Evaluation of a Victim-Centered,Trauma-Informed Victim Notification Protocol for Untested Sexual Assault Kits (SAKs)

Sample

Over the 1-year time frame for the evaluation of the victim notification protocol, three multidisciplinary NRT meetings were convened. A total of 40 cases were presented by the prosecutor’s office to the NRT for discussion. This number reflects the maximum caseload Detroit stakeholders could manage so that each victim who was notified would have investigators and advocates ready to help them immediately if they decided to reengage with the criminal justice system (unless a delay was requested by the survivor). The decision to notify a relatively small number of victims was a deliberate choice on the part of the action research project team and the NRT. Certainly, more victims could have been notified within 1 year, but given current staffing levels in Detroit organizations, it would not have been possible to reopen more cases and actively work with more survivors on reinvestigation and prosecution (if survivors decided to reengage). Detroit stakeholders did not want to notify victims and then subject them to a (another) lengthy delay while waiting for police and prosecutors to have availability to pick up their cases. In addition, consistent with methodological recommendations in formative evaluation projects (Davidson, 2005), the team wanted to try the protocol with a relatively small number of cases to assess its strengths and limitations prior to wide-scale implementation.

All cases that were presented to the NRT for consideration had CODIS hits. In the vast majority of these cases, the NRT decided to notify the victims, but there were three instances in which the team unanimously decided not to contact the survivors. All three were domestic violence–related intimate partner sexual assaults and there were multiple indications in the case records that the survivors had not wanted to pursue prosecution years ago; in addition, there were specific safety concerns documented in the records and team members were concerned that recontacting the victims would jeopardize their safety. ³

Four cases were selected for notification at the discretion of the prosecutor’s office (i.e., these cases were not brought before the NRT for review). In all four cases, the prosecutors felt immediate notification was necessary based on the CODIS hit information (e.g., the CODIS hit was to an offender currently in prison, who was about to be released on parole, and the new DNA match could be relevant in the parole hearing). These four cases were presented to the NRT (after the fact) so that all members of the team were aware of what had happened; in all four cases, the NRT unanimously agreed that the victims should have been notified and that emergency action had been warranted. Therefore, the total sample of cases selected for notification was 41 (37 cases approved by the NRT + four direct notification decisions made by the prosecutor’s office).

Data Collection Procedures

It was not possible for the research team to observe the notifications (either the first or the second contact notifications) to collect independent data regarding what happened in these interactions. Practically, the physical safety of the survivors and notifying personnel was a primary concern, so the NRT did not want any non–law enforcement personnel on site. Legally, if the research team had been allowed to observe the notifications, we would have become party to the case and we could have been called to provide testimony in subsequent legal proceedings; though we could not and would not provide any information (per research confidentiality), the prosecutors were concerned that this would significantly complicate already complicated cases. ⁴ Because these were open, active legal cases, our institutional review board (IRB) did not want the data collection procedures to interfere in any way with case proceedings. Ethically, it was also not in the best interest of the survivors and their right to privacy to have an unknown outside party witnessing the notifications. For all these reasons, the research team, the NRT, and our IRB agreed that only proxy data collection was appropriate. Therefore, we developed procedures for training notifying staff to collect relevant data and for monitoring data completeness and quality.

To collect data regarding how the investigators found survivors for notification, each detective kept daily electronic tracking logs that recorded (a) date and time of each action taken, (b) type of action (e.g., database search for last known address/phone number, phone call, in-person visit to address), (c) number of times each action was taken (e.g., number of databases searched, number of calls made to each phone number), (d) the outcome of each action taken (e.g., phone disconnected, no one at residence when visited), and (e) next steps to be taken in an effort to find the victim (e.g., return to address at different time of day). These records were maintained for every action taken to find each victim, up to and including the point at which the investigator was able to make successful contact with the victim and verify that she or he had indeed found the correct person. Members of the evaluation team monitored these logs to ensure that they were completed in a timely and thorough manner.

To collect data about what happened during the first contact notification (i.e., once the investigators had established that they had found the correct person), the investigators kept electronic tracking logs that recorded (a) date, time, and personnel in attendance for each notification; (b) location of the notification (e.g., on the phone, at victim’s residence); (c) a narrative description of what occurred in the notification (e.g., information provided to the victim, questions asked by the victim); (d) narrative description of the victim’s emotional and physical demeanor during the notification, noting any changes in demeanor over the course of the interaction (including verbatim quotes from the survivors); (e) whether the victim agreed to a follow-up second meeting to discuss the case in more detail; and (f) overall reflections on the notification (e.g., what went well, issues to bring back to the NRT for discussion). The investigators were responsible for recording these data fields within 48 hr of each notification. Before the first notification, we trained the investigators on how to complete these logs, which emphasized the importance of writing behaviorally focused descriptions of what happened in the interactions, without attributions, labels, interpretations, or analysis. Investigators were specifically instructed to provide detailed descriptions of what they directly observed and what victims stated in their interactions. We reviewed/practiced what would and would not be considered appropriate data collection. For example, a log entry that stated “victim was mad” would not be acceptable; investigators needed to provide specific descriptions of behaviors and statements made by the victim (i.e., what did the victim say/do that indicated anger). The evaluation team monitored these logbooks throughout data collection and provided corrective feedback to ensure consistent high-quality data collection.

