of rape or sexual assault will result in an exceptional clearance. These factors are found at the incident level, which are specific to the crime, and agency level, which are specific to the jurisdiction where the offense occurred and the responding LEA is located. For example, several characteristics of the incident and those involved are related to case outcomes. Research suggests the victim-offender relationship plays an important role, where incidents involving acquaintances are more likely to be exceptionally cleared (Spohn & Tellis, 2012; Walfield, 2016). This has been found in studies examining two specific forms of exceptional clearance—the result of victims declining to cooperate in the prosecution of the offender (Pattavina et al., 2016; Zeisel, 1982) and the result of the prosecution declining to pursue the case (Morabito et al., 2017; Richards et al., 2019).

Individual characteristics of those involved in the incidents have also been found to be associated with the likelihood of crime clearance (Bouffard, 2000; Pattavina et al., 2016; Richards et al., 2019; Stacey et al., 2017; Walfield, 2016). As the impetus to this study is the neglected topic of male victims of sexual assault, so too neglected is examination of the effects of the offender’s sex on the likelihood of exceptional clearance in these cases. In the Richards et al. (2019) study, the sex of the offender had no discernible effect on the likelihood of exceptional clearance for either outcome—due to the victim’s refusal to cooperate or the prosecutor declining the case.

The victim’s age has been found to be positively associated with the likelihood of exceptional clearance (Pattavina et al., 2016; Richards et al., 2019; Walfield, 2016). However, this finding is not consistent across all studies. For example, Bouffard (2000) did not find victim age to be associated with the likelihood of exceptional clearance when victim refusal and prosecutorial declination were disaggregated. Stacey et al. (2017) found juvenile victims to have the lowest likelihood of exceptional clearance, specifically when the offender was an adult. Similarly, the age of the offender has evinced some patterns of exceptional clearance. An increase in the age of an offender is associated with a decrease in the likelihood of an exceptional clearance which suggests cases involving older offenders paired with young victims are more likely to be taken seriously (Pattavina et al., 2016; Richards et al., 2019; Stacey et al., 2017; Walfield, 2016).

While there is a considerable amount of research that has examined the effects of race on crime clearance, there lacks consensus on what those effects are (Kochel et al., 2011; refer to Lytle, 2014), specifically the effects of race on exceptional clearance. Pattavina et al. (2016) found that victim race has no discernible effect in cases where the exceptional clearance was the result of victim refusal to cooperate, however it did in cases where the prosecutor declined to continue the case. Specifically, they found that victims identified as racial minorities are less likely to see their cases exceptionally cleared. This supports Walfield’s (2016) findings that White offenders and White victims are more likely to see an exceptional clearance. However, Bouffard (2000) and Richards et al. (2019) found no association between victim or offender race and the likelihood of exceptional clearance, and Stacey et al. (2017) found only the White-on-White dyad showed a significant difference (increase) in arrest probability.

Incident characteristics offer similar inconsistencies to individual characteristics with respect to directionality and their relationship with the likelihood of exceptional clearance. Incidents occurring at a private residence have been found by some to be less likely to result in exceptional clearance (Walfield, 2016), while others have found an increase in likelihood (Stacey et al., 2017), or no difference at all (Pattavina et al., 2016). Stacey et al. (2017), however, found that the location of the incident has different effects based on the type of exceptional clearance. For example, in cases where the victim refuses to cooperate with the prosecution, incidents occurring at a residence are more likely to result in exceptional clearance than those occurring at an indoor and public location, while being less likely than those occurring at a bar, restaurant, or hotel. When the prosecution declines the case, the same pattern emerges. In addition, incidents at a residence are more likely to be exceptionally cleared than those occurring at a school. At the macro-level, though NIBRS reporting is not yet generalizable (refer to McCormack et al., 2017), the region in which an incident of rape/sexual assault occurs has been linked to differential likelihoods of arrest and exceptional clearance. Walfield (2016) found that incidents of rape occurring in the South and West are the least likely to result in arrest, while those in the Northeast are significantly more likely than incidents in the South to see that case outcome. Additionally, the type of agency was important as well, with sheriff departments more likely to exceptionally clear cases than local agencies.

Interestingly, only Stacey et al. (2017) and Walfield (2016) found the seriousness of the offense as measured by weapon use to be associated with the likelihood of exceptional clearance. Other measures of offense seriousness have been found to be associated with exceptional clearance. For example, incidents in which an injury occurs are much less likely to be cleared by exceptional means (Pattavina et al., 2016; Richards et al., 2019; Stacey et al., 2017), and increasingly so with an increase in injury severity (Walfield, 2016). Similarly, forcible sex crimes that co-occur with other crimes such as burglary are less likely to result in exceptional clearance (Pattavina et al., 2016; Richards et al., 2019; Stacey et al., 2017; Walfield, 2016). In fact, these incident characteristics evince the largest effects in the respective studies. Compared to incidents without a concomitant offense, those with additional offenses are 2.6 (Stacey et al., 2017) to 3.7 (Walfield, 2016) times as likely to result in arrest. Pattavina et al. (2016) found crime co-occurrence is associated with a 71% to 73% decrease in the likelihood of exceptional clearance, while Richards et al. (2019) found that each additional offense in an incident is associated with a 63% to 69% reduction in the likelihood that it is exceptionally cleared.

Other incident characteristics have also been examined and, generally, have offered similar findings in their associations with the likelihood of exceptional clearance. Pattavina et al. (2016) and Richards et al. (2019) both found that there is no difference in this likelihood when the offender is suspected of being under the influence of drugs or alcohol and when s/he is not. This finding is evident in cases in which the victim refused to cooperate and when the prosecutor declined to pursue the case. However, when the incident qualifies as an attempted and not completed rape/sexual assault, research offers contrasting findings. While Walfield (2016) found incidents that were completed were far less likely to result in arrest (O.R. = .47), Stacey et al. (2017) found no effect.

Theoretical Framework

As the study and explanation of rape and sexual assault originated within a feminist framework, a focus has long been on the acts as based in gender and/or power dynamics. As a result, the male offender-female victim paradigm has come to dominate the discussion of, and subsequent research related to, rape and sexual violence. Relatedly, responsibility has historically been at least partially placed onto the victims and efforts into understanding why incidents of male sexual victimization may be treated differently have been absent or misguided at best (refer to Koss, 1993; c.f., Burt, 1980).

Specific to rape, researchers have argued that the criminal justice system differentiates between “real rape” and “simple rape” (Estrich, 1987).¹

This has also been referred to as “classic rape” (Williams, 1984) and “aggravated rape” (Brownmiller, 1975).

