Why Police “Couldn't or Wouldn't” Submit Sexual Assault Kits for Forensic DNA Testing: A Focal Concerns Theory Analysis of Untested Rape Kits

Abstract

In jurisdictions throughout United States, thousands of sexual assault kits (SAKs) (also termed “rape kits”) have not been submitted by the police for forensic DNA testing. DNA evidence may be helpful to sexual assault investigations and prosecutions by identifying offenders, revealing serial offenders through DNA matches across cases, and exonerating those who have been wrongly accused, so it is important to understand why police are not utilizing this evidence. In this study, we applied focal concerns theory to understand discretionary practices in rape kit testing. We conducted a three-year ethnography in one city that had large numbers of untested SAKs—Detroit, Michigan—to understand why thousands of SAKs collected between 1980 and 2009 were never submitted by the police for forensic DNA testing. Drawing upon observational, interview, and archival data, we found that while practical concerns regarding resources available for forensic analysis were clearly a factor, as Detroit did not have the funding or staffing to test all SAKs and investigate all reported rapes, focal concerns regarding victim credibility and victim cooperation were more influential in explaining why rape kits were not tested. Implications for the criminal justice system response to sexual assault and rape kit testing legislation are examined.

In jurisdictions throughout the United States, police frequently do not submit sexual assault kits (SAKs) for forensic DNA testing, and, instead, kits are shelved in police property, unprocessed, and ignored for years (Campbell et al. 2017a, 2017b; Pinchevsky forthcoming). A SAK (also termed a “rape kit”) is typically collected within 24–72 hours after a sexual assault in order to obtain biological evidence from victims’ bodies (e.g., semen, blood, saliva; Department of Justice 2013). This evidence can be analyzed for DNA and compared against other criminal reference DNA samples in Combined DNA Index System (CODIS), the federal DNA database, which can be which can be instrumental in solving crimes and prosecuting rapists (Campbell et al. 2017a, 2017b; Murphy et al. 2013; Strom & Hickman 2010). However, conservative estimates indicate there are at least 200,000 untested SAKs in U.S. police departments, and large stockpiles of kits have been documented in over five dozen jurisdictions, sometimes totaling more than 10,000 untested SAKs in a single city (Campbell et al. 2017a, 2017b). The growing national problem of untested rape kits has garnered the attention of Human Rights Watch (2009: 7) because “international human rights laws require police to investigate reports of sexual violence and take steps to protect individuals from sexual assault.” Likewise, the Department of Justice's (2015) report, Gender Bias in Law Enforcement Response to Sexual Assault, specifically highlighted the problem of untested rape kits as an example of biased and discriminatory police practices.

Ultimately, it is the responsibility of the police to submit SAKs for forensic DNA testing, and law enforcement personnel have tremendous discretion in what actions they do and do not take in sexual assault investigations (Department of Justice 2016; Human Rights Watch 2013). Focal concerns theory offers a useful framework for examining discretionary decision making in the criminal justice system (Spohn et al. 2014). Briefly, this theory stipulates that criminal justice system personnel (police, prosecutors, judges) define as particularly salient certain aspects of a crime/case, given their specific roles and responsibilities in the system (Steffensmeier et al. 1998). A common focal concern across all sectors of the criminal justice system is protecting public safety (Steffensmeier et al. 1998); as such, actions that can help solve crimes and prevent future attacks would be expected to be key priorities. In the context of sexual assault cases, rape kit testing may help protect public safety via DNA identification of unknown offenders, confirmation of offender identities, and the discovery of serial perpetrators by DNA matches across crimes (Campbell et al. 2017a, 2017b; Human Rights Watch 2009, 2010; Strom & Hickman 2010). And, yet, there is growing evidence that police do not test rape kits as a matter of routine law enforcement practice, so this contradiction suggests that other focal concerns may be taking precedence. Why then are police not submitting SAKs for forensic DNA testing?

In this article, we take up this question by examining why one city—Detroit, Michigan—had thousands of untested SAKs in police custody. In August 2009, approximately 11,000 rape kits were discovered in a Detroit police storage facility, the vast majority of these rape kits had never been submitted for forensic DNA testing. Our goal in this study was to understand why SAKs were not routinely tested in sexual assault cases that were reported by victims to the police from 1980 to 2009. Prior studies using focal concerns theory to understand the criminal justice response to sexual assault have been between case analyses to identify what differentiates cases in which an offender was arrested versus not arrested, cases that were prosecuted versus not prosecuted. However, focal concerns theory can also be a useful framework for in-depth study across cases that were not pursued by the criminal justice system to identify common patterns. What focal concerns did the police attune to in their decisions regarding rape kit testing? What other features of the case were more salient to the police than the potential evidentiary value of the SAK? Because DNA evidence can be instrumental in protecting public safety, it is important to understand why this resource is not being routinely utilized in sexual assault investigations.

Understanding Why Police Do Not Submit SAKs for Forensic DNA Testing

To set the stage for this study, we draw upon two distinct lines of inquiry that can inform our understanding of how police make decisions regarding SAK forensic testing. First, a small but growing are of research examines why police submit crime scene evidence for forensic testing, including sexual assault crimes. This body of work frames forensic testing as a specific choice-point in police practice in which law enforcement personnel consider the potential utility of testing to the investigation and whether there are resources available for such testing. Second, a large multidisciplinary literature examines discretionary decision making in sexual assault investigations. These studies cast a wider lens to consider why police invest investigational effort in some reported rape cases but not others, and how victim credibility is often more salient than evidentiary concerns. Taken together, these literatures help us understand the use of forensic testing as a specific investigational tool within the broader context of how police typically approach sexual assault investigations.

Police Practices for Forensic Testing of Crime Scene Evidence

In two national-scale surveys of law enforcement agencies’ forensic evidence testing practices in unsolved homicides, rapes, and property crimes from 1982 to 2007 (Lovrich et al. 2004; Strom & Hickman 2010), the most commonly cited reason for not submitting evidence for DNA testing was that the police did not have an identified suspect in the case. This finding may seem puzzling because DNA testing can help reveal offender identity. However, when forensic DNA technology first emerged in mid 1990s, it had limited utility for suspect identification because there was no national database of reference samples to which DNA profiles could be compared (Butler 2005). In 1998, the FBI established a federal criminal DNA database, CODIS, which allowed police to submit either an unknown or known DNA profile and search records against criminal reference samples from convicted offenders, arrestees, and other crime scenes; a match (termed a “hit”) could provide an investigational lead (i.e., identification or confirmation of the offender). Yet, police did not seem to change their testing practices after the emergence of CODIS. Police also indicated that they did not submit evidence for testing if the suspect had already been adjudicated or was expected to be adjudicated, if there was no specific request from the prosecutors to test the evidence, and if no charges were expected to be filed against the offender (Lovrich et al. 2004; Strom & Hickman 2010). These findings suggest that police have viewed DNA evidence as a confirmatory check for cases that will be going to trial, but not as an investigatory resource that can help build a case to determine if it should be referred for prosecution.