To collect data regarding what occurred during the second contact notification (i.e., the more detailed meeting with the investigators and the community-based advocates), the detectives continued to keep electronic tracking records, noting the same information as listed above for the first notification meeting, plus a new field regarding whether victims wanted to reengage with the criminal justice system for the reinvestigation and possible prosecution of their cases. Again, investigators were trained prior to starting data collection on the importance of logging detailed, behaviorally specific descriptions of these interactions. These records were also monitored throughout the evaluation to ensure timely completion and data quality.

Data Analysis Plan

Given the exploratory nature of this study, we selected Miles, Huberman, and Saldaña’s (2014) approach for coding and analyzing the data, which is a rigorous, multiphase process for identifying regularities within qualitative/mixed methods data. ⁵ In the first phase of analysis (data condensation), one research assistant computed descriptive statistics (frequencies, means, medians) for all quantitative data (e.g., number of strategies used to find each victim). For the qualitative data (e.g., descriptions of victims’ emotional reactions to notification), the research assistant reviewed the logbooks and then tagged and labeled content. For example, from a logbook entry that said, “After I [the investigator] told her it was about the report she got tears in her eyes . . . she stated she saw the news on the kits and thought hers was thrown out,” the research assistant applied the tag “victim cried.” Other common tags included “victim surprised,” “victim showed no emotion,” and “victim visibly angry.”

Also in this first phase of data analysis, the research assistant and the project’s principal investigator (PI) reviewed these descriptive tags to identify higher order codes. We identified three common themes/types of victims’ emotional reactions to the notification. The code “strong negative reaction” was applied to cases in which victims raised their voices to the investigators, exhibited physical signs of anger (e.g., slammed doors, stomping feet, harsh gestures), stated that they felt physically ill discussing the matter, and/or expressed that they did not want to discuss the assault (reasons for that varied, such as having current health concerns that survivors felt were more pressing in their lives right now than a prior sexual assault from years ago). The code “strong positive reaction” was applied to cases in which victims directly stated that they were happy to have been contacted, were pleased investigators had found them, expressed gratitude that the case was being reexamined; these victims typically welcomed investigators into their homes, offered them hospitality (e.g., something to drink), and sometimes hugged notifying staff. It was not uncommon that victims in this group cried during the notification discussions, and they told investigators that they were crying because they were happy something was finally happening in their cases. The code “absence of a strong negative or positive reaction” was applied to cases in which the victims did not have much by way of a demonstrable reaction to the notification. In these cases, victims were wary and hesitant to answer investigators’ questions, did not show anger or happiness, and typically spoke in flat/monotone voices without emotional inflection. Investigators often characterized these survivors’ reactions as “reserved,” “calm,” and “matter-of-fact.” Some victims in this group did cry during notification, but in contrast to those with strong positive reactions, the crying was brief and survivors did not state why they were crying. The two coders reviewed the data multiple times to verify consistent application of these three higher order codes, paying particular attention to instances in which a particular tag (e.g., “victim cried”) occurred in multiple higher order codes (e.g., different reasons for crying, different duration of crying).

In the second phase of analysis (data display), the research assistants and the PI worked together to create data display matrices to explore associations within the data. We assessed whether there were discernable patterns between the effort required to find victims (low, medium, high effort), victims’ emotional reactions to the notification (strong positive, strong negative, absence of strong emotional reactions), and victims’ decisions whether to reengage with the criminal justice system for reinvestigation and possible prosecution. We also explored how these features of the notification related to three variables that the literature has indicated could have bearing on victims’ reactions, namely, how much time had elapsed between the assault and notification (<9 years, 9 years [median/M], >9 years), victims’ age at the time of the assault (<16 years old [age of consent in Michigan], 16-24 years old, >24 years old), ⁶ and victim–offender relationship (stranger vs. known perpetrator). We created data display matrices “crossing” these variables to look for patterns in the data (Evergreen, 2014; Henderson & Segal, 2013), and supplemented these visual analysis methods with nonparametric statistical tests, given the small sample size in this project (Kitchen, 2009). Associations between two ordinal variables were evaluated with the gamma (G) statistic, those between nominal and ordinal variables with the eta (η) statistic, and those between two nominal variables with the phi (r_φ) statistic (Garson, 2012; Pett, 1997; Siegel & Castellan, 1988).