Estrich (1987) argued the criminal justice system and society are influenced by the extralegal characteristics of the incident as some cases are seen as more believable and harmful to the victim as a result of stereotypes about rape and victims of rape (Venema, 2016; Williams, 1984). These were cases in which the victim and offender were strangers, evidence of victim resistance or violence such as an injury or a weapon was present, multiple offenders perpetrated the crime, and co-occurring crimes transpired leading to the perception a legitimate rape took place (Du Mont et al., 2003). However, these types of rapes are rare, with the vast majority of incidents committed resembling “simple rape” (Black et al., 2011; Fisher et al., 2003; Kilpatrick, 2007; Tjaden & Thoennes, 2006; Wolitzky-Taylor et al., 2011), and thus are taken less seriously. Researchers have generally found cases that resemble “simple rape” were not only less likely to be reported, but were less likely to result in arrest in addition to increasing the likelihood they would be exceptionally cleared relative to arrest (Du Mont et al., 2003; Freeman et al., 1999; Lafree, 1981; Pattavina et al., 2016; Tasca et al., 2013; Tellis & Spohn, 2008; Walfield, 2016; Williams, 1984; c.f., Horney & Spohn, 1996).

However, not all researchers accept this explanation, as Addington and Rennison (2008) argue these findings could be a result of the solvability explanation, specifically, due to the importance and value of evidentiary factors. This is a result of aggravating features such as victim injury, weapon use, and the offense involving co-occurring crimes. Thus, the incident would be afforded more resources and investigatory time as there is more evidence to support the arrest and successful prosecution of offender. This is particularly important when law enforcement and prosecutorial resources are limited. Taking this into consideration, officers and prosecutors may then take a downstream orientation when deciding which cases to devote more time to and are likely to result in arrest, prosecution, and eventual adjudication by the judge and jury as well as consideration of the defense (Beichner & Spohn, 2005; Campbell et al., 2015; Frohmann, 1997; Lonsway & Archambault, 2012; Pattavina et al., 2016; Spohn & Holleran, 2001).

Current Study

Research Questions

The current examination begins once the police have found a crime. It is important to note at this point case attrition has already occurred for reported crimes if the investigation determines the report is baseless or fraudulent in which case law enforcement may “unfound” the case if they believe the incident occurred outside the jurisdiction, allegations were false, or allegations were baseless, and it is not actively investigated nor included in official statistics (Alderden & Ullman, 2012). To make a contribution to the literature, the present study seeks to address the gaps in prior research. To do this, the following research questions are addressed:

RQ1: To what extent do cases that remain open are closed due to exceptional clearance, and result in arrest different for male victims?

RQ2: How much variation is there across departments in their usage of arrest and exceptional clearance?

Methodology

Data.

To test these research questions, the law enforcement data in the study are derived from the NIBRS under the auspice of the FBI. NIBRS is the result of concentrated efforts at improving and expanding the summary data collected through the UCR, whereby some refer to it as “an extension of the UCR” (McCormack et al., 2017, p. 497). NIBRS collects extensive information supplied by law enforcement agencies in the United States for each verified crime, including victim and offender demographics as well as crime characteristics. NIBRS utilizes a gender-neutral definition and sexual assault was limited to the four forcible offenses: forcible rape, forcible sodomy, sexual assault with an object, and forcible fondling, and excluded statutory rape and incest. To ensure a sufficient number of cases were available for analysis, 2007 to 2016 NIBRS data were pooled which represent the 10 most recent years of available data. In accordance with prior research in this area, incidents were restricted to those that were reported to city or county law enforcement departments as these are the principal responders to calls for service (Morabito et al., 2017; Walfield, 2016).²

This excludes all university/college, state, special, and other agencies which resulted in the removal of 8.2% of the cases.

As a result of the differential response by the criminal justice system to pre-pubescent children and adolescents, incidents were limited to victims and offenders who were 14 years of age or older (Koss et al., 2007; Livingston et al., 2007). Incidents involving younger victims and offenders are typically viewed as child sexual abuse and seen as more serious, garnering a different and stronger response as the victims have not yet reached sexual maturity and lack the legal ability to consent. Due to the complexities involved with incidents involving multiple victims and/or multiple offenders, cases were restricted to one offender and one victim which ensures a better model fit. Approximately four out of five cases involved a single victim and a single offender, which matches the public health national estimates for male victims in the United States (Tjaden & Thoennes, 2006).³

The breakdown of the other categories is as follows: crimes involving one victim/multiple suspects: 4.9%, multiple victims/one suspect: 11.2%, and multiple victims/multiple suspects: 3.0%.

This resulted in a final sample size of 20,701 incidents involving a male victim 14 years of age or older from 2,945 agencies with a range of 1–388 incidents per agency (M = 7.0; SD = 18.4).

Outcome variable.

The dependent variable contrasts the case outcomes and is measured as a nominal variable using four categories: open, arrest, exceptional clearance due to the prosecution declining, and exceptional clearance due to victim refusal. Arrest was chosen as the reference category as this allows for a direct comparison between the two types of exceptional clearances and open cases in order to answer the research question posed and has been used in prior research (e.g., Richards et al., 2019). Exceptional clearances were limited to when the victim refuses to cooperate, or prosecution is declined for reasons other than a lack of probable cause. The three other types comprised approximately 3% of the exceptional clearances.

Independent variables.

The analysis examined three types of independent variables: victim, offender, and incident characteristics. Five victim and offender variables were included in the analysis: age, offender gender, victim-offender racial dyad, ethnicity, and victim-offender relationship. Age was measured similarly for victims and offenders and entered into the model in two forms: the first as a continuous variable which was standardized at its mean prior to entering it into the models, and the second as dummy variables to indicate if an individual was an adult. Offender gender indicates if the offender was female. Victim-offender racial dyad is measured as suggested by Black (1976): White victim/non-White offender (reference group), White victim/White offender, Black victim/non-White offender, Black victim/White offender. Non-White includes African-Americans, Native Americans, Asians, and other racial categories; due to low cell counts, these categories were collapsed (Roberts & Lyons, 2009; Stacey et al., 2017). Hispanic measures the ethnicity of the victim.⁴

NIBRS began collecting ethnicity data for the (suspected) offender in 2013, thus only the victim variable was utilized.

Victim-offender relationship was coded as romantic partner (e.g., current or former spouse, girlfriend, homosexual relationship), family member (e.g., sibling, stepparent), otherwise known (e.g., neighbor, employee), or stranger (Addington & Rennison, 2008; Walfield, 2016). Known offenders serve as the reference category.