Police agencies surveyed by Lovrich et al. (2004) and Strom and Hickman (2010) also noted they did not submit evidence for forensic testing if they believed their crime laboratories did not have sufficient resources to test evidence. Likewise, Tasca et al.'s (2013) regional-scale study in Arizona found that police did not rely on DNA evidence because they thought forensic analysis would take too long to be useful to the investigation. In Peterson et al.’ (2012) examination of untested SAKs in Los Angeles, lack of funding for testing and lengthy delays in processing were contributing reasons why police did not submit SAKs for DNA testing. The National Research Council (2009) noted that U.S. forensic laboratories are under-resourced to meet growing demand for crime scene evidence testing, including, but not limited to rape kit testing. These resource constraints likely bolster police perceptions that DNA evidence is not a resource that can be utilized in all investigations.

Police Practices in Sexual Assault Investigations

If forensic laboratories do not have the capacity to test all crime scene evidence, then police must decide what evidence from which cases will be submitted for testing. In the context of sexual assault investigations, police exercise considerable discretion, up to and including decisions whether to conduct an investigation at all, as noted in recent high-profile critiques of police practices in Washington, DC (Human Rights Watch 2013) and Baltimore, MD (Department of Justice 2016). Spohn and colleagues have advanced focal concerns theory as a conceptual model for understanding discretionary decisions in sexual assault investigations (Kaiser et al. 2017; Spohn & Tellis 2010; Spohn et al. 2014). Spohn et al. (2014) noted that the focal concerns of law enforcement include the seriousness of the crime, the degree of injury to the victim, and the blameworthiness and dangerousness of the offender. However, there are also practical concerns that police must attune to, and certainly resource constraints for forensic testing would be such a factor. But, Spohn et al. (2014) argued that another key practical consideration for the police is the likelihood of conviction. This “downstream orientation” focuses on what factors may help secure a conviction and privileges those considerations in upstream decision making (Frohmann 1997). For example, a victim's testimony in court could affect convictability, so a victim's credibility is paramount to the police (Kerstetter 1990). Closely related to perceived credibility is the degree to which victims are perceived as cooperative with law enforcement during the investigation. The extent to which there is physical evidence that can corroborate victims’ statements is also important. These three factors—credibility, cooperation, and evidence of the crime—affect police effort in sexual assault investigations in general and could affect decisions regarding rape kit testing specifically.

Whether victims are deemed credible is based on an expansive set of factors that encompasses who they are (gender, age, race/ethnicity, socioeconomic status); who assaulted them (a stranger or someone known them); where the assault occurred (neighborhood and time of day); why they might be making a report of rape (altruistic or revenge motives); how they behaved prior to the assault (engagement in “risky behaviors,” such as alcohol/drug use); how they behaved during the assault (physical resistance); and how they behaved after the assault (prompt reporting to the police, visible signs of distress) (Frohmann 1997; Jordan 2004; Kaiser et al. 2017; Kelley & Campbell 2013; Morabito et al. 2016; Murphy et al. 2014; Schwartz 2010; Spohn & Tellis 2010; Spohn et al. 2014; Tasca et al. 2013; Venema 2016). Perceived victim credibility is based on a complex set of attitudes and beliefs as to what constitutes a “real rape” and real victims (Estrich 1987; Lonsway & Archambault 2012). Jordan (2004) noted that it is nearly impossible for any victim to meet all of the implicit and explicit criteria of what constitutes a “real rape” in the eyes of the police. Recently, Morabito et al. (2016) suggested that there is a cumulative “piling on” effect whereby police deliberately look for multiple indicators of what they believe is poor credibility so they can discount victims’ accounts of rape.

A particularly salient finding in this body of work is that if the police think there is even a possibility that the act in question was consensual, they are less likely to view the victim as credible. For example, in interview studies, police directly state that they find victims less credible if they knew the perpetrator and had prior social/sexual contact, which, to their thinking, may mean that rape allegations could be fabricated because women regret having sex and/or want to seek revenge on their partners (Jordan 2004; Spohn et al. 2014; Venema 2016). Adolescents are often singled-out by police as being particularly less credible, as law enforcement believe that their claims of rape are fabricated to cover up for “bad behavior” (being out late, drinking) and to try to avoid getting into trouble with their parents for those behaviors (Schwartz 2010; Venema 2016). Police also highlight that claims of rape are not credible from women who they believe are engaged in prostitution (Venema 2016). The sexual assault detectives interviewed by Schwartz (2010: 34) dismissed rape allegations by women they believed were involved in sex work as “economic crimes,” meaning that they alleged rape when they were not paid.

What police say about victims in research interviews is consistent with their actions on the job. In reviews of sexual assault police reports, it is not uncommon to find instances of law enforcement personnel writing about victims in harsh, often judgmental language. For example, Tasca et al. (2013: 1167) cited a police report in which a detective wrote, “Victim is a prostitute, addict, smoked crack with suspect;” (see also Shaw et al. 2017). These indexes of perceived credibility—and others—are consistent predictors of case closures, exceptional clearances, unfounding decisions, false report designations, or otherwise no actions taken in reported rape cases (Kaiser et al. 2017; Kelley & Campbell 2013; Morabito et al. 2016; Murphy et al. 2014; Pattavina et al. 2016; Spohn & Tellis 2010; Spohn et al. 2014; Tasca et al. 2013). From this body of work, it stands to reason that police may not submit rape kits for forensic DNA testing if they believe that victims are not credible.

Closely related to perceived victim credibility is victim cooperation (Kaiser et al. 2017; Spohn & Tellis 2010). Frohmann (1997) noted that prosecutors need victims’ engagement for successful prosecutions, so if the police perceive that victims are not cooperative, it raises concerns about whether they should invest effort in such cases (Venema 2016). For example, in Schwartz's (2010) interviews with sexual assault investigators, detectives said that a major problem they faced in their work was victims not telling the complete truth immediately. Schwartz termed this “impression management”: victims are thought to withhold or hide information, often because they are embarrassed about some aspect of the incident, so they try to put their behavior in the best possible light. When the “full story” or contradictory information comes out later in the investigation, this bolsters detectives’ beliefs that victims are not truthful. The detectives in Schwartz's (2010) study felt this pattern was especially common with adolescents, so this population was viewed with particular distrust.

However, interview studies with rape survivors suggest a more complex picture regarding victim cooperation and its role in sexual assault investigations. Patterson (2011a,2011b) found that how victims are treated by detectives in the investigational interview affects how much they disclose. In other words, the complete forthrightness that police demand is very much affected by the police themselves. When interviewed kindly by detectives, victims disclose more (Kaiser et al. 2017; Patterson 2011a,2011b). When victims are subjected to harsh repeated questions, accusations of lying, and threats of punishment for not telling the full truth, victims may withdraw and disengage (Spohn & Tellis 2010). Interviews with adolescent rape victims suggest this kind of treatment is particularly common in their interviews with detectives (Greeson et al. 2016). Campbell (2008) referred to this pattern of treatment as “secondary victimization”: insensitive, victim-blaming treatment that makes survivors feel as though they are being re-raped by the systems that are supposed to help them (see also Bumiller 2009; Corrigan 2013). When victims experience secondary victimization, they may choose to disengage, largely for self-protective reasons (Greeson & Campbell 2011; Patterson et al. 2009). Thus, victims may become noncooperative or can be made to disengage.

