A Decade of Decision Making: Prosecutorial Decision Making in Sexual Assault Cases

Abstract

In the United States, it is estimated that fewer than 30% of sexual assaults are reported to law enforcement, less than 5% will result in an arrest, and approximately 3% will result in a felony conviction. The present study examines a census of sexual assault reports from 2012 to 2020 in a large police department in the Midwest region of the United States, considering only those cases presented to the district attorney’s office for consideration (n = 700). Victim characteristics, suspect characteristics, and incident characteristics were examined as relating to a prosecutor accepting a case for charges. Further, a series of “ideal victim” characteristics were identified and used to create an additive scale based on prior research. The most significant predictors of prosecutor acceptance in the full sample were a lack of “date rape” drugs, continued cooperation from the victim, the suspect having a prior arrest for sexual assault, and the assault being reported within 24 h of occurring. A separate model was conducted for only those cases with continued cooperation from the victim. The researchers discuss the implications of these significant factors and suggest training to change false perceptions of victims among criminal justice officials and to increase victim reporting/cooperation in cases of sexual assault.

1. Introduction

Many people have the misconception that once an arrest happens, justice has been served. However, this is only the first step. Next, prosecutors play an important role in deciding whether a case will move forward to court or be dismissed. This role is one of the most important aspects of the criminal justice process; however, it is also one of the least understood by the general public (O’Neal et al. 2015). In the United States, prosecutors hold the power to administer justice. They decide who will be charged, what the charges will be, if plea bargaining will be an option, and possible sentences for the offender (Morabito et al. 2019; O’Neal et al. 2015; Spohn and Tellis 2012). To make these decisions, prosecutors must weigh multiple factors to determine which direction to go. In sexual assault cases, these decisions are even more complicated, as they tend to be “he said, she said” arguments. Prosecutors are undeniably aware that the case may be presented in front of a jury of citizens who may hold stereotypical misperceptions about rape and rape victims (Taylor and Joudo Larsen 2005). The current study will build on existing research regarding the considerations of prosecutors, both legal and extralegal when determining whether to pursue charges in a sexual assault case.

1.1. Review of Literature

In determining whether to pursue a sexual assault case, prosecutors consider legal evidence such as the strength of physical evidence against the defendant (Henning and Feder 2005; Lapsey et al. 2022; Spohn and Holleran 2001), the seriousness of the crime itself (Alderden and Ullman 2012; Henning and Feder 2005; Spohn and Holleran 2001; Spohn and Tellis 2019), if a weapon was present (Beichner and Spohn 2005; Chandler and Torney 1981; Lapsey et al. 2022; Spohn and Holleran 2001), the amount of harm the victim suffered, or if the victim resisted their attacker (Spohn et al. 2001; Spohn and Horney 1993; Spohn and Spears 1996; Taylor 2022). Extralegal factors such as the victim’s credibility (Alderden and Ullman 2012; Beichner and Spohn 2005; Campbell et al. 2009; Frohmann 1991; Spears and Spohn 1997), victim’s age and occupation (McCahill et al. 1979; Spears and Spohn 1997; Spohn and Tellis 2019), a behavioral activity that could be considered risky such as drugs, drinking, sexual history, etc. (Beichner and Spohn 2005; Bryden and Lengnick 1997; McCahill et al. 1979; Spears and Spohn 1997; Spohn and Holleran 2001; Spohn and Horney 1993; Spohn and Spears 1996; Spohn and Tellis 2019; Spohn et al. 2001, 2002), and the victim’s moral character (Beichner and Spohn 2005; Bryden and Lengnick 1997; McCahill et al. 1979; Spears and Spohn 1997; Spohn and Tellis 2019) could all be called into question. These factors are weighed in the decision-making process by prosecutors to determine the strength and certainty of a case (Albonetti 1987; Pattavina et al. 2021; Spohn et al. 2001). Prosecutors will exercise discretion by reducing or rejecting charges against the defendant in order to avoid uncertainty in court (Albonetti 1987; Spohn et al. 2001). Jurors want to view victims of sexual assault as sympathetic, credible, hurt from the act, and the case to be clear-cut with no alternative theories to fit the narrative (Estrich 1987; Spohn et al. 2001). Jurors want this image of a “real rape” victim, as notably stated by Estrich (1987), and prosecutors aim to present this in court.

This area of criminal justice decision-making has been analyzed from the perspective of focal concerns (Steffensmeier et al. 1998). Unfortunately, prosecutors are often presented cases with only limited information on the victim or offender. To make decisions quickly, they adopt a “perceptual shorthand” using anecdotal experiences and stereotypical perceptions to informally determine the likelihood of a case being successful (Albonetti 1987). These considerations come in three areas: offender blameworthiness, protecting the community, and concerns about practical consequences. While originally oriented to explain judicial sentencing decisions, this theory has recently been applied to prosecutor decision-making (Spohn and Tellis 2019). Using these informal considerations of legal and extralegal factors, prosecutors do not approve charges for a vast number of sexual assault cases (Albonetti 1987; Alderden and Ullman 2012). These cases are deemed to have a high likelihood of uncertainty and are thrown out of the running due to certain factors, including, but not limited to, unreliable credibility from the victim, weak physical evidence, and other discrepancies. This leads to the suspect walking away without any reprimand while the victim receives no justice (Alderden and Ullman 2012; Morabito et al. 2019; Spohn and Tellis 2019).

In Campbell et al. (2009), the authors investigated a large Midwestern county, including urban, suburban, and rural areas, with a sexual assault nurse examiner (SANE) program to determine what factors may affect the outcome of 137 adult sexual assault case investigations and prosecutions. Reports were obtained from September to December 1999. After analyzing the data, some factors that were significant in predicting a higher-level outcome, such as a greater chance of conviction in court, for prosecution in an adult sexual assault case were the victim’s age (between 18 and 21), when the victim was assaulted, the victim/offender relationship, and positive DNA evidence. These factors could be used to show physical evidence of the assault while also determining the severity of the crime and the victim’s overall credibility. However, some factors were also significant in predicting a decreased chance of obtaining higher-level outcomes: cases where the victim had consumed drugs and/or alcohol and when there was a large time gap between the assault and the victim obtaining a medical forensic exam. Prosecutors may view or feel that a jury may see these actions as related to the victim’s credibility/moral character. Further, the deterioration of physical evidence would not provide the prosecution with a strong case.