The group of incident-based predictor variables was derived from prior research and six indicators of crime seriousness were included. Weapon was measured as a categorical variable indicating the presence of a weapon (e.g., firearm, knife, blunt object), personal weapon (e.g., hands, feet, asphyxiation), or no weapon which served as the reference category (Budd et al., 2017). While NIBRS utilizes eight categories for injuries, due to small cell counts, this was aggregated into a dichotomous variable that indicates the presence of any physical injury (e.g., broken bones, possible internal injury, severe laceration, unconsciousness) coded as a dummy variable. Concomitant measured if there was more than one offense (e.g., robbery and forcible sodomy) committed during the incident (Addington & Rennison, 2008; Sewall et al., 2013). Offender substance use was a dummy variable indicating if the offender was suspected of being under the influence of alcohol and/or drugs; this is not captured by NIBRS for the victim. While NIBRS records nearly 50 locations, this was recoded into four mutually exclusive categories: residence, indoors (e.g., shopping mall, public building, bar), outdoors (e.g., field, playground, rest area, construction site, road, parking lot), or service locations (e.g., hospital, school, religious institution, hotel; Williams & Bierie, 2015). Completed was measured as a dummy variable indicating if the sexual assault was attempted or completed.

Two agency-level variables were included. To control for any geographic variability, the region identifies the region the police department was located in. This follows the U.S. Census categories (South, Northeast, Midwest, and West), with the South serving as the reference category. Local law enforcement was coded as a dichotomous variable contrasting county (sheriff) departments with local law enforcement agencies.

Missing data.

The amount of missing data is minimal, and primarily limited to one or two variables for most cases. However, data were not missing at random. Cases that resulted in arrest were the least likely to have missing data whereas those that remained open had the most missing data. As such, discarding the data via listwise deletion would substantially bias the results. To account for this, multiple imputation was performed. Imputation is a common method in which a “complete” dataset is created by using the observed data, including auxiliary variables, to impute missing values, even in instances of data missing not at random (MNAR; Allison, 2001). To capture the uncertainty when imputing data, multiple datasets are created then pooled to obtain parameter estimates and standard errors in accordance to Rubin’s rule for multivariate analyses (Rubin, 1976). As Graham (2009, p. 55) argues, multiple imputation of data MNAR “are always at least as good as the old procedures (e.g., listwise deletion, except in artificial, unrealistic circumstances), and MI/ML [Multiple Imputation–Maximum Likelihood] methods are typically better than old methods, and often very much better.”

The R package “Amelia II” 1.7.4 was used to create 10 datasets with imputed values for all missing data (Honaker et al., 2011). Amelia II uses an expectation-maximization with bootstrapping (EMB) algorithm to produce the maximum likelihood and allows for logical bounds on the imputed data. A sensitivity analysis was performed in addition to various diagnostics to ensure the multiple imputation process produced sensible results (Allison, 2001). Listwise deletion was used for three variables: victim age, offender age, and offender gender. Only four variables were missing data: victim ethnicity (31.3%), victim-offender relationship (9.7%), racial dyad (6.7%), and weapon (4.0%).

Analytic procedure.

Descriptive statistics are presented using a random imputed dataset. Following this analysis, a series of regression models were estimated to predict the case outcome; results represent the pooled estimates from the 10 imputed datasets. Because the dependent variable is nominal and has four possible discrete outcomes, MNLM is utilized. As a result of the clustered nature of the data (i.e., reported incidents are “nested” within police departments), independence between the cases within police departments cannot be assumed and may lead to biased standard errors, increasing the likelihood of a Type I error (Snijders & Bosker, 2012). To account for this, multilevel modeling is used to estimate the present models. To ensure this approach was necessary, the intraclass coefficient (ICC) was first calculated which measures the proportion of total variance between agencies; coefficients were greater than .10 in both the null and full models.⁵

The following formula is used to compute intraclass correlation (ICC), also referred to as the variance partition coefficient: ρI= /(+ύ2/3) where is the department-level variance (Snijders & Bosker, 2012).

Second, F-tests were significant when comparing the multilevel models to the single-level models. Collectively, this indicates that the multilevel model produced a better fit (Snijders & Bosker, 2012). Variance inflation factor scores were calculated and confirmed multicollinearity was not an issue; no values exceeded 3. Due to the large sample size which increases the likelihood of a significant finding, only variables that are significant at p < .01 level are discussed. Data preparation and all analyses were performed in R 4.0.2 (R Development Core Team, 2017) and the models estimated using the R packages “lme4” 1.1–17 (Bates et al., 2015) and “mitools” (Lumley, 2014).

Results

Table 1.

Victim, Offender, and Incident Characteristics Post Imputation (n = 20,701).

Dependent Variable	n	%
Case outcome
Arrest†	4,392	21.2
Open	13,602	65.7
E.C.–Prosecutor	1,707	8.2
E.C.–Victim	1,000	4.8
Victim/offender
Victim age (M/SD)	23.8	12.8
Offender age (M/SD)	35.4	15.7
Adult victim	11,158	53.9
Adult offender	18,466	89.2
Female offender	3,935	19.0
Racial dyad
White victim-Non-White offender†	2,522	12.2
Non-White victim-White offender	1,306	6.3
White	14,133	68.3
Non-White	2,729	13.2
Hispanic victim	2,137	10.3
Victim-offender relationship
Known†	13,679	66.1
Stranger	2,226	10.8
Family	3,546	17.1
Intimate partner	1,250	6.0
Incident
Weapon present
None†	6,830	33.0
Strong arm	12,505	60.4
Weapon	1,366	6.6
Injury sustained	2,367	11.4
Concomitant crimes	502	2.4
Location
Residence†	13,802	66.7
Indoors	1,709	8.3
Outdoors	2,027	9.8
Service	3,163	15.3
Offender substance use	1,793	8.7
Completed	19,998	96.6
Macro
Local law enforcement	14,754	71.3
Region
South†	6,408	31.0
Midwest	7,705	37.2
Northeast	1,963	9.5
West	4,625	22.3

Note. n = 20,701, N = 2,945.

†Serves as reference category in the regression models.

E.C. = Exceptionally cleared.

Descriptives are presented using a random imputed dataset.Table 1 presents the descriptive statistics. Overwhelmingly, the majority (65.7%) of cases remained open with only one in five resulting in arrest. Exceptional clearances were more likely to occur due to the prosecutor refusing (8.2%) than the victim declining (4.8%); of the cleared cases collectively, nearly 40% (n = 2,707) were exceptionally cleared. Male victims were likely to be victimized by other males as only one in five cases involved a female offender. While most offenders were adults (89.2%), only half of the victims (53.9%) were adults at the time of the incident and victims (M = 23.8; SD = 12.8) on average were 12 years younger than their offender (M = 35.4). Most cases were intra-racial with White-on-White crimes the most frequent (68.3%). Only one in 10 cases involved a Hispanic victim. In cases where the victim and perpetrator were not strangers, 17% were family members and 6% were intimate partners, whereas the remainder knew each other in some other manner. The characteristics of the incidents closely resemble “simple rape:” few resulted in any injury (11.4%), were committed by a stranger (10.8%), weapons were rarely used (6.6%), and even less co-occur with other crimes (2.4%). The majority of incidents occurred at a residence (66.7%) followed by service locations (15.3%). The offender was suspected of substance abuse in only 8.7% of the incidents, and nearly all offenses were completed (96.6%).