In sexual assault police reports, this context is stripped away and what is documented in writing is simply that the victim did not cooperate with police (Murphy et al. 2014; Shaw et al. 2017; Spohn & Tellis 2010; Tasca et al. 2013). Several studies have found that law enforcement does not routinely invest substantial effort in rape investigations, but whatever effort is put forward, however slight, is expected to be returned in kind, and promptly. If victims do not return a call or attend a meeting, they are labeled as uncooperative (Murphy et al. 2014; Shaw et al. 2017; Spohn & Tellis 2010; Tasca et al. 2013). If victims do not consent to a medical forensic exam/SAK, this may also be noted as a specific indicator of poor cooperation (Murphy et al. 2014; Tasca et al. 2013). However, many victims do have a SAK collected, but the police do not consistently submit kits for testing, perhaps because there were other indicators of perceived lack of engagement. In the end, if victims are not perceived as cooperative, the likelihood increases that their cases will be cleared by exceptional means or otherwise closed/not pursued (Murphy et al. 2014; Pattavina et al. 2016; Spohn & Tellis 2010; Tasca et al. 2013).

Convictability is also affected by the quality and quantity of the evidence supporting the state's case, which could include physical evidence, forensic evidence (DNA, injury documentation), witness statements, and suspect interviews. Spohn et al. (2014) noted that police often do not have much by way of evidence at the beginning of an investigation, and fall back on stereotypes, usually about victims’ credibility, to guide decisions. Law enforcement personnel could collect and test evidence (e.g., rape kits) so they would not have to rely on stereotypic schemas (Venema 2016), but because police have limited resources, they may not want to invest time and effort in cases in which they are concerned about victim credibility and cooperation (Murphy et al. 2014; Spohn & Tellis 2010; Tasca et al. 2013; Venema 2016). Cases that do not have corroborating physical or forensic evidence are less likely to result in an arrest, and thus are less likely to be charged or convicted (Murphy et al. 2014; Spohn & Tellis 2010; Tasca et al. 2013). But, the growing national problem of untested rape kits suggest that thousands of cases may in fact have corroborating evidence, but the SAKs have been shelved (literally) and reported rapes are being shelved (figuratively). As such, there is a pressing need to understand why this evidence is not tested and used by law enforcement personnel.

The Current Study

Prior research suggests that police do not submit crime scene evidence for forensic DNA testing because they do not believe that such evidence will be beneficial during the investigation stage of the criminal justice system. It is a costly resource that will likely be needed for the cases that do advance to prosecution, so DNA testing may not be feasible as routine investigational practice. From this point of view, the national problem of untested rape kits may be due to resource constraints, which has certainly been a common theme in media coverage of this problem. However, focal concerns theory suggests that there may be additional reasons why the limited resources of the criminal justice system are not invested in sexual assault cases. Police have considerable discretionary power in rape investigations, and a large body of research suggests that they privilege concerns about victim credibility and cooperation over evidentiary concerns. Thus, decisions about how to expend limited resources may be based on long-standing beliefs about helping only “real victims” (Estrich 1987; Lonsway & Archambault 2012). To understand the focal concerns of police regarding rape kit testing, we conducted an in-depth qualitative study in one community that had large numbers of untested SAKs to explore why they did not routinely submit rape kits for forensic DNA testing, considering both the resources available for SAK testing as well as their general norms and practices for rape investigations.

Detroit, Michigan discovered that it had large numbers of untested SAKs in August 2009 when representatives from the local police, state police, and the prosecutor's office toured a remote property storage facility to discuss how to best manage the volume of evidence in police custody. During that tour, an assistant prosecutor noticed dozens of large storage boxes on shelving units, and when asked what they were, police personnel indicated that they contained rape kits. When pressed for details about the kits, police officials were not able to verify how many SAKs were in police property and how many of those SAKs had been tested. An initial review by a state-level agency found that there were approximately 11,000 SAKs in police custody, dating back to 1980, the vast majority of which had never been submitted for DNA testing (Michigan Domestic and Sexual Violence Prevention and Treatment Board 2013). Local, county, and state officials demanded an in-depth examination of the problem. Coincidentally, the National Institute of Justice had just released a funding announcement to support action research projects on untested SAKs, as an increasing number of U.S. cities were making similar discoveries (National Institute of Justice 2010). Detroit stakeholders and the state agency that conducted the initial review of the kits reached out to the senior author of this article to discuss developing a joint collaborative application for this grant solicitation, given our work together on prior projects on sexual assault and the criminal justice system. Detroit's researcher-practitioner application was selected for one of two NIJ-funded action research projects on unsubmitted rape kits. This project brought together practitioners from law enforcement, prosecution, forensic science crime laboratories, medical/nursing, and victim advocacy to work together with researchers (the authors of the article) to develop long-term solutions for resolving these untested kits (see Campbell et al. 2015b).

The funding solicitation specifically stated that a key purpose of these projects was to “understand the underlying nature of the problem” (National Institute of Justice 2010: 1). At that time this project began, law enforcement officials were defending their decisions not to test these rape kits for DNA and did not perceive that there was a problem to be solved. By contrast, practitioners from other disciplines—victim advocacy, nursing/medicine, prosecution, and forensic sciences—were alarmed that so many kits had not been tested, particularly because so many of these victims were black women and/or poor women. When stakeholders have such vastly different perspectives on an issue, qualitative methods are an ideal choice, as they are well-suited for capturing complex and divergent points of view (Denzin & Lincoln 2011; Lincoln & Guba 1985; Patton 2015). To understand why the police decided not to submit thousands of rape kits for DNA testing, and why they did not think it was problematic that they had 11,000 SAKs in storage, we embarked upon a three-year ethnography that included sustained observational work, longitudinal qualitative interviews, and archival record reviews of sexual assault police reports associated with untested rape kits. Our goal was to understand why the police believed that these rape kits did not merit testing.

Methods

Ethnographic Observations: Sample and Procedures

The Detroit SAK Action Research Project (“the SAK collaborative”) met in-person, bi-monthly for three years to resolve the untested SAKs in this jurisdiction (see Campbell et al. 2015b for an extended discussion of how this partnership was developed and sustained over time). Multiple representatives from the police department attended these meetings, including senior leadership (e.g., deputy chief, commander of criminal investigations), the supervising sergeant of the sex crimes unit, as well as several front-line case detectives. The other participating organizations in the collaborative included the county prosecutor's office; the state forensic crime laboratory; the local sexual assault nurse examiner (SANE) program; and local, state, and national victim advocacy organizations. Each of these organizations also had multiple participants, including high-level leadership and front-line practitioners from each group. The research team members were participant-observers in these collaborative meetings (∼186 hours of observation). All members of the SAK collaborative were briefed individually and as a group regarding Institutional Review Board (IRB) procedures for ethnographic observations so that they understood that their remarks would be written down; informed written consent was obtained from all collaborative participants (100 percent participation). After each meeting, the researchers wrote fieldnotes and conducted preliminary open coding of those notes within 72 hours of each observation in order to monitor data quality, identify new areas of inquiry, and assess saturation of themes (Emerson et al. 1995; Patton 2015).