Beichner and Spohn (2005, p. 488) examined prosecution charging decisions across two different jurisdictions: Kansas City, Missouri, and Miami, Florida. Kansas City held a specialized unit that focused on sexual assault cases, while Miami did not have a specific unit. During interviews with the prosecutors in these areas, it was revealed that both jurisdictions base their decisions on the standard of proof beyond a reasonable doubt, as required for a conviction in criminal cases. Both jurisdictions used a combination of legal and extralegal variables when making a decision. During these interviews, prosecutors revealed that jurors have a hard time relating to specific victims, so the ideal victim to have is a “pristine victim upon whom the jury can place no blame”. The results indicated minor differences between the specialized unit in Kansas City and the nonspecialized office in Miami; only 47% of the cases presented to prosecutors in Miami and 49% of the cases in Kansas City resulted in charges being filed. Another commonality between both jurisdictions was the significance of extralegal victim characteristics. In Kansas City, if someone were to engage in risk-taking behavior such as hitchhiking, using alcohol and/or drugs, walking alone at night, inviting the suspect into their residence, or willingly accompanying the suspect to the suspect’s residence, the chances of the prosecution approving charges decreased significantly. In Miami, any information that questioned or tarnished the victim’s moral character, such as out-of-wedlock pregnancy, a prior criminal record, patterns of alcohol and/or drug abuse, or prior sexual activity, would cause a significant decrease in charges being approved. Despite being two different jurisdictions and almost 1500 miles apart, both regions showed a specific interest in the victim’s credibility when making the decision to prosecute sexual assault cases.

Alderden and Ullman (2012) examined 465 sexual assault cases from a large Midwestern police department in 2003 to identify factors that could impact sexual assault case processing, focusing on those that influenced first police officers’ and later prosecutors’ decisions about legitimacy and conviction surrounding a sexual assault case. When examining prosecutor approval of charges for a case, three factors were considered significant: victim/suspect relationship, victim discrepancies, and victim injury. Cases that involved intimate partners decreased the odds of gaining approval for felony charges by 98% compared with cases involving strangers. Discrepancies in the victim’s statement were seen as a lack of truthfulness from the victim and resulted in a 90% decrease in approval of felony charges. The last factor, victim injury, showed an eightfold increase in approval for felony charges when the victim sustained injuries in the assault. While most of the cases in this study were presented to the prosecution for approval of felony charges, less than 10% were approved.

In an attempt to identify the chance of prosecution and the factors that influence the decision to charge in intimate partner sexual assault (IPSA) cases, O’Neal et al. (2015) analyzed 47 IPSA cases that were referred from the Los Angeles Police Department to the LA prosecution office. While certain factors, such as a lack of evidence, were often claimed as justifiable reasons for dismissals, other factors were discovered to play a role in affecting a prosecutor’s judgment call. Prosecutors only moved forward with charges if conviction was likely and dismissed cases where the chances of conviction were slim. Cases that held strong physical evidence, such as witnesses, sexual assault exams, injuries to the victim, and/or evidence from the crime scene, were found to be good indicators for pursuing a conviction. Some of the factors that negatively impacted the chance of prosecution were a lack of cooperation from the victim, questions concerning the victim’s credibility, and/or if there were no other additional domestic violence factors involved. In this study, all cases filed by the prosecutor were devoid of victims engaging in potentially questionable or risk-taking behaviors that could be misconstrued by a jury, which further indicates the importance of victim credibility in the decision process.

When evaluating the relationship between the suspect and the victim, prior research does not show a clear relationship with prosecution decision making. Some research says that a stranger’s sexual assault is investigated more thoroughly, is less likely to be rejected by the prosecutor, and results in the filing of more serious charges in comparison to an acquaintance or intimate partner assault (Beichner and Spohn 2005; McCahill et al. 1979; Spohn and Holleran 2001; Taylor 2022). However, other studies, such as Spohn et al. (2001), have discovered that sexual assault cases involving intimate partners or acquaintances were more likely to be prosecuted when compared to sexual assault cases where the suspect was a stranger. This could potentially be explained by the added benefit of being able to locate the suspect more readily for law enforcement officials when the offender is known (Spencer and Stith 2020).

Spohn and Tellis (2019), examining police and prosecutor decision-making in 650 Los Angeles-based cases, found that victim cooperation is a key factor in prosecutors deciding to charge a case, along with whether the victim did not have a motive to lie. These were the only significant variables when examining prosecutorial decisions after a suspect was arrested. When expanding to cases that were presented to the prosecutor before an arrest as well as after, prosecutors were significantly influenced by a victim’s risky behavior, delayed reporting (within an hour of the assault occurring), victim cooperation, and the seriousness of the charge presented. This study highlights the importance of looking at all sexual assaults reported and forwarded to a prosecutor, not just those that have resulted in an arrest, especially considering the private nature of the sexual assault and the decreased likelihood of officers apprehending the suspect during or immediately following the incident.

Pattavina et al.’s (2021, p. 1065) study of rapes reported to six different agencies also highlights the importance of victim cooperation. The final part of the authors’ analysis examined the predictors of a case being exceptionally cleared, or “officially closed or ended without prosecution”. The most significant predictor of cases being exceptionally cleared is if the offense was between intimate partners. Other factors that were significantly related to a case not moving forward with charges were if the offender/victim were acquaintances, combined with the previous finding indicating that stranger cases were most likely to proceed to prosecution, the victim being labeled as engaging in risky behavior or seen as having credibility issues, and being older.

These complicated dynamics are also apparent during pretrial investigations (Ruback and Ivie 1988). Compared to when rape is completed by a stranger, rapes completed by a spouse or other romantic partner are less likely to have witnesses (Felson et al. 1999), which can be an important case consideration for law enforcement presenting a case to prosecutors (Pattavina et al. 2021). Rapes involving strangers are also more likely to be reported by a third party compared to other relationship dynamics (Felson et al. 1999).