Table 2.

Multilevel Multinomial Logistic Regression Analysis Predicting Case Outcomes (Reference Category = Arrest; n = 20,701).

	Open			Exceptionally Cleared–Prosecution			Exceptionally Cleared–Victim
	OR	b	SE	OR	b	SE	OR	b	SE
Victim age	1.09	.09	.03	1.22	.20	.05	–	.02	.06
Offender age	.89	–.12	.02	.84	–.18	.05	.78	–.25	.06
Adult victim	1.67	.51	.05	1.91	.65	.10	4.18	1.43	.13
Adult offender	.76	–.27	.08	.58	–.55	.14	.54	–.61	.17
Female offender	1.20	.19	.05	–	.01	.11	–	.11	.13
Racial dyad (ref = White victim–Non-White offender)
Non-White	–	.07	.09	–	.04	.15	–	.39	.18
Non-White victim–White offender	–	–.01	.10	–	.10	.18	–	–.10	.23
White	–	–.12	.07	–	.00	.12	–	–.06	.14
Hispanic victim	–	–.03	.08	–	–.10	.15	–	.10	.18
Victim-offender relationship (ref. = Known)
Stranger	.67	–.41	.07	.32	–1.12	.15	.71	–.34	.13
Family	1.50	.40	.06	1.65	.50	.12	–	.15	.16
Intimate partner	–	.07	.10	1.56	.45	.16	1.90	.64	.19
Weapon present (ref. = None)
Strong arm	–	–.07	.05	–	.11	.10	–	.10	.12
Weapon	–	.03	.09	–	.13	.18	–	–.01	.21
Injury	.84	–.18	.06	.62	–.48	.13	.71	–.34	.13
Concomitant crimes	.43	–.85	.11	.33	–1.10	.27	.43	–.85	.26
Location (ref. = Residence)
Indoors	.74	–.30	.08	.72	–.33	.14	.61	–.49	.18
Outdoors	.74	–.30	.07	.64	–.44	.14	.62	–.48	.16
Service	–	–.03	.07	1.31	.27	.12	–	–.19	.14
Offender substance use	.64	–.45	.07	.70	–.35	.14	–	.05	.15
Completed	.70	–.35	.12	–	–.41	.22	–	–.46	.24
Local law enforcement	–	.05	.07	–	.13	.14	–	–.01	.15
Region (ref = South)
Midwest	1.76	.57	.08	–	–.28	.15	.48	–.74	.17
Northeast	–	–.21	.10	.13	–2.04	.25	.16	–1.86	.27
West	1.30	.27	.09	–	–.27	.17	–	–.18	.19
Model diagnostics
Variance (ICC)	.18			.38			.38
AIC	18409.31			5970.40			4366.35
Log-likelihood	–9177.66			–2958.20			–2156.18
n	17,994			6,099			5,392

Note. Only odds ratios (OR) are presented for variables significant at the .01 level.

ICC = Intraclass correlation.

Results from the multivariate analyses can be found in table 2. Generally, results across the three models indicate that open cases, exceptionally cleared by the prosecution, and exceptionally cleared due to the victim refusing to cooperate had similar patterning relative to those resulting in arrest. Incidents that involved adult offenders and juvenile victims were more likely to result in arrest than in any other outcome. Cases involving female offenders were significantly more likely to remain open relative to ending in arrest, but the likelihood of either type of exceptional clearance did not differ from arrest. While the racial dyad and the victim ethnicity were not significant, the victim-offender relationship was. Across all three models, cases involving a stranger relative to a known offender were more likely to result in arrest, whereas the other two relationships had differing impacts. While cases involving family members relative to known offenders were more likely to stay open and result in prosecution declining, this did not reach significance for the victim refusing. Cases involving intimate partners were more likely to be exceptionally cleared whether by the prosecution or the victim; the likelihood of open cases did not differ from arrest.

Of the incident characteristics, the largest effect was concomitant crimes; across all three models, this substantially increased the odds of arrest. While the presence of a weapon was not a significant predictor of any case outcome, sustaining an injury increased the odds of arrest across all three models. The location was important as well and relatively consistent across all three comparisons. Relative to cases occurring in a residence, those that were either indoors or outdoors were more likely to result in arrest. Incidents that occurred at a service location were only significantly different than those at a residence to be cleared exceptionally due to the prosecutor.

The study found significant variation between departments in their handling of forcible sexual offenses. While the intraclass correlation was only .18 when comparing open with cases that resulted in arrest, this more than doubled to .38 for both types of exceptional clearances, indicating that 38% of the variation in these comparisons was between-department variation which suggests significant discrepancies in how agencies utilize this type of case outcome.

At the agency level, of the 1,767 agencies that had cleared cases, an average of 25.8% (SD = 36.3) were exceptional clearances (interquartile range [IQR] = 0–50%), suggesting a small proportion of agencies have high rates of exceptional clearances. Of the 202 agencies with at least 20 reported incidents, nearly 25% (n = 49) did not use exceptional clearances at all whereas four agencies recorded only exceptional clearances. Of these 202 agencies, which represented nearly half of all reported incidents, 36% of cases were cleared and arrests occurred slightly more often than exceptional clearances. Table 3 provides a sample of how the excess of exceptional clearances can inflate an agency’s clearance rate (agencies 1–4) while it may mask the effectiveness of other agencies that do not, or rarely, administratively close cases (agencies 5–8). For example, in the agency with the highest percentages of cases cleared (95%) only 5 out of the 14 cleared cases were by arrest.

Table 3.

Subset of Case Outcomes per Agency.

Agency	n (%)	1	2	3	4	5	6	7	8
# of incidents	20,701	321	245	70	22	203	71	63	47
Open	13,602(66)	99(31)	75(31)	18(26)	8(36)	110(54)	39(55)	16(26)	28(60)
Cleared	7,099(34)	222(69)	170(69)	52(74)	14(64)	93(46)	32(45)	47(73)	19(40)
Arrest	4,392(21)	72(22)	32(13)	23(33)	3(14)	85(42)	24(34)	45(71)	19(40)
Exceptional Clearance	2,707(13)	150(47)	138(56)	29 (41)	11(50)	8(4)	8(11)	2(3)	0 (0)
U.S. Census Region		West	Midwest	South	South	Midwest	West	West	Northeast
Agency Type		Local	Local	Sheriff’s	Sheriff’s	Local	Local	Local	Local

Note. Agencies have been de-identified.