Qualitative Interviews: Sample and Procedures

All members of the SAK collaborative (across all disciplines and organizations listed above) were asked to participate in one-on-one interviews (i.e., 100 percent participation rate). In those interviews, we asked participants to nominate other key individuals within their organizations that we should also interview, given their knowledge and expertise (e.g., key individuals who had changed positions or retired but were once closely involved in these issues; 100 percent participation rate). Over the three years of primary data collection in this project, we conducted 42 interviews with Detroit stakeholders, spanning all organizations and staffing levels within each organization (16 cross-sectional interviews and 26 longitudinal interviews). The interviews were semi-structured qualitative assessments that examined many topics, including: (1) the participants’ current job position and their role in SAK testing, investigation, prosecution, and/or victim advocacy; (2) their organization's past and present procedures regarding SAK testing and sexual assault investigations; (3) their beliefs regarding why some SAKs were submitted for forensic testing and others were not; and (4) their understanding of the resources available in their organization for SAK testing, sexual assault investigations and prosecutions, and/or victim advocacy. Written informed consent was obtained from all participants and the interviews were digitally recorded with participants’ permission and transcribed verbatim.

Archival Records: Sample and Procedures

To supplement the information from our ethnographic observations and qualitative interviews, we requested archival records from each organization within the SAK collaborative regarding their staffing, resources, services provided, and policies and procedures regarding SAK testing and sexual assault investigations/victims services. A total of 93 individual documents were provided to the research team. Each record was reviewed by two research team members to determine what information should be extracted/captured for later analysis (akin to conducting preliminary open coding of narrative data). We also reviewed sexual assault police reports associated with untested SAKs as another source of information and insight into why law enforcement personnel did not routinely submit rape kits for forensic DNA analysis. From the ∼11,000 SAKs in Detroit, we drew a stratified random sample of 1,600 that included both stranger-perpetrated sexual assaults and non-stranger-perpetrated sexual assaults (see Campbell et al. 2015a). We requested the police reports for the sampled cases, and law enforcement personnel were able to locate 1,268 reports (the remaining had been lost over time). These records were coded for another component of project (i.e., extraction of data fields pertaining to assault characteristics, see Campbell et al. 2015a); in addition, after each batch of coding, the researchers wrote qualitative field notes to document key themes in the reports (e.g., typical level of investigatory effort in the cases, common comments made by the police about the victims). During the review and coding of these police files, the research team members also made note of reports that could become typical case studies of key themes in the data (Emmel 2013).

Qualitative Data Analytic Approach

To analyze these varied types of data, we used Miles et al.'s (2014) analytic framework, which is a rigorous, pragmatic approach for analysis of narrative data. In the first phase of analysis—data condensation—focuses on organizing the data, developing preliminary themes, and preparing the data for more refined analyses. Similar to Corbin and Strauss's (2008) concept of open coding, the ethnographic observation fieldnotes, interview transcripts, and archival documents were independently reviewed by two analysts throughout data collection, who (separately) tagged and labeled concepts to define and develop preliminary thematic categories. After every 2–3 new observations, interviews, and/or records, the new data were checked against the existing codes to see if there were new codes to be created and new patterns emerging. Coding was revised to accommodate the new data, and if warranted, data collection protocols (e.g., interview questions, archival records to request) were revised to seek out additional/clarifying information.

Also as part of this first phase of analysis, we conducted a triangulation assessment to gauge the quality and credibility of the data before continuing with more detailed coding (Creswell & Clark 2011; Lincoln & Guba 1985). Using a visually-intuitive green-yellow-red color-coding system, we coded the extent to which each preliminary open code could be triangulated. “Green” data had multiple data types (e.g., interviews and archival records), across multiple organizations confirm the presence and meaning of a theme; “yellow” data had conflicting information across organizations and no other data types were available to resolve the information/interpretation; “red” data had only one data source supporting a theme. In these triangulation analyses, the vast majority of the open codes were “green,” but there was one “yellow” theme that we elected to continue to analyze, and we will specifically highlight those data in the presentation of the results. No “red” data were further analyzed. To round out this first phase of data analysis, we then developed refined sub-codes for each “green” and the one “yellow” triangulated open codes, which is akin to Charmaz's (2006) stage of focused coding. This refined coding was conducted by three coders so that each sub-code was reviewed and discussed by multiple analysts (MacQueen et al. 2008).

The second phase of data analysis—data display—focused on organizing comparisons and contrasts of the data, akin to Corbin and Strauss's (2008) concept of axial coding. We constructed micro-level tables that organized data from different sources by the individual themes (e.g., data from observations, interviews, and archival records regarding the triangulated sub-theme “police stereotypes re: adolescent sexual assault victims”). Then, these tables were combined into macro-level tables that examined the associations between themes (e.g., “police attitudes re: adolescent sexual assault victims” and “police investigational effort”). This process resulted in multiple organizations of the data that from which we could examine interrelations among the identified themes.

The third phase of analysis—drawing and verifying conclusions—involved constructing mechanistic linkages between themes in the data. For this phase, we drew upon Erickson's (1986) analytic induction method, which is an iterative procedure for developing and testing empirical assertions in qualitative research. An assertion refers to a hypothesized pattern in the data (e.g., “police officers held negative beliefs about adolescent sexual assault victims and because of those views, they were less likely to invest investigational effort in their cases, more likely to question young women's credibility, resulting in a SAK not being submitted for testing”). Two analysts worked together to develop assertions from the data, and then one analyst used Erickson's (1986) methods to evaluate the adequacy of the assertions. To determine whether an assertion was substantiated/well-founded, the analyst assembled confirming and disconfirming evidence, looking for five types of evidentiary inadequacy: (1) inadequate amount of evidence; (2) inadequate variety in the kinds of evidence; (3) faulty interpretative status of evidence (i.e., doubts about the accuracy of the data due to social desirability bias); (4) inadequate disconfirming evidence (i.e., no data were collected that could disconfirm a key assertion); and (5) inadequate discrepant case analysis (i.e., no cases exist that are contrary to a key assertion) (Erickson 1986: 140). Assertions were revised or eliminated based on their evidentiary adequacy until a set of well-warranted assertions remained, similar to Glaser's constant comparison process (Glaser 2007).