Rapes with known offenders are more likely to be dropped before a trial occurs (Miethe 1987). More recent findings suggest women who knew their offender in some way were significantly less likely to cooperate with a police investigation (Felson and Lantz 2015). Further complications for law enforcement could include victims withdrawing their statements, failing to arrive for their interviews, or wishing the case to be discontinued (Spohn and Tellis 2012; Spohn et al. 2001). Koss et al. (1988) report that approximately one-fifth of victims of stranger rape report the crime to the police, while only 1.7% of victims of acquaintance rape report the crime to the police.

Further, victims who experience mental health issues after their assault are less likely to continue with the investigation (Ullman and Siegel 1993). Ullman and Siegel (1993) found that fear and anxiety were more present in victims of stranger rape, resulting in a higher fear of retaliation and a higher likelihood of being uncooperative in an investigation. These factors all weigh on a victim when deciding whether to report their story, along with feelings of shame/embarrassment, retaliation from others, a belief that rape is a minor incident, mistreatment or being discredited by law enforcement officials, and many other factors (Bachman 1998; Herman 2003).

1.2. “Ideal Victim” Characteristics

When the public thinks of the term “victim”, they often hold prior beliefs and ideas in their minds about what a victim should be. Christie (1986) defined an “ideal victim” by five characteristics: they are weak; the individual was performing a respectable errand (such as caring for a relative); they were in a suitable location where no blame could be placed upon them; the offender was stronger and larger than the victim; and the offender had no prior relationship with the victim. Christie (1986) further added that a condition or sixth characteristic of an “ideal victim” is that they are strong enough to make their case known and successfully achieve the title of being an “ideal victim”. To be seen as an “ideal victim” by Christie’s definition, an individual must have all five characteristics present in their situation.

Using Christie’s (1986, p. 19) parameters, an “ideal victim” of sexual assault is “the young virginal woman on her way home from visiting sick relatives, severely beaten or threatened….”. What is interesting about this example is that Christie’s definition of an “ideal victim” is also considered the least common sexual assault victim (Christie 1986; Taylor 2022). Ussery (2022) reported that using this characterization of a victim excludes all intimate partner relationships from gaining “ideal victim” status, even though over half of sexual assault cases involve someone the victim knows. Under this definition, sex workers would also be ineligible for status as an “ideal victim”.

As applied to decision-making in sexual assault cases, multiple studies have shown that prosecutors are less likely to move forward with charges if the victim was partaking in “risk-taking” behavior, such as using drugs, drinking, going somewhere alone with their offender (Grubb and Turner 2012; McCahill et al. 1979; Nelson and Amir 1975; Tellis and Spohn 2008), knew their offender or was acquainted with them in some way (Albonetti 1987; Campbell et al. 2009; Simon 1996; Stanko 1981), or had a questionable moral character (Larcombe 2002; McCahill et al. 1979; Spohn et al. 2002). Using Christie’s parameters as a guide for identifying an “ideal victim” shows that much of the time, an individual will not receive this victim status.

When not receiving victim status, individuals form a distrust of the criminal justice system since they do not meet the expectations of what Estrich (1987) deemed a “real rape”. Estrich (1987) explained that there are two distinct forms of rape: aggravated and simple. Cases involving aggravated rape include specific factors such as a stranger and/or multiple assailants, the threat of a weapon/violence, and resistance markers. Simple rape cases lack these characteristics. Estrich (1987) indicated that the criminal justice system identified and accepted aggravated rape cases as “real rape” due to markers of resistance being present. A victim cannot only verbally say “no” to their attacker but must also display evidence of true resistance from their assailant. Having these markers present solidifies the case as a “real rape” which results in the criminal justice system believing and trusting the victim. The absence of these markers leads law enforcement officials to distrust, scrutinize, and/or blame the victim.

While it is not possible to conduct research on active juries, research on mock juries’ perceptions of sexual assault has uncovered long-standing misperceptions that juries hold regarding the “ideal victim” of sexual assault. Kalven and Zeisel (1971), completing one of the premier studies on the topic, noted that juries often placed at least some blame for sexual assault on the victim, with juries more likely to convict in “aggravated rape” cases that include more stereotypical factors such as a stranger offender, clear violence, or multiple offenders. Variations of this finding have persisted through the decades (e.g., Ellison and Munro 2009; Whatley 1996). A recent meta-analysis by Dinos et al. (2015) found that rape myth acceptance (RMA) by mock jury members was significantly related to decision-making. With nine studies ranging from 1984–2011, five studies found a medium/large effect of RMA on decision-making in sexual assault cases. Rape myths can be closely aligned with perceptions of the “ideal victim”, as they both involve stereotypical and often unrealistic impressions of rape and rape victims. How the jury will respond to a case is a critical consideration for a prosecutor.

Past research has shown that prosecutors consider multiple factors besides legal evidence when deciding how to move forward with a sexual assault case. While evidence has depicted many options that could play a significant role in the criminal justice process, one common element depicted is victim credibility (Alderden and Ullman 2012; Beichner and Spohn 2005; Campbell et al. 2009; Frohmann 1991; Spears and Spohn 1997; Spohn et al. 2002). Prosecutors give considerable weight to this factor, as it is highly correlated with the likelihood of conviction. If the victim’s credibility is called into question, the case may begin to fall apart. There have been fewer investigations of the impact of other ‘ideal victim’ characteristics in this decision-making process.

2. Materials and Methods

The present study provides an updated and more extensive investigation into prosecutor decision-making in sexual assault cases. Data collection covered a nine-year period of sexual assault reports from a large midwestern city. This project was initiated by the unit of study with an interest in exploring changing trends in adult sexual assault reports. The primary researcher met with the unit commander, and the project was supported by the Chief of Police. The primary researcher’s institution approved this project as a secondary data analysis since there was no interaction with victims or offenders.

Unlike prior studies in this area, all sexual assault reports during the study period were analyzed, not just those with certain characteristics, such as a completed sexual assault nurse examiner’s kit, or those that were handled by a specialized prosecution unit. The police department under study did have a separate sexual assault unit specifically for adult crimes. All reports of sexual assault were either initiated in the unit after a victim called in or were referred to the unit after a patrol officer took the initial report. “Ideal victim” characteristics are examined in each case to determine if the presence of these, especially the presence of more characteristics, is related to an increased likelihood of prosecutors accepting a case. Further, we explicitly examine the importance of victim cooperation in the decision to pursue charges.