Discussion

Results from the study indicate that the wide usage of exceptional clearances is an issue for a number of agencies across rape and sexual assault, regardless of the gender of the victim, supporting prior research (Bouffard, 2000; Pattavina et al., 2016; Richards et al., 2019; Spohn & Tellis, 2012; Stacey et al., 2017; Walfield, 2016). Furthermore, cases that were exceptionally cleared resembled open cases more than they did for arrest across a number of variables: victim age, offender age, victim-offender relationship, location, injury, and concomitant crimes, providing partial support for H₁. Importantly, these findings support prior research involving female victims (Pattavina et al., 2016; Richards et al., 2019; Spohn & Tellis, 2012; Walfield, 2016). This is somewhat troubling given the fundamental difference between these two general outcomes as the offender has been successfully identified by law enforcement in exceptionally cleared cases, assuming that the requirements were met, yet unlike in arrest the offender remains free and is not prosecuted.

Collectively, these variables offer some support to the “real rape” hypothesis (Estrich, 1987) as extralegal variables predicted exceptional clearance and arrest as well as the solvability perspective (Addington & Rennison, 2008) which provides evidence and physical proof that a crime had been committed. Additionally, the covariates related to both types of exceptional clearances were nearly identical, and there were two instances when variables were only significant when contrasting cases resulting in arrest with those that remain open: female offender and completed sex offenses.

The large intraclass correlations for the three regression models provide evidence for the significant disparity in how cases are cleared by law enforcement agencies across the United States, with both exceptional clearance models displaying an intraclass correlation more than double the open model (open model = .18; prosecutor declining model = .38; victim refusal model = .38). This finding offers support of prior studies conducted on male offenders-female victims. It indicates there is far greater variation across law enforcement agencies with regard to the rate of exceptional clearance compared to variation in the rate of arrest. For instance, one agency in Colorado reported that 69% of crimes were cleared yet, when limiting this to arrest, it only amounted to 22% of cleared cases. Just as Poggio et al. (1985) warned and Spohn and Tellis (2012) had reported, the use of exceptional clearance as an analog to arrest in crime clearance statistics results in an inaccurate representation of police practices and case outcomes.

While numerous studies have examined how specific police departments respond to sexual violence among female victims, such as when and under what circumstances the victim refuses to cooperate (e.g., Bouffard, 2000; Murphy, Edwards, Bennett, Bibeau, & Sichelstiel, 2014; Spohn & Tellis, 2012), there has been a lack of in-depth research utilizing police incident reports on male victims, their reporting and general experiences, and how these cases are processed within the criminal justice system (Du Mont et al., 2013). It is unknown if the findings relating to female victims apply to males as well. Furthermore, the utilization of case records would allow researchers to review exceptionally cleared cases to ensure that they meet the criteria required by the FBI given that Spohn and Tellis (2012) reported that both the city and county police agencies in Los Angeles were not following protocol. For example, an investigation launched by Fort Lauderdale’s major newspaper, the Sun Sentinel, examined exceptional clearances in Broward Sheriff’s Office in Florida. The investigation found extensive misuse of exceptional clearance for property crimes with “some districts ‘bank[ing]’ cases to boost their statistics in slow months and avoid drawing attention to improbably high clearances in a single month,” which resulted in officers facing criminal charges (O’Boye et al., 2004). This will help elucidate when and where case attrition occurs and under what conditions.

Finally, despite the wealth of data collected by NIBRS, it does suffer from some limitations. Importantly, of the violent crimes, sex crimes are the least likely to be reported which is even more problematic for male victims (Hart & Rennison, 2003; Pino & Meier, 1999; Tjaden & Thoennes, 2006). As a result, these cases may not be generalizable. As researchers have noted, there are issues with the coverage and representativeness of NIBRS data (McCormack et al., 2017). Because participation is voluntary, not all agencies submit data to NIBRS—approximately one-third of agencies report. Smaller agencies and departments in the South and Midwest are over-represented and large cities are not included (Roberts, 2009). For instance, in the current dataset, no city serving a population greater than one million people is included.

While NIBRS remains the best source of data for crime incident data generally, the inclusion of other datasets could be a fruitful line of research. This can be done at the agency, city, county, or state level. For instance, NIBRS can be linked to a multitude of other datasets to examine agency and community characteristics. This includes criminological datasets such as the Law Enforcement Management and Administrative Statistics (LEMAS) and the UCR as each LEA has its own unique Originating Agency Identifier (ORI). Other omnibus and social indicator datasets such as the American Community Survey (ACS) and the U.S. Census have been used in conjunction with NIBRS (e.g., Stolzenberg et al., 2006). For instance, LEMAS contains variables relating to the presence of a victim assistance unit, the percentage of female officers, education requirements, and other covariates that have been found to be associated with case outcomes (e.g., Roberts & Roberts, 2016; Walfield, 2016). As it was necessary to pool nearly a decade’s worth of data to ensure a sufficient enough sample size in the current study, only static covariates at the police department level could be included in the present study (i.e., region and type of agency). Pattavina et al. (2016) found dynamic agency characteristics to be significant predictors of exceptionally cleared cases for female victims of rape. Specifically, they found at the macro-level that agencies in cities with higher crime rates increased the likelihood of the victim refusing to cooperate whereas a decrease in the rate of prosecution was associated with prosecution declining to pursue the case due to workload issues. Results regarding the percentage of female officers have been mixed, with some studies showing a greater percentage leads to an increase in arrests (Meier & Nicholson-Crotty, 2006; Walfield, 2016) while others, such as Pattavina et al. (2016), report no effect. Furthermore, studies should include both male and female victims with separate multivariate analyses based on the sex of the victim.

Additionally, some covariates that have been found to be important for an arrest are not collected by NIBRS (Black et al., 2011; Bouffard, 2000; O’Neal et al., 2016). For instance, NIBRS does not include whether the victim was using either alcohol or drugs at the time of the incident (O’Neal et al., 2016), the length of time between when the crime occurred and when it was reported (Tasca et al., 2013), the collection of forensic and/or physical evidence (Bouffard, 2000; O’Neal et al., 2016), or if a stranger had been successfully identified. NIBRS does not include cases in which male victims were forced to penetrate other individuals, an important limitation given the high prevalence of this for male victims as, according to the CDC, nearly 1.6 million males had been forced to do this (Black et al., 2011). Finally, these incidents only represent reported cases to law enforcement as prior research using NCVS data found differences between reported crimes and non-reported crimes (Pino & Meier, 1999).