In this final phase of analysis, we also used several strategies to verify the trustworthiness of the analyses and conclusions (Creswell 2012; Lincoln & Guba 1985). Throughout data collection and analysis, we maintained prolonged and persistent engagement in the setting, conducted memoing and peer debriefing, and kept an audit trail of all coding decisions and analysis methods. We also conducted member checks by presenting these findings to the SAK collaborative for their review and feedback; a draft final written report was also circulated to all participating organizations for comment. The SAK collaborative organizations (including the police department) did not contest the major findings, though they did ask clarifying questions that suggested we needed to add more contextual details to our presentation of the findings, which we have done. However, as expected, there was one specific finding (pertaining to the “yellow” triangulated code) that the police did contest, which is noted as such in the “Results” section, along with their objection and our rationale for reporting it.

Results

“The Kit Isn't the Issue”: Reframing the Problem of Unsubmitted SAKs

The primary goal of this study was to understand why police did not submit thousands of rape kits for forensic DNA analysis, so in our interviews with law enforcement personnel, we asked specific questions about SAK submission guidelines and decisions (e.g., “Can you tell me about why a SAK would or would not be submitted,” “Can you tell me about how you decide whether to submit a SAK for testing”). Police stakeholders often said that we were “barking up the wrong tree” because kit submission was only one part of a larger decision-making process, as one police official explained:

“The kit isn't the issue … it's the investigation, figuring out what we could do … if [there's] something we could do, then we would. (Q: would that include submitting the kit for testing?) Depends. If it was a stranger … sometimes, yes, sometimes no … you (referring to the research interviewer) keep asking about the kit. Wrong question. What mattered was the case and whether it was real and whether we could do anything about it.”

In the SAK collaborative team meetings, police personnel often expressed frustration at what they termed the “preoccupation” with the unsubmitted rape kits. From their point of view, the kits had not been submitted for testing because they did not merit testing—and they did not merit testing because the cases were “weak.” Police noted repeatedly that submitting a SAK for forensic testing was not a discrete, separate decision based on whether they thought testing the kit would help the investigation; whether a kit would be tested depended on preceding decisions regarding the overall merit of the case.

In light of this finding, our focus shifted to understanding how police approach sexual assault investigations more generally. Cases associated with unsubmitted kits were not, as a group, thoroughly investigated. In many instances, it would be difficult to claim that they were investigated at all. The vast majority of the police reports we reviewed were 1–3 pages long, which included the responding officer's summary of the victim's account of the assault and sometimes detectives’ notes about his/her interaction with the victim. It was rare to find documentation regarding a search for the suspect, a suspect interview, witness interviews, crime scene evidence, or other investigatory efforts. The final case dispositions were often missing in the files, and police personnel did not consistently or correctly use the FBI sexual assault clearance codes. Rather, the police seemed to follow their own local coding system, and most of the cases we reviewed were marked “UTEEC” (unable to establish the elements of the crime), “CRTP” (complainant refuses to prosecute), “to locate” (offender was still to be located), or “pending victim contact” (the victim needed to make contact with the police for the investigation to move forward). When we asked why so many of these cases were closed and the kits were not tested, one police official summed up what we heard from many in law enforcement when he/she said: “The kits [that weren't] tested were cases that we couldn't or wouldn't do anything about.” Why would the police decide a case is something that they “couldn't” or “wouldn't” act upon?

“Couldn't Do Anything About It”: The Impact of Chronic Resource Constraints

In our ethnographic observations, stakeholders across all organizations in the SAK collaborative noted that the police department had struggled for decades with chronic resource constraints. Staffing shortages contributed to an organizational mindset that it was impossible to respond appropriately to all reported crimes, so there would always be cases they “couldn't do anything about.” From the 1980s to the early 2000s, the sex crime unit had, on average, 20–30 investigators and staff (e.g., a designated property officer). In 2002, the staffing levels were cut by approximately 50 percent (down to 12 investigators/staff), and in 2008, the sex crimes unit had another 50 percent cut in staffing, down to 6–8 investigators/staff. Police officials spoke at length about how difficult it has been to sustain quality police work under decades of resources cuts, as one leader explained:

“It sounds like an excuse, [and] there's no excuse, [but] when you have fewer investigators and the same caseload, you have less time to investigate each case…[when] you don't have enough people to investigate it, you know, that can come back to really haunt you.”

High caseloads and understaffing also led to chronic burnout, which also negatively affected investigational quality, as one police stakeholder noted:

“I do think it's a resource issue; I do think that officers are so overwhelmed with the next case walking through the door that they take short cuts, and that they don't do everything they should do. I think the burnout is very high.”

How do police approach sexual assault investigations when they are under-staffed, exhausted, and burned-out? Figure 1 depicts the process that emerged from our ethnographic observations, qualitative interviews, and review of 1,268 sexual assault police reports associated with untested rape kits. Law enforcement officials were candid that the investigations for many of the sexual assault cases associated with unsubmitted SAKs were not thorough. One stakeholder summed up how many of these cases were handled: “Okay, I made a couple phone calls, I threw my card in the door and that's the end of that. Close it.” Speaking to this issue of the thoroughness of investigation, one police official said:

Figure 1. Police Processing of Sexual Assault Cases Associated with Unsubmitted Sexual Assault Kits.

“When you have fewer investigators and the same caseload … the reality of it that you may not cross all your t's and dot all your i's. …where do you cut? Canvassing, looking for additional witnesses … personal contacts … Those are things, that's where you'll see corners cut or we should be reaching out more talking to people, we just don't, we don't have a chance to.”

Another strategy for deciding how to “allocate limited time for unlimited cases” was to wait for the victim to initiate follow-up contact with the police. As one police supervisor described:

“I get calls from victims saying my detective hasn't called me and I'll look in the notes and it says, I made a phone call, I called at 3 on Tuesday afternoon, there was no answer, I left a message’ … [the investigator] may not put in the effort that we would like to see him put in … he may just say, make a phone call and that's it. Then if she wants to complain or she wants to prosecute, I'm sure she'll be contacting me.”

In other words, police would shift the responsibility to the victim to contact them and pursue the case, or as one investigator said, “she has to prove she wants this … then I'll take a look.” When asked about the fairness of this practice, police readily acknowledged that it was not fair to victims, but that it was one of many strategies investigators used to manage workload. As shown in Figure 1, resource constraints led to “cutting corners” and decreased effort on cases; when victims did not respond to investigators’, admittedly, minimal efforts to contact them and develop the investigation, police often labeled victims as “not cooperative” or “refused to prosecute,” and the case was not pursued further.

However, chronic resource scarcity explains only why police “couldn't” invest effort in all reported sexual assaults. Understanding the reasons why they “wouldn't” invest effort requires delving deeper into how law enforcement personnel were approaching their work on sexual assault cases and how they were treating victims. Highlighting this point that resource depletion is only part of the story, one member of the SAK collaborative noted:

“This isn't the whole story … sure, for thirty years, and even longer than that, this city hasn't had [what it] needs to care for its citizens. But a lot of caring did happen. This didn't. This crime (sexual assault) wasn't cared about … these citizens (sexual assault victims), specifically, weren't cared about … there weren't enough resources to go around … what little there was, it didn't go here (to help rape victims and test SAKs) … There was a choice, lots of choices, choices every day not to help a victim.”