2.1. Sample

Data for the present study was collected from a large police department in the Midwest region of the United States. The police department had a specialized unit for adult sex crimes. Researchers spent 18 months at the department headquarters reviewing all reported adult cases designated as rape, sodomy, or sexual assault.1 A census of adult reports, where both the offender and victim were at least 17 years of age, were collected for the years 2012–2020.2 Generally, the department classified rape cases as non-consensual completed or attempted penetration by a penis or object (n = 485) and sodomy as non-consensual oral penetration by a penis or digital/oral penetration of the vagina or anus (n = 136). It is important to note that the Federal Bureau of Investigation, the major federal law enforcement arm of the United States, has overlapping definitions of rape and sodomy. For this reason, these categories of assault will not be looked at separately, as cases could often fit into either definition. Lastly, sexual assault was a catch-all category of unwanted sexual touching under clothing (n = 79). This may include penetration, but it also includes attempted penetration or contact without penetration. This category of assault was less common in the data set, likely due to significantly decreased reporting rates for sexual assaults that do not result in injury (Chen and Ullman 2010; Felson and Paré 2005). Further, attempted rape or sodomy with no penetration may be counted in any of the categories. For these reasons, reports were classified based on the type of physical contact.

The final data set consisted of 2587 reported cases. As this research focuses solely on prosecutor decision-making, only cases that were referred to the prosecutor by detectives will be examined. These cases could have been referred pre- or post-arrest (Spohn and Tellis 2019).3 This resulted in a sample of 700 cases over 9 years, or less than one-third of reported cases. As data collection ended in 2021, cases reported in 2020 were not as common in the dataset as investigations were ongoing by detectives and had not yet reached the point of being ready for the district attorney.

2.2. Variables

Variable coding was based on information available in individual case files. Initial reports are most commonly taken by patrol officers who respond to calls, not by specialized detectives. Detectives are then assigned the case and will follow up with victims, witnesses, etc. Information was limited to these reports. Researchers did not make any assumptions about the cases. For example, if no mention of substances was made in the report but the assault happened at a location that served alcohol, researchers would not assume that the victim or suspect were drinking alcohol. The first month of data collection was a collaborative process, with researchers reading a case together, coding separately, and then reviewing. This process helped achieve a general understanding of how certain variables would be coded. After this initial process, researchers coded individually but regularly discussed any ambiguous or unclear variables to reach a consensus.

The dependent variable for this analysis is prosecutor acceptance. All cases in the sample had been forwarded to the district attorney’s office for review by sex crime detectives. If a case was not forwarded by detectives, it was not included in the analysis. This was a dichotomous variable, indicating whether the district attorney filed charges in the case or declined. This is not a measure of if the suspect took a plea bargain or was convicted.

Independent variables are abundant and can be broken into three categories: incident characteristics, victim characteristics, and suspect characteristics. These were based on what was widely available in reports—those that would not result in a large amount of missing data. They were then divided based on what the researchers felt fit into Christie’s (1986) elements of the “ideal victim” and those that did not. Descriptive statistics for all variables, as well as bivariate correlations with the dependent variable, can be seen in

Table 1. Descriptive statistics with relationship to prosecutor acceptance (n = 700).

Incident Characteristics	n (M)	% (SD)	Min/Max	χ2/t (p-Value)
Assault type			0–4	31.067 (<0.001)
Multiple types of penetration	133	19.00
Anal penetration	35	3.57
Vaginal penetration	389	55.57
Oral penetration	25	5.00
No penetration	118	19.00
Victim/offender relationship			0–2	21.757 (<0.001)
Current or former romantic partners	195	27.86
Other known offender	290	41.43
Strangers	185	26.43
Location			0–2	9.543 (0.008)
Victim’s/mutual residence	232	33.14
Other residence	267	38.14
Other location	196	28.00
Time of assault			0–1	3.373 (0.066)
Daytime (7 a.m.–7 p.m.)	234	33.43
Nighttime (7 p.m.–7 a.m.)	433	61.86
Presence of witnesses			0–1	6.425 (0.011)
At least 1 witness present	226	32.29
No witnesses reported	463	66.14
SANE conducted			0–1	0.104 (0.747)
Yes	470	67.14
No	227	32.43
Victim characteristics
Sex			0–1	0.282 (0.596)
Female	669	95.57
Male	31	4.43
Race			1–3
White	356	50.86		2.662 (0.103)
Black	298	42.57		3.319 (0.068)
Other	45	6.43		0.121 (0.728)
Age	(30.96)	(11.29)	17–84	−0.669 (0.504)
Resistance used			0–1	0.414 (0.520)
Physical or verbal resistance	522	74.57
No resistance noted	178	25.43
Injuries present			0–1	17.138 (<0.001)
Yes	186	26.57
None noted	514	73.43
Under the influence			0–1	6.257 (0.012)
Substances noted in the report	223	31.86
No substances noted	477	68.14
Suspected “date rape” drugs			0–1	14.970 (<0.001)
Yes	90	12.86
Not indicated in the report	608	87.11
Credibility/mental health issues			0–1	2.572 (0.109)
Yes	55	7.86
Not indicated in report	645	92.14
Criminal record			0–2	0.389 (0.823)
None	489	69.86
One type of offense	101	14.43
Multiple types of offenses	52	7.43
Initially cooperative			0–1	0.702 (0.402)
Yes	671	95.86
No	27	3.87
Stayed cooperative			0–1	8.850 (0.003)
Yes	540	77.14
No	123	18.41
Delayed reporting			0–1	26.612 (<0.001)
Reported within 1 day	538	76.87
Delayed by more than 1 day	162	23.14
Suspect characteristics
Sex			0–1	0.563 (0.755)
Male	690	98.57
Female	8	1.14
Race			1–3
White	208	29.71		0.004 (0.948)
Black	429	61.29		0.521 (0.470)
Other	62	8.86		1.797 (0.180)
Age	(37.33)	(12.05)	17–84	−1.602 (0.110)
Weapon used			0–1	16.213 (<0.001)
Yes	130	18.57
No weapon noted	570
Criminal record	81.43		0–2	13.604 (0.001)
None	343	49.00
One type of offense	150	21.43
Multiple types of offenses	141	20.14
Sexual assault charge history			0–1	23.498 (<0.001)
Yes	36	5.14
No history noted	598	94.32

Note: Incident and victim characteristics percentages are all taken from a possible total of 700. Cases with missing data not removed at the descriptive level.