The data limitations notwithstanding, these findings suggest consideration be given to the use of exceptional clearance as a common practice by which sexual assault cases are cleared and to the extralegal factors associated with sex assaults involving male victims remaining open. First, the extensive use of exceptional clearance by some departments should be examined for accuracy and appropriateness of its use (Richards, et al., 2019; Spohn & Tellis, 2011). This can be done at the federal level, as the FBI is the clearinghouse for official statistics (Walfield, 2016), the local level by the prosecutor’s office responsible for the exceptional clearances, or through an alternative means of external review (Colon et al., 2018). Patterns that emerge at the local level may better inform the jurisdiction of the reasons for prosecutorial declination and victim refusal. Identifying these correlates may ensure equal attention is afforded to all cases and their processing, and all offender and victim subgroups experience the criminal justice process similarly. Second, the identification of two specific associations in the present study may warrant attention by police. Male victims of sexual assault were more likely to see their cases remain open, relative to arrest, when the perpetrator was female compared to when the perpetrator was male. As Walfield (2018) found, one of the most adhered to rape myths is believing a male cannot be raped by a female. Likewise, in a study of law enforcement perceptions of victim credibility, male victims were the second most likely to not be believed, only ahead of prostitutes (Page, 2010). Further, attempted assaults were also more likely to remain open than result in arrest. Departments should ensure their training and procedures reflect the FBI-compliant definition of rape and their agents respond similarly when the incidents do not conform to the “real rape” stereotype.

Conclusion

In accordance with prior studies on female victims, exceptional clearances are used widely, albeit somewhat inconsistently, when clearing forcible sexual offenses. Findings reveal that the cases that do remain open were similar to those that were exceptionally cleared whether it is due to the victim’s refusal to cooperate or a result of prosecutorial discretion to decline the case, providing partial support that the cases that do not resemble “real rape” were taken less seriously. As Richards et al. (2019, p. 21) note, these findings signal “that stereotypes regarding genuine and worthy victims continue to impact both system actors’ and victims.” Furthermore, to provide a more accurate portrayal of clearance, both the FBI and agencies should report arrest and exceptional clearances separately to not only hold agencies accountable but to ensure they are transparent and not abusing this type of case closure to influence their clearance rates.

Footnotes

Acknowledgments

The authors take responsibility for the integrity of the data, the accuracy of the data analyses, and have made every effort to avoid inflating statistically significant results. Thank you to the reviewers and editor for their helpful comments.

Authors’ Note

An earlier version of this article was presented at the 2018 American Society of Criminology annual meeting in Atlanta, GA.

Declaration of Conflicting Interests

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

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Scott M. Walfield

Author Biographies

Scott M. Walfield, PhD, is an assistant professor in the Department of Criminal Justice at East Carolina University. His primary research interests include crime clearance, victim reporting, rape/sexual assault, and sex offender management and policy. He has been published in various venues including Sexual Abuse: A Journal of Research and Treatment, Psychology, Public Policy, and Law, and Justice Research and Policy, among others. He was the winner of the 2015 Association of Criminal Justice Sciences (ACJS) Michael C. Braswell/Anderson Publishing Outstanding Student Paper Award.

Philip D. McCormack, PhD, is an assistant professor in the Criminal Justice Department at Saint Anselm College. His research interests include quantitative methods, crime clearance, intimate partner violence, and examining racial disparity across a variety of social outcomes. His work has been published in such outlets as Crime & Delinquency, Race and Justice, and the Journal of Interpersonal Violence. He currently serves as an independent evaluator on a BJA-funded study examining the efficacy of programming on reentry.

Kaitlyn Clarke, PhD, is an assistant professor in the Criminal Justice Department at Saint Anselm College. Her research interests include wrongful convictions, qualitative methods, and class disparities across a variety of social outcomes. Much of her work examines the consequences of wrongful conviction on the reentry/reintegration process. Her most recent work has been published in Sociological Perspectives.

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(2), 205–226. https://doi.org/10.1007/s10940-008-9043-9

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(5), 525–551. https://doi.org/10.1177/1077801212453867

Allison

P. D.

(2001). Missing data . Sage.

Atassi

, Dissell

, & Baird

(2011, January 23). Cleveland police aren’t following departmental rules in clearing rape cases, analysis shows. Plain Dealer News . http://blog.cleveland.com/metro/2011/01/cleveland_police_arent_followi.html

Basile

K. C.

, Chen

, Black

M. C.

, & Saltzman

L. E.

(2007). Prevalence and characteristics of sexual violence victimization among U.S. adults, 2001 -2003. Violence and Victims , 22(4), 437–448. https://doi.org/10.1891/088667007781553955

Bates

D. M.

, Mächler

, & Bolker

(2015). Fitting linear mixed-effects models using lme4. Journal of Statistical Software , 67(1), 1–48. https://doi.org/10.18637/jss.v067.i01

Bayley

D. H.

(1994). Police for the future . Oxford University Press.

Beichner

, & Spohn

(2005). Prosecutorial charging decisions in sexual assault cases: Examining the impact of a specialized prosecution unit. Criminal Justice Policy Review , 16(4), 461–498. https://doi.org/10.1177/0887403405277195

Black

(1976). The Behavior of Law . Academic Press.

10.

Black

M. C.

, Basile

K. C.

, Breiding

M. J.

, Smith

S. G.

, Walters

M. L.

, Merrick

M. T.

, Chen

, & Stevens

M. R.

(2011). The National Intimate Partner and Sexual Violence Survey (NISVS): 2010 Summary Report . Centers for Disease Control and Prevention.

11.

Bouffard

J. A.

(2000). Predicting type of sexual assault case closure from victim, offender, and case characteristics. Journal of Criminal Justice , 28(6), 527–542. https://doi.org/10.1016/S0047-2352(00)00068-4

12.

Brownmiller

(1975). Against our will: Men, women, and rape . Simon & Schuster.

13.

Budd

K. M.

, Bierie

D. M.

, & Williams

(2017). Deconstructing incidents of female perpetrated sex crimes. Sexual Abuse: A Journal of Research and Treatment , 29(3), 267–290. https://doi.org/10.1177/1079063215594376

14.

Burt

M. R.

(1980). Cultural myths and supports for rape. Journal of Personality and Social Psychology , 38(2), 217–230. https://doi.org/10.1037/0022-3514.38.2.217

15.

Campbell

B. A.

, Menaker

T. A.

, & King

W. R.

(2015). The determination of victim credibility by adult and juvenile sexual assault investigators. Journal of Criminal Justice , 43, 29–39. https://doi.org/10.1016/j.jcrimjus.2014.12.001

16.

Colon

K. M.

, Kavanaugh

P. R.

, Hummer

, & Ahlin

E. M.

(2018). The impact of race and extra-legal factors in charging defendants with serious sexual assault: Findings from a five-year study of one Pennsylvania Court Jurisdiction. Journal of Ethnicity in Criminal Justice , 16(2), 99–116.

17.

Du Mont

, Miller

K.-L.

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(4), 466–486. https://doi.org/10.1177/1077801202250960

18.

Estrich

(1987). Real rape: How the legal system victimizes women who say no . Harvard University Press.