“Wouldn't Do Anything About It”: The Impact of Victim-Blaming Beliefs

Police noted that there were reported sexual assault cases that they “wouldn't do anything about,” and stakeholders from all organizations in the SAK collaborative—except the police—argued that gender, race, and social class were clearly reasons why.¹ As one stakeholder said, “I think that's probably the number-one reason [why kits aren't submitted], it affects mostly women … if men were getting raped, I think that it wouldn't be like that.” Similarly, another member of the collaborative said, “It's not that complicated to figure out … this is a crime that affects women, and in this city, that means Black women, poor Black women … there's a good chunk of the explanation right there.” Prosecutors, victim advocates, forensic nurses, and forensic scientists argued that intersectional sexism, racism, and classism was fundamental to why the police would not help so many rape victims:

“Many of them are poor … many of them are living not only a legacy of racism but active racism, active misogyny … and they have multiple problems [in their lives] … and the criminal justice system [is just going to] exacerbate [that.]”

Police officials expressed strong disagreement with these assertions, highlighting that many members of the department are African American and many are female, and that among the higher leadership ranks of the department, women and African Americans have been well-represented throughout the years.

Whereas there was disagreement about whether general stereotypes regarding gender, race, and social class were a factor in why police would not help many rape victims, there was clear evidence across all data sources and across all organizations that police held negative beliefs about specific populations of rape victims, which adversely impacted case investigations and ultimately SAK submissions. First, police personnel often assumed that victims reporting sexual assaults were engaged in prostitution/sex work. In both the stakeholder interviews and in the police reports, there were frequent references to “deals gone bad”:

“[sometimes it was] a deal gone bad, she got herself caught. (Q: ‘got herself caught,’ what does that mean?) She was prostituting and she agreed to the money and he didn't pay her …she says it's rape. It's not, it's a deal gone bad.”

In our interviews, we asked police how and why they suspected a victim might have been involved in prostitution, which one investigator described as:

“It's not one thing, usually, neighborhood, street, circumstance of the assault, like if she accepted a ride with someone … how she looked … can't put your finger on it exactly, but you do this long enough, you can tell.”

When asked if women involved in sex work could indeed be victims of rape, some police officials agreed (e.g., “of course they can be raped … some rapists prey on them ‘cause people won't believe them”), but many did not. Either way, their overriding concern was that if it was a “deal gone bad,” then they would not want to invest limited resources investigating the case. When asked if it was possible that investigators would label a case in that way in order to manage their workloads, one police official stated: “I'd like to think that didn't happen, but it yeah, I'm sure it did.”

If police believed that victims may have been involved in prostitution, they readily acknowledged that they treated them differently, often acting in deliberately intimidating ways to “test them,” as on officer put it, to see if they were telling the truth, and sometimes to “nudge them” out of the system and discourage them from continued pursuit of their report:

“I guess one of the ways you could discourage is being the jaded police professional… . Just coming off wrong … now in my mind they're prostitutes… my line of questioning or … demeanor [shows that] … [I'm] discouraging them.”

After being treated in such a way, victims might not cooperate with the police, thereby bolstering investigators’ assessments that the case should not be pursued further.

Bringing these ideas together, Figure 2 presents a case example police report in which the victim is assumed to be a prostitute and that belief appears to have dictated the course of the investigation, to the point of discounting possible evidence of a crime. In this case, a 22-year-old woman was attacked outside a liquor store by a known acquaintance, who pulled her into a vacant house, threatened her with a gun, and then sexually assaulted her. Afterward, she ran to a nearby fire station and the staff took her to the hospital. As can be seen in Figure 2, the officer explicitly states his/her belief that this was a “deal gone bad.” The victim's concerns about discussing the assault with the police are re-framed as duplicity on her part and the investigator told the victim that felony charges could be filed against her if she was not telling the truth. Though we do not know how the victim interpreted the investigator's comments about felony charges being filed against her, it seems likely that such actions were at the very least intimidating, and at worst, threatening. The report also shows the “let-the-victim-contact-us” strategy discussed before, whereby police appear to be managing their workloads by expecting survivors to pursue them to investigate the case. The report ends with a mention of physical injuries to the victim, but these facts do not appear to be a key concern to the investigator.

Figure 2. Sexual Assault Police Report of Suspected Prostitution “Deal Gone Bad.”

The second common stereotype that emerged in the stakeholder interviews and police reports was that the credibility of adolescent victims was questionable and that they made claims of rape to cover up for “bad” behavior as one detective described:

“The young girls, they say something went down so they won't get in trouble with their mammas ‘cause they were out late or with older men or doing something they shouldn't be doing.”

Police noted that they would “lay it on thick” with adolescent victims, describing the criminal justice system process in deliberately scary terms, warning them about potential negative consequences if they were found to be lying.

Figure 3 presents a police report case example that illustrates these themes and processes. A 14-year-old girl reported that she had been abducted by two men, taken to an abandoned building and raped. At the beginning of the case report (top part of Figure 3), the officer states his/her belief that the report is false—a statement that preceded any details about the assault itself. The last sentence of this report highlights that the victim was supposed to be home at 7:00 p.m., but did not return until 8:30 p.m. The relevance of this statement is unclear, though it implies that the account may have been fabricated to cover up for being late. The investigator's supplemental notes (bottom part of Figure 3) elaborate on these themes, highlighting how the officer did not believe the victim's account of the assault. Referring to the victim in writing as a “heffer” also clearly shows the investigator's disrespect and disbelief. This case was closed (UTEEC), and it does not appear from the documentation in the file—which consists of only these two documents—that there was any attempt to review the scene of the crime, canvass for witnesses, or search for the assailants. When this case example was presented to police stakeholders as part of the member-checking process, they expressed alarm at the tone, content, and language of the report, but they acknowledged that disbelief of adolescent victims was a pervasive problem.

Figure 3. Sexual Assault Police Report of an Adolescent Victim “Covering Up Bad Behavior.”

The third common stereotype was to question the legitimacy of victims who knew their assailants: police doubted victims’ credibility if they knew or were even minimally acquainted with the assailant. In the stakeholder interviews, police expressed frustration about these kinds of cases because the accused perpetrators often claim that the incident was consensual, which law enforcement felt was difficult to prove or disprove: “[it's] impossible to prove lack of consent.” When asked about why it was “impossible,” one investigator clarified that they can establish the elements of the crime, including lack of consent, but that it is often time-consuming to do so and time to invest in such cases is often limited. Investigators emphasized that given how pressed they were to keep up with their caseloads, they were particularly frustrated and skeptical of any case that seemed, to them, like a “wake up and regret report” or “revenge report,” meaning, the victim was making an accusation of rape to “get back at” a friend or partner because they were upset or regretting what had happened. When asked about how common it was that known associates, friends, and/or partners rape their partners, police acknowledged that it does happen, but, in their belief, not that often: “Truly rape? Sometimes. But not most of the time.” In terms of how police responded to victims who knew their assailants, stakeholders acknowledged that they sometimes treated victims of non-stranger assault brusquely, as one investigator described his/her approach to known-offender cases:

“I don't believe them, sure I let them know that … if this boyfriend-girlfriend stuff, then that's not my business and I tell them that … if they say never mind, ok, complainant refused to prosecute. Close it and move on to the next one.”