.

2.2.1. Incident Characteristics

Based on the definitional issues described above, assaults were divided by type of penetration (Campbell et al. 2009), with vaginal penetration only being the most common (n = 389; 55.57%).

The victim/offender relationship was separated into three broad categories. Current or former romantic partners included any romantic and/or sexual relationship, including “friends with benefits” and similar contexts. Other known offenders were anyone else known to the victim whom they had not previously engaged in sexual conduct with, such as friends, acquaintances, and coworkers, among others. Lastly, strangers were individuals that the victims had not met prior to the date of the assault. This includes suspects who attacked the victim suddenly but also suspects that the victim met that night for the first time (e.g., at a bar or party). Other known offenders were the most common in this data (n = 290; 41.43%), supporting past research that known offenders are most often perpetrators of sexual assault (DeLoveh and Cattaneo 2017; Fisher et al. 2003; Tjaden and Thoennes 2006). Stranger assaults were approximately one-quarter of the cases here (n = 185; 26.43%). This number is likely inflated due to the higher likelihood of victims reporting stranger assaults (Clay-Warner and McMahon-Howard 2009).

Assault locations were also divided into three categories. The first encompassed places where a victim resided, including the victim’s residence, a mutual residence that the victim and suspect shared, or a facility where the victim resided long-term. Reported assaults occurred most often at other private residences where the victim did not live long-term (n = 268; 38.14%). Assaults happening at “other locations”, including outside, in a car, or in public, were the least common, also supporting prior research on sexual assaults most likely occurring in private settings.

Most reported assaults occurred during evening/nighttime hours (7 p.m.–7 a.m.; n = 433; 61.86%) and without the presence of witnesses (n = 463; 66.14%). A sexual assault nurse examination (SANE) was present in over two-thirds of cases (n = 470; 67.14). According to detectives in the instant department, biological evidence could be collected via a SANE up to seven days following the assault.

2.2.2. Victim Characteristics

Victims in the sample were overwhelmingly female (n = 669; 95.57%).4 Approximately half of the victims in the sample were white (n = 356; 50.86%), which was slightly lower than the census percentage of white residents. Victims were significantly more likely to be African American compared to the full population of the city (n = 298; 42.57% of the sample).5 The average age of victims was 30.96 (SD = 11.29; range: 17–84).

The next set of victim characteristics was based solely on initial patrol reports, then by follow-up detective interviews if available. Resistance is a dichotomous variable indicating any form of resistance noted by the victim. This includes either verbal or physical resistance, or both. Victims reported using some form of resistance in approximately 75% of cases (n = 522). In 18.29% of cases, victims explicitly reported not using resistance for a variety of reasons. To keep the highest number of cases in the multivariate analysis, this variable is considered as the victim clearly stating they used resistance (1) vs. no mention of resistance in the report (0). The same approach was taken for injuries, noted in 26.57% of reports, and substance use, mentioned by victims or responding officers in 31.86% of reports. If these variables were not mentioned in the report, they were treated as a (0) in multivariate analyses.

Suspected “date rape” drugs were determined by either the victim noting that they may have been drugged (e.g., the victim had two beers and then could no longer remember the evening) or the assigned detective requesting a drug-facilitated sexual assault test. The latter is only done in cases of suspected involuntary intoxication. This was a suggested possibility in 12.86% of cases. Credibility and mental health issues were coded based on prior research examining prosecutor decision-making (Alderden and Ullman 2012; Beichner and Spohn 2005) and included risk-taking behaviors (e.g., accepting a ride from a stranger, walking alone at night) along with labels by detectives as an “emotionally disturbed person”. A limited number of cases were classified in this way (n = 55; 7.86%). Related to credibility, the victim’s criminal history was obtained by looking at their city-based criminal record. Being a suspect in a crime or being arrested was counted in this category to capture prior involvement with the criminal justice system.6 Offense categories were property, person, or public order/drug crimes, with coding indicating no history, one type of offense, or multiple types of offenses. Most victims had no criminal history in the city of study (n = 489; 69.86%).7

Lastly, cooperation was measured as the victim being initially cooperative with responding patrol officers (n = 671; 95.86% of victims)8 and staying cooperative with their assigned detective for the duration of the case. In the current sample, continued cooperation is significantly higher than in the full sample, as the cases looked at here are those that detectives have determined should be sent to the district attorney for possible criminal charges (77.14% in the present sample compared to 34.51% in the full sample). Reports in the present sample were most likely to be done within 24 h of the assault taking place (n = 538; 76.86%).

2.2.3. Suspect Characteristics

The suspects in the sample were almost entirely male (n = 690; 98.57%). The majority of suspects were African American (n = 429; 61.29%), with a significantly higher proportion than the general city population. The average suspect age was older than that of the victims (M = 37.33; SD = 12.05; range: 17–84).

Weapons were not common in reported cases (n = 130; 18.57%). This is a higher percentage than estimated in sexual assaults generally and agrees with research suggesting that more serious crimes are more likely to be reported (Clay-Warner and McMahon-Howard 2009; Fisher et al. 2003). Criminal records were available for suspects who had committed other crimes only within the city limits of the Department of Study.9 Approximately half of the suspects (n = 343; 49.00%) had no criminal history noted, 20.14% had been suspects in multiple types of other offenses, and 5.14% (n = 36) had been charged, not just suspected, in a previous sex crime.

2.2.4. “Ideal Victim” Characteristics

Christie’s (1986) ideal victim characteristics are interpreted here through ten possible dichotomous variables. These can be seen in

Table 2. “Ideal victim” characteristics (n = 700).

	n	%	Min/Max
Vaginal penetration	520	74.29	0–1
Strangers	185	26.43	0–1
Victim’s residence	232	33.14	0–1
Daytime (7 a.m.–7 p.m.)	234	33.43	0–1
White victim	356	50.86	0–1
Elderly victims (65+)	7	1.00	0–1
No substances present	477	68.14	0–1
No credibility issues	645	92.14	0–1
Injuries present	186	26.57	0–1
Victim resisted	522	74.57	0–1
	M	SD	Min/Max
“Ideal victim” scale	4.84	1.35	1–10

Note: Incident and victim characteristics percentages are all taken from a possible total of 700. Cases with missing data not removed at the descriptive level.