Finally, the third phase of analysis (drawing and verifying conclusions) addresses the trustworthiness of the conclusions drawn about the data. Consistent with the standards outlined by Lincoln and Guba (1985) for the use of qualitative methods in evaluation projects, we used multiple strategies to assess the credibility, transferability, dependability, and confirmability of our work (see Campbell, Fehler-Cabral, et al., 2015, for details). Briefly, the research team had prolonged engagement with all organizations represented in the NRT, conducted ongoing observation of the NRT meetings and all other action research project team meetings for more than 4 years, triangulated data sources, conducted member checks of our findings with members of the action research project team, practiced memoing throughout the project, and maintained audit trails throughout data collection and analysis.

Is It Possible to Find Victims Years After They Reported the Assault to the Police?

Overall, of the 41 cases selected for notification, eight notifications were still pending at the time the evaluation concluded. Of the 33 cases that were resolved during the evaluation period, the investigators were able to locate and notify 31 victims (95% find rate). ⁷ Thirty of these 31 victims were female (97%); one was male (3%). Most were African American (n = 24, 77%), six were White/Caucasian (19%), and one was Hispanic (3%). On average, these victims were 29 years old at the time they were sexually assaulted (SD = 12.18 years), with a range of 12-54 years old. At the time of notification, they were on average 38 years old (SD = 12.20 years), range 20-62 years old. On average, the notifications were made 9 years after the victim had been sexually assaulted and made a police report. Most of these survivors had been sexually assaulted by a stranger (n = 24, 77%) and seven had been raped by someone known to them (23%). Most of these notifications were made by female investigators affiliated with the prosecutor’s office (79%); there were no differences in victims’ emotional reactions or decisions to reengage with the criminal justice system as a function of the gender of the notifying investigator (η = .04, non-significant [NS], and r_φ = .11, NS, respectively).

To find these victims years after the assault, investigators first searched three law enforcement databases (LEIN—Law Enforcement Information Network; TLO Online Investigative Services; NEXIS/LEXIS) to find the victims’ last known addresses and phone numbers, and then initiated contact to verify whether they had in fact found the right person. On average, investigators made three contact attempts (range = 1-16 contacts) on each case, leading up to and including the contact in which they confirmed they had found the correct individual. Investigators averaged two in-person visits per case (range = 0-6 attempts). Sometimes investigators were successful in locating victims on the first attempt, but typically, investigators had to return to the same home on multiple occasions before they were able to speak with the victim, or had to visit a series of addresses, ruling them out one by one. To protect victims’ privacy and safety, the investigators never left any information disclosing that the reason they were trying to reach the victim was in reference to a sexual assault case. For victims whose most current addresses were outside the Detroit metro area or out of state, phone contact was the sole strategy used. Investigators averaged one phone call per case (range = 0-12 calls).

Although there was considerable variability in the number of in-person and phone contact attempts that were necessary before investigators were successful in finding each victim (zero to six and zero to 12, respectively), in most cases, survivors could be located with relatively low investigational effort: 65% (n = 20) were found by conducting database searches, plus zero to four phone calls, plus zero to one in-person visit to one address. The remaining cases required more extensive investigational effort to locate the victims: 19% (n = 6) required an additional one to two in-person visits (to one to two addresses); and 16% (n = 5) required extensive efforts ( six to 12 phone calls and four to six in-person visits to multiple addresses). There was no significant relationship between the effort necessary to find the victim (low, medium, high) and survivors’ emotional reactions to notification and their decisions regarding reengagement with the criminal justice system (G = .006, NS, and η = .19, NS, respectively). In other words, survivors who were difficult to find were no more or less likely to have a strong emotional reaction to notification, and were no more or less likely to reengage with the criminal justice system for the reinvestigation and possible prosecution of their cases.

How Did Victims React When Notified That Their Rape Kit Had Finally Been Tested?

We identified three common themes/types of victims’ emotional reactions to notification (see “Method” section): strong negative reactions (e.g., victims raised their voices to the investigators, exhibited physical signs of anger, and/or expressed that they did not want to discuss the assault), strong positive reactions (e.g., victims directly stated that they were happy to have been contacted, were pleased investigators had found them, expressed gratitude that the case was being reexamined), and absence of strong emotional reactions (e.g., victims were wary and hesitant to answer investigators’ questions, did not show anger or happiness, and typically spoke in flat/monotone voices without emotional inflection).

We examined whether victims’ emotional reactions varied as a function of time since the assault to the notification, victim’s age at time of the assault, and victim–offender relationship. In Figure 2, the rows represent the time since the assault; the columns, victims’ age at time of the assault; and the cell divisions, victim–offender relationship. The symbols inside the cells represent the victims’ emotional reactions: strongly positive (plus sign), strongly negative (strikethrough sign), absence of a strong reaction (open circle). The number of symbols in each cell reflects how many victims had that pattern of associations; for example, in the upper most left cell, there was only one survivor who was assaulted when she or he was less than 16 and was notified less than 9 years after the assault; this individual had been sexually assaulted by a stranger, and she or he exhibited a positive reaction to the notification.