Other police stakeholders noted that victims ought to expect “what they get” if they invite someone over or agree to go somewhere with them. As one investigator stated, “it might not be right, but it's what happens, you go over there, what do you think's gonna happen?”

In the SAK collaborative meetings, police officials emphasized that they did not see the need to test rape kits associated with non-stranger-perpetrated assaults because the identity of the offender was not in question. Other stakeholders challenged that reasoning, emphasizing that confirmation of identity is important in court, and SAK testing could identify serial perpetrators through DNA matches across cases. However, police officials continued to assert that SAKs associated with non-stranger rapes should not be tested because these cases were not, in their view, real rapes. In one-on-one interviews, law enforcement personnel also emphasized that SAK testing was not utilized because they did not believe that these cases were actual crimes, so testing would have been “double moot.”

Figure 4 depicts multiple excerpts of a police report that highlight these dynamics. The initial responding officer's report is three detailed, hand-written pages, sequencing how a 14-year-old girl victim was threatened with being beaten up and killed, orally and vaginally penetrated by force, and subjected to attempted anal penetration by a known acquaintance in his 20's (first excerpt, handwritten); these details were condensed to the second excerpt shown in Figure 4: “compl (complainant) invited known perp (perpetrator) over to watch TV; perp forced sexual intercourse and forced compl to perform fellatio.” The victim's behavior is now the leading element in case file. The initial case disposition was “to locate,” meaning that the assailant (who is known to the victim), needed to be found by police and interviewed. The case was then handed off to a sex crimes unit investigator (third section, cursive handwriting), who questions the victim's statement (“statement and time frame has some holes in it”). The victim provided additional details about the perpetrator, namely that he had been arrested before, but the investigator characterizes the victim as “hostile” when asked for more information. Given that police personnel acknowledged that they sometimes ask questions in deliberately intimidating and challenging ways in order to discourage victims’ continued engagement, it seems reasonable to wonder what had transpired in this exchange. In the end, the final case disposition was changed to “unable to establish elements of the crime.” There is no documentation in this file that the police ever searched for, contacted, or interviewed the named suspect.

Figure 4. Sexual Assault Police Report of a Known-Offender Assailant.

Discussion

Focal concerns theory highlights a fundamental contradiction in the law enforcement response to sexual assault: the police are responsible for investigating reported crimes and protecting public safety, yet their actions and on-the-ground decision making in rape cases often seem at odds with that charge (Spohn & Tellis 2010; Spohn et al. 2014). The growing national problem of untested rape kits is a case-in-point. Submitting SAKs for DNA forensic testing could provide useful information for an investigation, such as identifying offenders and serial crimes, yet in Detroit, and in many cities throughout the United States, police do not routinely submit kits for testing. In this three-year ethnography, our goal was to understand how and why Detroit, Michigan amassed thousands of untested SAKs in police property storage.

There was strong triangulation across observational, interview, and archival record data indicating that Detroit criminal justice organizations did not have sufficient resources for SAK testing and sexual assault investigations. They “couldn't” act on these cases because the police did not have sufficient staffing to investigate all reported sexual assaults in this jurisdiction, so “cutting corners” became normative. Cases were closed, often labeled as “complainant refused to prosecute [CRTP]” after minimal investigational effort. Prior research has found that police often do not invest substantial effort in rape cases and that they label victims as uncooperative if they do not immediately respond to the efforts the police do make (Murphy et al. 2014; Shaw et al. 2017; Spohn & Tellis 2010; Tasca et al. 2013). In this study, we also found that police managed an unmanageable work flow by nudging victims out of the system—by being jaded, by being rude, by questioning them aggressively, by threatening them—and in the end, they blamed victims for the fact that no action had been taken in their case due to their “lack of cooperation” (see also Greeson et al. 2016; Patterson 2011a,2011b; Spohn & Tellis 2010). Yet, resource constraints were not the sole reason why SAKs were not tested. Submission rates for rape kit testing did not significantly change in Detroit with the influx of additional financial and staffing resources (see Campbell et al. 2017b). Furthermore, in our interviews with the police, law enforcement personnel stated with remarkable consistency that they did not submit rape kits for forensic DNA analysis because they did not believe the victims and did not think these cases were worthy of investigational effort.

As to why the police “wouldn't” investigate so many reported rape cases, stakeholders from the prosecutor's office, victim advocacy programs, crime lab, and forensic nursing program argued that intersectional racism, sexism, and classism were factors as so many of the victims in this study were poor, African American women. Frohmann (1997) noted that the criminal justice system ascribes moral judgments of victims’ character based on race and social class, which is why women of Color are so often disbelieved or blamed for their assaults. Similarly, in-depth reviews of police inaction in reported rapes in Washington DC (Human Rights Watch 2013) and Baltimore (Department of Justice 2016) found that systemic racism, sexism, and classism biased police actions, so that poor, racial/ethnic minority women were particularly likely to be discredited and treated harshly. However, in Detroit, police stakeholders strongly refuted the assertion that they were biased, noting that their police force had women, African Americans, and African American women in senior leadership. However, theories of institutional racism emphasize that it is possible to have individuals in leadership positions who are women and/or people of Color within institutions that have systemic, structural-level policies and practices that promote oppression and systematically deny or limit access to goods, services, and opportunities in society, including the right to safety and protection (Bonilla-Silva 1997; Crenshaw 1991; Sidanius & Pratto 2001). The criminal justice system is widely understood to be a system that creates and reinforces social hierarchies based on race, gender, and poverty (Alexander 2012; Frohmann 1997; Richie 2012). Crenshaw (1991) argued that the intersectional identities of women of Color place them at particular risk to be victims of crime, and then to be re-victimized by the criminal justice system. Thus, the failure to protect women, particularly African American women, many of whom were poor, by not investigating crimes committed against them—and to argue that this inaction was not about gender, race, and class—is wholly consistent with structural, intersectional analysis of the criminal justice system.