. “The victim is weak” has been defined as a male offender and a female victim. Since this is the case for nearly 100% of the cases in this sample, sex has been excluded from the multivariate analyses. Weakness was also defined by Christie (1986) as being young or elderly. As there are no juveniles in this sample, the variable created for this criterion was “elderly” and included victims aged 65 and older.

“Victim is carrying out a respectable project” is often more thought of, as the victim cannot be doing anything unrespectable (Taylor 2022). In this data, alcohol/substance use issues and credibility issues were examined, with (1) being “no issues” present.10 As mentioned above, this was solely based on the department records management system and mainly consisted of initial reports written by responding patrol officers. Lastly, time was considered here as being out at night, which may be considered “unrespectable”. The time of the incident was divided into two periods: 7 a.m.–7 p.m. being the “appropriate” hour to be out of one’s home or have visitors versus the nighttime hours of 7 p.m.–7 a.m.

The location of the assault should also be somewhere a victim “could not possibly be blamed for being” (Christie 1986, p. 19). This was considered to be the victim’s residence, as defined above. Previous research has not considered residences to be blameworthy places, but the researchers feel that this is optimistic, as victims could also be blamed for attending house parties or meeting a potential romantic partner at a private residence. Fansher and McCarns (2019) reported that nearly 30% of college-aged online daters meet for the first time at a private residence, which is arguably a behavior that may elicit blame from a prosecutor or jury.

Injuries and victim resistance are the items chosen to represent that “the offender was big and bad”. Prior research does not consider the presence of resistance but commonly only injury (e.g., Taylor 2022). Injuries are arguably more likely to be present when a victim shows resistance. Further, a common question a victim may receive is, “Why didn’t you fight back?” (Heritage 2021; Patterson 2011), which supports the addition of this consideration as an “ideal victim” characteristic.

Cases involving strangers are also important for the consideration of an “ideal victim”. Additionally, these analyses will consider white victims to be more “ideal” due to their increased credibility among police officers, prosecutors, and juries (Myers and Hagan 1979). Lastly, as rape and sodomy overlap in definition, vaginal penetration, versus anal or oral, was considered “ideal” as it fits into historical definitions of rape in the United States.

These ten items were added to create a scale ranging from 1 to 10. The average case had 4.84 of these characteristics and was normally distributed (see Figure 1). The most important note here is that these variables are not correlated. The Kuder–Richardson coefficient, used to examine a reliability coefficient for a scale of dichotomous items and interpreted similar to Cronbach’s alpha (Casi 1998), was only 0.032. This suggests that the items are not correlated. The scale will remain in the first multivariate analysis with the understanding that the research question only seeks to see if the presence of more “ideal victim” characteristics influences prosecution decisions.

3. Results

3.1. Birvariate Results

Overall, prosecutors accepted 33.43% (n = 234) of the 700 cases presented to them by law enforcement. Bivariate comparisons between prosecutor acceptance and other variables of interest were done using chi-square tests, with victim age and offender age analyzed using t-tests. Bivariate results can be found in Table 1.

The majority of incident characteristics were related to prosecutor acceptance, beginning with assault type (χ² = 31.067; p < 0.001). Based solely on bivariate statistics, prosecutors were most likely to accept cases without penetration (54.24%), followed by those with only oral penetration (40.00%). An important note is that cases without penetration included many “attempted” assaults, with approximately 40% of these cases having a witness to the assault.

The victim/offender relationship also showed a significant relationship with prosecutor acceptance (χ² = 21.757; p < 0.001). Stranger cases had the highest percentage of prosecutor acceptance (n = 87; 47.03%), followed by other known offenders (n = 82; 28.28%), and cases involving current or former partners were least likely to be accepted by a prosecutor (n = 54; 27.69%). Assault location was also significant at the bivariate level (χ² = 9.543; p = 0.008), with assaults occurring at a non-residence most likely to result in prosecutor acceptance (n = 79; 40.31%), followed by victim/mutual residence (n = 82; 35.34%), with assaults occurring at other residences least likely to be accepted for prosecution (n = 72; 26.97%).

The presence of a witness increased prosecution (χ² = 6.425; p = 0.011). Approximately 40% of cases with witnesses were accepted for prosecution (n = 91), compared to 30% of cases without a witness (n = 140). The time of the assault was approaching significance (χ² = 3.373; p = 0.066); 38.46% of cases occurring during daytime hours were accepted (n = 90), compared to 31.41% of cases occurring during nighttime hours (n = 136). Interestingly, the presence of a sexual assault nurse examiner kit was not significantly related to prosecutor acceptance at the bivariate level.

For victim characteristics, demographics were generally not related to prosecutor acceptance. The only variable approaching significance was victims identifying as black/African American compared with other race victims (χ² = 3.319; p = 0.068).

The presence of injuries was related to a higher likelihood of prosecutor acceptance (χ² = 17.138; p < 0.001), with 45.70% (n = 85) of injury cases being accepted, compared to 28.99% (n = 149) of cases without injuries. Victims being under the influence was negatively related to prosecutor acceptance (χ² = 6.257; p = 0.012). Approximately one-quarter of these cases were accepted (n = 60; 26.91%), compared to 36.48% (n = 174) of cases where there was no indication that the victim was under the influence. Similarly, cases with suspected “date rape” drugs were less likely to be accepted for prosecution (χ² = 14.970; p < 0.001; 15.56% of cases with suspected drugs accepted vs. 36.18% of cases without).

Cases with delayed reporting of more than one day were significantly less likely to be accepted (χ² = 26.612; p < 0.001). Only 16.67% (n = 27) of cases with delayed reporting were accepted, compared to 38.47% (n = 207) of cases reported within a day of the assault. Lastly, continued cooperation was significant for prosecutor acceptance (χ² = 8.850; p = 0.003). Approximately one-third (n = 191) of cases with continued cooperation by the victim were accepted, compared to 21.14% (n = 26) of cases without.11

The use of resistance, criminal record, and initial cooperation were not significantly related to prosecutor acceptance at the bivariate level, nor were the credibility/mental health issues of the victim.