19.

Federal Bureau of Investigation (FBI). (2004). Uniform Crime Reporting Handbook . U.S. Department of Justice, Federal Bureau of Investigation.

20.

Federal Bureau of Investigation (FBI). (2017). Summary of NIBRS 2016 . U.S. Department of Justice, Federal Bureau of Investigation. https://ucr.fbi.gov/nibrs/2016/tables/data-declarations/dd_participation_by-state_2016.pdf

21.

Feeney

(2000). Police clearances: A poor way to measure the impact of Miranda on the police. Rutgers Law Journal , 32, 1–114.

22.

Fisher

B. S.

, Daigle

L. E.

, Cullen

F. T.

, & Turner

M. G.

(2003). Reporting sexual victimization to the police and others: Results from a national-level study of college women. Criminal Justice and Behavior , 30(1), 6–38. https://doi.org/10.1177/0093854802239161

23.

Freeman

R. C.

, Williams

M. L.

, & Saunders

L. A.

(1999). Drug use, AIDS knowledge, and HIV risk behaviors of Cuban-, Mexican-, and Puerto-Rican-born drug injectors who are recent entrants into the United States. Substance Use & Misuse , 34(13), 1765–93.

24.

Frohmann

(1997). Convictability and discordant locales: Reproducing race, class, and gender ideologies in prosecutorial decisionmaking. Law and Society Review , 31(3), 531–556.

25.

Gorby

D. M.

(2013). The failure of traditional measures of police performance and the rise of broader measures of police performance. Policing: A Journal of Policy and Practice , 7(4), 392–400.

26.

Graham

J. W.

(2009). Missing data analysis: Making it work in the real world. Annual Review of Psychology , 60(1), 549–576. https://doi.org/10.1146/annurev.psych.58.110405.085530

27.

Hart

T. C.

, & Rennison

C. M.

(2003). Reporting crime to the police: 1992–2001 (NCJ-1957100). Bureau of Justice Statistics, U.S. Department of Justice.

28.

Hirschel

, & Faggiani

(2012). When an arrest is not an arrest: Exceptionally clearing cases of intimate partner violence. Police Quarterly , 15(4), 358–385. https://doi.org/10.1177/1098611112452152

29.

Honaker

, King

, & Blackwell

(2011). Amelia II: A program for missing data. Journal of Statistical Software , 45(7), 1–47.

30.

Horney

, & Spohn

(1996). The influence of blame and believability factors on the processing of simple versus aggravated rape cases. Criminology , 34(2), 135–162. https://doi.org/10.1111/j.1745-9125.1996.tb01200.x

31.

Human Rights Watch. (2013). Capitol offense: Police mishandling of sexual assault cases in the district of Columbia . Human Rights Watch.

32.

Kilpatrick

D. G.

(2007). Drug-facilitated, incapacitated, and forcible rape: A national study . https://www.ncjrs.gov/pdffiles1/nij/grants/219181.pdf

33.

Kochel

T. R.

, Wilson

D. B.

, & Mastrofski

S. D.

(2011). Effect of Offender Race on Officer’s Arrest Decisions. Criminology , 49(2), 473–512. https://doi.org/10.1111/j.1745-9125.2011.00230.x

34.

Koss

M. P.

(1993). Detecting the scope of rape: A review of prevalence research methods. Journal of Interpersonal Violence , 8, 198–222. https://doi.org/10.1177/088626093008002004

35.

Koss

M. P.

, Abbey

, Campbell

, Cook

, Norris

, Testa

, Ullman

, West

, & White

(2007). Revising the SES: A collaborative process to improve assessment of sexual aggression and victimization. Psychology of Women Quarterly , 31(4), 357–370. https://doi.org/10.1111/j.1471-6402.2007.00385.x

36.

Lafree

G. D.

(1981). Official Reactions to Social Problems: Police Decisions in Sexual Assault Cases. Social Problems , 28(5), 582–594. https://doi.org/10.2307/800232

37.

Liem

, Suonpää

, Lehti

, Kivivuori

, Granath

, Walser

, & Killias

(2019). Homicide clearance in western Europe. European Journal of Criminology , 16(1), 81–101. https://doi.org/10.1177/1477370818764840

38.

Livingston

J. A.

, Hequembourg

, Testa

, & VanZile-Tamsen

(2007). Unique aspects of adolescent sexual victimization experiences. Psychology of Women Quarterly , 31(4), 331–343. https://doi.org/10.1111/j.1471-6402.2007.00383.x

39.

Lonsway

K. A.

, & Archambault

(2012). The “justice gap” for sexual assault cases: Future directions for research and reform. Violence against Women , 18(2), 145–68. https://doi.org/10.1177/1077801212440017

40.

Lumley

(2014). Mitools: Tools for multiple imputation of missing data. R package version 2.3 . https://CRAN.R-project.org/package=mitools

41.

Lytle

D. J.

(2014). The effects of offender characteristics on arrest: A meta-analysis. Journal of Criminal Justice , 42(6), 589–597. https://doi.org/10.1016/j.jcrimjus.2014.10.001

42.

Maguire

E. R.

, King

W. R.

, Johnson

, & Katz

C. M.

(2010). Why homicide clearance rates decrease: Evidence from the Caribbean. Policing and Society , 20(4), 373–400. https://doi.org/10.1080/10439463.2010.507869

43.

McCormack

P. D.

, Pattavina

, & Tracy

P. E.

(2017). Assessing the coverage and representativeness of the National Incident-based Reporting System. Crime & Delinquency , 63(4), 493–516. https://doi.org/10.1177/0011128717694595

44.

Meier

K. J.

, & Nicholson-Crotty

(2006). Gender, representative bureaucracy, and law enforcement: The case of sexual assault. Public Administration Review , 66(6), 850–860. https://doi.org/10.1111/j.1540-6210.2006.00653.x

45.

Murphy

S. B.

, Edwards

K. M.

, Bennett

, Bibeau

S. J.

, & Sichelstiel

(2014). Police reporting practices for sexual assault cases in which “the victim does not wish to pursue charges”. Journal of Interpersonal Violence , 29(1), 144–156. https://doi.org/10.1177/0886260513504648

46.

Morabito

M. S.

, Pattavina

, & Williams

L. M.

(2017). Active representation and police response to sexual assault complaints. Journal of Crime and Justice , 40(1), 20–33. https://doi.org/10.1080/0735648X.2016.1216730

47.

National Research Council. (2014). Estimating the incidence of rape and sexual assault . The National Academies Press. https://doi.org/10.17226/18605.

48.

O’Boye

, Maines

, & Hernandez

(2004, February 4). Broward Sheriff’s high success rate questioned. Sun Sentinel . https://www.sun-sentinel.com/news/fl-xpm-2004-02-28-0402280040-story.html

49.