Taken together, the results of this study suggest that this “new” problem of untested rape kits is actually a rather old problem: untested SAKs have been accumulating in police storage facilities for decades because the criminal justice system has not dedicated sufficient resources to combat sexual violence, which is due in no small part to entrenched gender, race, and class biases about rape and credibility of rape victims. Rape kit evidence collection and DNA testing were considered a potential solution to the problem of police and prosecutorial inaction by providing tangible evidence that could support prosecution (Martin 2005). Systematic collection and use of DNA evidence could contribute to increased prosecution, but generally has not thus far, as Du Mont and White (2007, 2013) have noted that forensic evidence is infrequently utilized by the criminal justice system in sexual assault cases. Rape survivors endure an invasive, often degrading, full-body medical exam (Bumiller 2009; Corrigan 2013; Quinlan et al. 2009; cf: Campbell et al. 2008, 2013; Du Mont et al. 2014; Fehler-Cabral et al. 2011)—and often nothing happens with the evidence collected. As cities throughout the United States are uncovering large stockpiles of untested SAKs, these discoveries re-open conversation about how forensic DNA testing could help sexual assault investigations and prosecutions. However, strategies that seek to strengthen carceral solutions to sexual assault have been critiqued by sociolegal theorists (Bernstein 2012; Bumiller 2009, 2014; Corrigan 2013), noting that such approaches have not solved the problem of the gender-based violence and indeed create new problems as feminist agendas become coopted by neoliberalism and conservative anti-crime advocates (Coker & Macquoid 2015; Whittier 2016). DNA evidence, in particular, strengthens the capacity of a surveillance state and carceral approaches (Lynch et al. 2008; Quinlan et al. 2009).

We cannot reconcile feminist carceral theory and anti-carceral theories, but we do highlight a core tenant in both sociolegal traditions—the importance of respecting and supporting survivors’ choices. As Bumiller (2009: 96, 131) argued, “Despite feminist-inspired reforms, many of which were explicitly designed to empower ‘victims’ to exercise more ‘choice’ or protect against retraumatization, the voices and desires of women are often muted by the more dominant discourse of the state … the opportunities for meaningful choice and autonomy remain elusive for victims”. The rape survivors in this study were most often poor, black women, and despite the historic abuse of African American women by the criminal justice system (Crenshaw 1991; Richie 2012), they sought legal recourse. They went to hospital emergency departments for post-assault medical care and consented to SAK collection, released the rape kit for DNA testing, and filed a police report. Some scholars have questioned whether these actions are indeed indicative of a desire for criminal justice system involvement because current North American systems of post-assault care specifically direct victims to these services (e.g., victims are told to have a medical forensic rape exam) (Bumiller 2009; Corrigan 2013; Parnis & Du Mont 2006; Quinlan et al. 2009). Empirical studies on this issue find that victims seek post-assault medical care for myriad reasons—for healthcare, for emotional support and advocacy, for forensic evidence collection, and/or for reporting to the police (Du Mont et al. 2009, 2014; Patterson & Campbell 2010).

In this action research project, the multidisciplinary team spent considerable time discussing the need to respect victims’ choices, particularly in light of the fact that we did not know with certainty why each of these ∼11,000 survivors sought a rape exam, consented to the collection and release of forensic evidence, and filed a police report. However, based on the information available about these cases, there were ample indications that for these rape survivors, they did indeed wish to pursue the prosecution of the crimes committed against them. The rape survivors in our workgroup emphasized that reporting to the criminal justice system may not be the choice for all victims—and it was not for many of them—but it was the choice of these survivors. Respecting their choices meant that we needed to understand why these kits had not been submitted for testing (the goal of this research), and to begin to reach out to survivors to ascertain their wishes and amplify their voices. As the project unfolded and we were able to connect with survivors, the majority indicated that their wish now was the same as it had been years ago: to report and prosecute the assault (see Campbell et al. Forthcoming). These were hard decisions for the Detroit community, decisions that indeed speak to broader theories of the state, but they are also decisions that affect the lives of individual survivors. This is not a study of rape victims who did not engage the criminal justice system—it is a study of survivors who sought legal remedies and were denied them. The choice whether to engage the criminal justice system and the meaning of justice is different for each survivor, and we make no claims in this article or in our action research project that a carceral approach to justice works for all survivors.

In addition to these conceptual caveats, we also acknowledge several limitations of this study with respect to its scope and methods. Our study identified the reasons why Detroit amassed so many untested SAKs, and given the uniqueness of Detroit among U.S. urban cities, our results may not generalize to other communities that have large numbers of untested rape kits. Detroit is a racially homogeneous city (83 percent black in the 2010 Census), with a high violent crime rate (highest rate in the nation in the 2010 UCR), and severe economic hardships. These factors are clearly related to the chronic resource scarcity documented in this study, which likely affected the magnitude of the problem in this city. While the scale of the problem in Detroit is not atypical (other urban cities have reported similar numbers of untested SAKs: 8,000–12,000 in Los Angeles, Memphis, and Houston) and our findings regarding police disengagement in sexual assault cases is well-documented in the literature, perhaps the reasons why kits are not tested are different in other cities. Research with multicity samples is warranted for a more complete understanding of why SAKs are not tested for DNA.

As a study of untested SAKs, our project focused on reported rape cases not pursued by the criminal justice system. Across-case analyses of the “not's” are a useful complement to more traditional between-case analyses that compare cases pursued to those not pursued; however, this means that we do not have information about SAKs that were tested and the rape cases that were arrested and prosecuted. In the years 1980–2009, most reported rapes in Detroit did not result in an arrest and most SAKs were not tested for DNA (see Campbell et al. 2017b), so what we studied was the most typical scenario in this community. In jurisdictions that have more variability in SAK testing rates, it would be useful to compare cases in which rape kits were and were not tested to assess whether indexes of victim credibility and cooperation were related to testing decisions. Regional-scale studies have found that extra-legal characteristics do affect whether SAKs are submitted for testing (Patterson & Campbell 2012; Shaw & Campbell 2013; Valentine et al. Forthcoming), which underscores the need for more research on how convictability concerns affect the utilization of DNA evidence.

With these caveats in mind, this study highlights that despite efforts to make SAK evidence collection more accessible and less traumatizing for rape victims (Department of Justice 2013; Du Mont et al. 2014; Martin 2005), the criminal justice system has not been utilizing this evidence to pursue sexual assault investigations and prosecutions. To those critical of carceral solutions, these results reinforce concerns that the state will not help survivors, so alternative meanings and methods of justice are necessary. For policy makers who support carceral solutions, the systemic neglect of DNA evidence is cause for action, and indeed, a growing number of states no longer allow rape kit testing to be a discretionary decision by the police. To date, nine states mandate testing previously unsubmitted SAKs and 16 mandate submission and testing of new/current cases (Joyful Heart Foundation 2017). However, the results of this study offer a cautionary tale about the long-term impact of such policy initiatives if the sole focus is ensuring kit testing without addressing the underlying reasons why so many SAKs are not tested in the first place. Mandating rape kit testing and providing federal funding for testing (e.g., Department of Justice Backlog Reduction Grants and Sexual Assault Kit Initiative Grants) might alleviate resource constraints law enforcement agencies have struggled with for decades (Lovrich et al. 2004; Strom & Hickman 2010), but these solutions do not necessarily address why police wouldn't test SAKs and investigate reported sexual assault cases. It is possible that if rape kit testing became a routine part of police practice, forensic testing results and CODIS hits may challenge law enforcement's long-standing biases about victim credibility by providing evidence that supports their accounts of the crime. Yet DNA evidence will not neutralize gender, race, and social class oppressions, and addressing these biases is fundamental to changing the criminal justice system response to sexual assault.