Offender characteristics were least likely to be related to prosecutor acceptance. The use of weapons increased the likelihood of prosecutor acceptance at the bivariate level (χ² = 16.213; p < 0.001). Nearly half of the cases (n = 63; 48.46%) where a weapon was used were accepted by prosecutors, compared with 30.00% (n = 171) of cases without a weapon. The presence of a criminal record for the offender was also related to acceptance (χ² = 13.604; p = 0.001). Offenders with multiple types of offenses on their record were more likely to see a case accepted (44.68%), followed by those with one type of offense on their record (33.33%). Offenders with no criminal record were least likely to see charges filed in a sexual assault case (27.41%). Offenders with a prior sexual assault charge were most likely to see charges filed in the present case (χ² = 23.498; p < 0.001; 69.44% of cases). Offender sex, race, and age were not significantly related to prosecutor acceptance at the bivariate level.

3.2. Multivariate Results

Binary logistic regressions were conducted for multivariate analyses, with prosecutor acceptance as the outcome variable. The “ideal victim” scale was included in the first analysis, along with all other independent variables that were presented in Table 1. If a variable was part of the “ideal victim” scale, it was not individually included in the first multivariate analysis, as doing so would count the variable twice, once in the scale and once by itself. Cases with missing data on any of the multivariate analyses variables were excluded.

Table 3. Binary logistic regression predicting prosecutor acceptance with scale (N = 556).

	b	SE	Std.
Ideal victim	0.136	0.081	0.089
Victim/offender relationship: Partners	−0.461	0.242	−0.102 *
Assault Location: Non-residence	0.104	0.244	0.022
Witness(es)	0.494	0.212	0.113 *
SANE conducted	−0.255	0.246	−0.056
Victim sex (1 = Female)	0.152	0.495	0.016
Suspected “date rape” drugs	−1.166	0.385	−0.190 **
Victim criminal history	−0.099	0.174	−0.030
Initial cooperation	−0.334	0.573	−0.031
Continued cooperation	0.998	0.308	0.188 ***
Delayed reporting (0 = Report within 1 day)	−0.965	0.303	−0.191 ***
Suspect sex (1 = Female)	0.784	1.129	0.037
Suspect race (1 = White)	0.455	0.234	0.100 *
Suspect age	0.013	0.009	0.075
Weapon used	0.836	0.256	0.157 ***
Suspect criminal history	0.177	0.136	0.070
Suspect history of sexual assault	1.292	0.468	0.144 **
Pseudo R2	0.128 ***

* p < 0.05, ** p < 0.01, *** p < 0.001.

displays multivariate predictors of prosecutor acceptance using the additive “ideal victim” scale along with the other variables of interest. This model predicted 12.83% of the variance in prosecutor acceptance (Pseudo R² = 0.1283; p < 0.001). The “ideal victim” scale was not significantly related to prosecutor acceptance. The most significant predictor of prosecutor acceptance in this model was a report being made within one day of the assault occurring (delayed reporting; β = −0.191; p < 0.001), followed by no suspected “date rape” drugs (β = −0.191; p < 0.01), and continued cooperation by the victim (β = 0.188; p < 0.01). Cases were also more likely to be accepted if the suspect had a prior sexual assault charge (β = 0.144; p < 0.01), the suspect was white (β = 0.100; p < 0.05), a weapon was used during the assault (β = 0.157; p < 0.001), the assault was not between current/former intimate partners (β = −0.102; p < 0.05), and there was at least one witness to the crime (β = 0.113; p < 0.05).

Non-residence assault locations, current or former partner relationships, and suspect non-sexual criminal history were not significant predictors of prosecutor acceptance, despite being significant at the bivariate level. The victim’s race remained non-significant.

Based on the non-significance of the “ideal victim” scale, the second analysis, seen in

Table 4. Binary logistic regression predicting prosecutor acceptance with individual items (N = 538).

	B	SE	Std.
Vaginal penetration	−0.802	0.295	−0.155 **
Victim/offender relationship
Partners	−0.383	0.289	−0.081
Strangers	0.811	0.274	0.161 **
Assault location
Victim/mutual residence	0.439	0.269	0.096
Non-residence	0.126	0.285	0.026
Time (1 = 7 a.m.–7 p.m.)	0.017	0.225	0.004
Witness(es)	0.403	0.222	0.088
SANE conducted	0.114	0.286	0.024
Victim sex (1 = Female)	0.807	0.544	0.080
Victim race (0 = Non-white)	−0.248	0.242	−0.057
Victim age	−0.012	0.011	−0.060
Resistance used	−0.180	0.268	−0.035
Physical injuries present	0.677	0.243	0.139 **
Victim substance use (1 = No issues)	0.035	0.259	0.008
Suspected “date rape” drugs	−1.300	0.412	−0.202 **
Victim credibility (1 = No issues)	0.732	0.437	0.090
Victim criminal history	−0.159	0.184	−0.045
Initial cooperation	−0.351	0.584	−0.031
Continued cooperation	1.032	0.314	0.185 ***
Delayed reporting (0 = Report within 1 day)	−0.774	0.313	−0.146 **
Suspect sex (1 = Female)	1.123	1.236	0.050
Suspect race (1 = White)	0.694	0.265	0.146 **
Suspect age	0.022	0.010	0.117 *
Weapon used	0.582	0.277	0.104 *
Suspect criminal history	0.145	0.144	0.055
Suspect history of sexual assault	1.298	0.475	0.138 ***
Pseudo R2	0.173 ***

* p < 0.05, ** p < 0.01, *** p < 0.001.

, examines the scale variables separately, combined with the same additional variables of interest used in the previous model, to predict prosecutor acceptance. This model had only one additional variable of significance compared to Table 3 but increased variance to 17.32% (Pseudo R² = 0.1732; p < 0.001). Only three of the ten characteristics originally included in the “ideal victim” scale were predictive of prosecutor acceptance. Cases involving strangers increased prosecutor acceptance (β = 0.161; p < 0.01), along with the presence of injuries (β = 0.139; p < 0.01). Cases involving vaginal penetration were less likely to be accepted than cases with other types of penetration or no penetration (β = −0.155; p < 0.01). Continued cooperation from the victim was also a significant predictor of acceptance (β = 0.185; p < 0.001).