OPEN IN VIEWER

Figure 2.

Victims’ emotional reactions to notification, as a function of time since assault, victim age at time of assault, and victim–offender relationship.

As seen in Figure 2, how victims reacted was associated with how long ago the assault occurred. On average, 9 years had passed since the assault and notifications, and 29% of victims who were notified more than 9 years after the assault had a strong negative reaction to the notification. By comparison, only 8% of the victims notified at the 9-year point and 0% of the victims notified less than 9 years after the assault had strong negative reactions. The longer the period of time between when the assault occurred and when the kit was tested and the victim was notified, the more likely it was that a survivor had a negative reaction to the notification (G = −.52, p < .05). Strong negative reactions were also somewhat more common among victims who had been 16-24 at the time of the assault (23%, compared with 13% of victims above 24 years old and 0% of victims below 16), but this effect was not significant in the nonparametric tests (G = −.22, NS). Victims’ emotional reactions to the notification did not vary as a function of victim–offender relationship. Those assaulted by known perpetrators did not have different emotional reactions to the notification as those who had been raped by a stranger (14% of known perpetrators had strong negative reactions vs. 16% for stranger perpetrators; η = .01, NS). However, only seven of the notifications were made in cases of nonstranger-perpetrated sexual assaults, so there may not be enough cases in this evaluation to detect a pattern.

Did Victims Decide to Reengage With the Criminal Justice System?

In the first contact meeting, the investigators asked victims whether they would like to have a follow-up second meeting to discuss the case in more detail, review options, and meet with a community-based advocate. In three cases, it became clear at the first contact that there was no need for a second follow-up meeting: In one case, the statute of limitations had already expired ⁸ ; in the second case, the CODIS hit turned out to be a match to the victim’s consensual partner (not the assailant), so there was no legal case (as of yet) to be pursued; in the third case, the victim was in a group home for individuals with serious mental and physical disabilities, and she had no memory of the assault (and so no legal case could be considered). In all three instances, the investigators provided victims with their contact information (if they had any follow-up questions or concerns) and a booklet of community resources. Given the circumstances of these three cases, the sample size for the evaluation shifted from N = 31 to N = 28 (i.e., there were only 28 cases in which a second contact was possible and legal action could be considered).

Eighteen of the 28 victims who had first contact notifications (and there was reason to have a follow-up meeting) had a secondary meeting (64%). At the second follow-up meeting, the investigator(s) and community-based advocates (all were females) worked together to explain the next steps in the investigation and ascertained victims’ willingness to participate in that process. Overall, 16 of the 28 victims decided that they did want to have continued participation in the investigation and possible prosecution of their cases (57%). As might be expected, victims’ emotional reactions during the notification were related to their willingness to engage in further contact with the criminal justice system. All victims who had a strong negative emotional reaction did not want any further contact (100% of those who had a negative emotional reaction). Most victims who had a strong positive reaction were more likely to agree to further contact with the criminal justice system (78% of those who had a positive emotional reaction). Survivors who did not have a strong positive or negative emotional reaction also typically decided that they did want continued contact with the criminal justice system (64% of victims who did not have a strong emotional reaction; η = .55, p < .05).

We also examined how victims’ willingness to have continued contact with the criminal justice system was related to the time since the assault, the victims’ age at the time of the assault, and victim–offender relationship (see Figure 3). Time since the assault was associated with victims’ willingness to have continued contact with the criminal justice system. Specifically, 62% of victims who had been assaulted more than 9 years prior to the notification were unwilling to engage further with the investigator(s), whereas only 18% of victims who were notified 9 years after the assault and 50% victims notified less than 9 years after the assault were unwilling to continue their participation (η = .41, p < .05). In other words, the more time that had elapsed since the assault and the notification, victims were less willing to reengage with the criminal justice system. Victims who were between 16 and 24 years old at the time of the assault were somewhat less willing to interact further with the investigator(s) as compared with victims who were older or younger at the time of the assault (62% unwilling for 16- to 24-year-olds, 31% unwilling for victims above the age of 24, 0% for victims below 16 years old), but this was not significant in nonparametric tests (η = .38, p = .11). Finally, the nature of the relationship between the victim and the perpetrator was related to victims’ willingness to participate in the reopening of their cases. Most of the victims assaulted by someone they knew (86%) were unwilling to engage further, whereas only 29% of victims assaulted by a stranger were unwilling to have continued contact (r_φ = −.50, p < .01).

OPEN IN VIEWER

Figure 3.

Victims’ decisions to reengage with the criminal justice system, as a function of time since assault, victim age at time of assault, and victim–offender relationship.