O’Neal

E. N.

, Beckman

L. O.

, & Spohn

(2016). The sexual stratification hypothesis: Is the decision to arrest influenced by the victim/offender racial/ethnic dyad. Journal of Interpersonal Violence , 34(6), 1287–1310. https://doi.org/10.1177/0886260516651093

50.

Page

(2010). Police officers and rape myth acceptance. Feminist Criminology . 5(4), 315 –334. https://doi.org/10.1177/1557085110384108

51.

Pattavina

, Morabito

M. S.

, & Williams

L. M.

(2016). Examining connections between the police and prosecution in sexual assault case processing: Does the use of exceptional clearance facilitate a downstream orientation? Victims and Offenders , 11(2), 315–334. https://doi.org/10.1080/15564886.2015.1046622

52.

Pino

N. W.

, & Meier

R. F.

(1999). Gender differences in rape reporting. Sex Roles . 40(11–12), 979–990. https://doi.org/10.1023/A:1018837524712

53.

Poggio

E. C.

, Kennedy

S. D.

, Chaiken

J. M.

, & Carlson

K. E.

(1985). Blueprint for the Future of the Uniform Crime Reporting Program . https://www.ncjrs.gov/pdffiles1/bjs/98348.pdf

54.

R Development Core Team. (2017). R: A language and environment for statistical computing . R Foundation for Statistical Computing. https://www.R-project.org/

55.

Richards

T. N.

, Tillyer

M. S.

, & Wright

E. M.

(2019). When victims refuse and prosecutors decline: Examining exceptional clearance in sexual assault cases. Crime & Delinquency , 65, 474–498. https://doi.org/10.1177/0011128719828351

56.

Roberts

(2009). Contributions of the National Incident-based Reporting System to substantive knowledge in criminology: A review of research. Sociology Compass , 3(3), 433–458. https://doi.org/10.1111/j.1751-9020.2009.00209.x

57.

Roberts

, & Lyons

C. J.

(2009). Victim-offender racial dyads and clearance of lethal and nonlethal assault. Journal of Research in Crime and Delinquency , 46(3), 301–326. https://doi.org/10.1177/0022427809335168

58.

Roberts

, & Roberts

J. M.

(2016). Crime clearance and temporal variation in police investigative workload: Evidence from National Incident-based Reporting System (NIBRS) Data. Journal of Quantitative Criminology , 32, 651–674.

59.

Rubin

D. B.

(1976). Inference and missing data. Biometrika , 63(3), 581–592. https://doi.org/10.2307/2335739

60.

Sewall

L. A.

, Krupp

D. B.

, & Lalumière

M. L.

(2013). A test of two typologies of sexual homicide. Sexual Abuse: Journal of Research and Treatment , 25(1), 82–100. https://doi.org/10.1177/1079063212452617

61.

Snijders

T. A.

, & Bosker

R. J.

(2012). Multilevel analysis (2nd ed.). Sage.

62.

Sparrow

M. K.

(2015). Measuring performance in a modern police organization. Psycholigcal Issues in Human Resource Management , 3(2), 17–52.

63.

Spohn

, & Holleran

(2001). Prosecuting sexual assault: A comparison of charging decisions in sexual assault cases involving strangers, acquaintances, and intimate partners. Justice Quarterly , 18(3), 651–688. https://doi.org/10.1080/07418820100095051

64.

Spohn

, & Tellis

(2012). The criminal justice system’s response to sexual violence. Violence Against Women , 18(2), 169–192. https://doi.org/10.1177/1077801212440020

65.

Stacey

, Martin

K. H.

, & Brick

B. T.

(2017). Victim and offender race and the police clearance of sexual assault. Race and Justice , 7(3), 226–255. https://doi.org/10.1177/2153368716643137

66.

Stolzenberg

, Eitle

, & D’Alessio

S. J.

(2006). Race, economic inequality, and violent crime. Journal of Criminal Justice , 34(3), 303–316.

67.

Tasca

, Rodriguez

, Spohn

, & Koss

M. P.

(2013). Police decision making in sexual assault cases: Predictors of offender identification and arrest. Journal of Interpersonal Violence , 28(6), 1157–1177. https://doi.org/10.1177/0886260512468233

68.

Tellis

, & Spohn

(2008). The sexual stratification hypothesis revisited: Testing assumptions about simple versus aggravated rape. Journal of Criminal Justice , 36(3), 252–261. https://doi.org/10.1016/j.jcrimjus.2008.04.006

69.

Tjaden

, & Thoennes

(2006). Extent, nature, and consequences of rape victimization: Findings from the National Violence Against Women Survey .

70.

Turchik

J. A.

(2012). Sexual victimization among male college students: Assault severity, sexual functioning, and health risk behaviors. Psychology of Men & Masculinity , 13(3), 243–255. https://doi.org/10.1037/a0024605

71.

Venema

R. M.

(2016). Police officer schema of sexual assault reports: Real rape, ambiguous cases, and false reports. Journal of Interpersonal Violence , 31(5), 872–899. https://doi.org/10.1177/0886260514556765

72.

Walfield

S. M.

(2016). When a cleared rape is not cleared: A multilevel study of arrest and exceptional clearance. Journal of Interpersonal Violence , 31(9), 1767–1792. https://doi.org/10.1177/0886260515569062

73.

Walfield

S. M.

(2018). “Men cannot be raped:” Correlates of male rape myth acceptance. Journal of Interpersonal Violence . https://doi.org/10.1177/0886260518817777

74.

Wellford

, & Cronin

(1999). An analysis of variables affecting the clearance of homicides: A multistate study . Justice Research and Statistics Association.

75.

Williams

K. S.

, & Bierie

D. M.

(2015). An incident-based comparison of female and male sexual offenders. Sexual Abuse: A Journal of Research and Treatment , 27(3), 235–257. https://doi.org/10.1177/1079063214544333

76.

Williams

L. S.

(1984). The classic rape: When do victims report? Social Problems , 31(4), 459–467.

77.

Wolitzky-Taylor

, Resnick

H. S.

, McCauley

, Amstadter

A. B.

, Kilpatrick

D. G.

, & Ruggiero

K. J.

(2011). Is reporting of rape on the rise? A comparison of women with reported versus unreported rape experiences in the National Women’s Study-Replication. Journal of Interpersonal Violence , 26(4), 807–32. https://doi.org/10.1177/0886260510365869

78.

Yeung

, Greenblatt

, Fahey

, & Harris

(2018, November 15). When it comes to rape, just because a case is cleared doesn’t mean it’s solved. ProPublica . https://www.propublica.org/article/when-it-comes-to-rape-just-because-a-case-is-cleared-does-not-mean-solved

79.

Zeisel

(1982). The limits of law enforcement. Vanderbilt Law Review , 35, 527–540.