As seen in the previous model, suspected “date rape” drugs decreased prosecutor acceptance (β = −0.202; p < 0.01), along with cases reported more than one day after the assault (β = −0.146; p < 0.001). With regards to the suspect, prosecutor acceptance was more likely if the offender had a prior sexual assault charge (β = 0.138; p < 0.01), was white (β = 0.146; p < 0.01), was older (β = 0.117; p < 0.05), and used a weapon (β = 0.104; p < 0.05).

The presence of witnesses was no longer significant when separating out “ideal victim” characteristics. Victim age and race, the presence of substances, the time of the assault, the location of the assault, victim credibility issues, and the use of resistance by the victim were not significantly related to prosecutor acceptance in this sample. A suspect’s non-sexual criminal history remained non-significant.

3.3. Exploratory Analysis

Based on the importance of victim cooperation in cases of sexual assault (Beichner and Spohn 2005; O’Neal et al. 2015; Pattavina et al. 2021; Spohn and Tellis 2019), an exploratory analysis was conducted on only cases in which victims remained cooperative with detectives and prosecutors. This decreased the sample from 700 to 455. Separated “ideal victim” characteristics, along with other variables of interest, were input into a backward stepwise binary logistic regression, with variables being removed if they were above p < 0.10. The results of this exploratory model can be seen in

Table 5. Stepwise binary logistic regression predicting prosecutor acceptance with cooperative victims only (N = 455).

	B	SE	Std.
Vaginal penetration	−0.583	0.255	−0.113 *
Injuries present	0.816	0.236	0.171 ***
Strangers	0.951	0.249	0.193 **
Witness(es) present	0.415	0.229	0.092
Suspected “date rape” drugs	−1.579	0.428	−0.250 ***
Delayed reporting	−0.400	0.155	−0.173 **
Suspect history of sexual assault	1.891	0.534	0.210 ***
Pseudo R2	0.157 ***

* p < 0.05, ** p < 0.01, *** p < 0.001.

.

Nineteen variables were initially in the model. The final model produced only six significant variables, predicting 15.70% of the variance (R² = 0.157; p < 0.001) of prosecutor acceptance in this sub-sample of cooperative victims. Prosecutor acceptance was more likely in cases involving strangers (β = 0.193; p < 0.001) and in cases where injuries were present (β = 0.171; p < 0.001). Cases involving vaginal penetration, compared to other types of penetration or attempted penetration cases, were less likely to be accepted (β = −0.113; p < 0.05).

The most impactful variable in this model was suspected “date rape” drugs. The presence of this variable significantly decreased the likelihood of prosecutor acceptance (β = −0.250; p < 0.001). Suspect history of sexual assault charges was the second most impactful variable in this model, significantly increasing prosecutor acceptance (β = 0.210; p < 0.001). Lastly, delayed reporting was associated with a decreased likelihood of prosecutor acceptance (β = −0.173; p < 0.01).

4. Discussion

The present study sought to examine predictors of prosecutor acceptance of sexual assault cases referred by law enforcement. In contrast to previous United States-based studies in this area, a census of data was collected on all reported cases over a nine-year period and did not only include cases filtered through a specialized program or prosecution unit. Several variables, including a series of variables designated to indicate an “ideal victim”, were explored as influencing prosecutor decisions to accept a reported case of sexual assault for criminal charges.

Notably, the additive presence of “ideal victim” characteristics was not a significant predictor of prosecutor decision-making, and the full model, with all variables analyzed separately, explained greater variance in the dependent variable.

The presence of injuries was, unsurprisingly, a significant predictor of prosecutor acceptance. The expectation of injuries dates to the original legal definitions of rape, requiring that “force” be used during the assault. Presumably, the only way to prove that force was used in an assault was through victim injuries. This misconception has held despite “force” requirements being removed from current definitions. Much research in this area looks at victims who report to an emergency room following an assault, arguably overestimating the number of sexual assaults that result in injury. More general research has produced varied estimates on the presence of moderate or severe injuries following sexual assault, ranging from 19% (Zilkens et al. 2017) to two-thirds (McGregor et al. 1999). In the present study, less than one-quarter of respondents had physical injuries noted on their reports, indicating that a relatively small portion of cases referred by law enforcement received a preference from prosecutors.

The significance of injuries in prosecutor acceptance is supported by prior research finding that injuries will increase the likelihood of a prosecutor accepting a case (Alderden and Ullman 2012; Beichner and Spohn 2005; McGregor et al. 1999; O’Neal et al. 2015). However, Pattavina et al. (2021) found that injury was related to an officer’s decision to consult with a prosecutor about a case, not necessarily with regard to the prosecutor’s decision to pursue charges. Beichner and Spohn (2005) and Spohn and Tellis (2019) also did not find injury to be a significant factor in their study sites, possibly suggesting that in these samples, the presence of physical injury was an important consideration prior to a case going before a prosecutor, resulting in law enforcement consideration acting as a mediator between physical injury and prosecutor acceptance.

The presence of weapons was also significantly related to prosecutor acceptance of a sexual assault case, as seen in prior literature (Beichner and Spohn 2005; Spohn and Holleran 2001; Spohn and Tellis 2019). The United States (Department of Justice 2013) estimates that a weapon was used in 11% of sexual assaults from 1994 to 2010, compared to 18.57% in the current sample. This supports the notion that when a victim perceives the crime as more serious, they are more likely to report it. However, this again highlights a small portion of reported cases receiving prosecutor preference.

Another significant characteristic was that the offender was a stranger to the victim. In this study, a stranger was anyone with whom the victim did not have a preexisting relationship and could be someone that the victim met that day/night or an offender who attacked the victim without any previous contact. Prosecutors in this study were more likely to accept cases from a stranger offender, supporting prior research (Beichner and Spohn 2005; McCahill et al. 1979; Pattavina et al. 2021; Spohn and Holleran 2001; Taylor 2022). This is again problematic when looking at the overall scope of sexual assault offenders. The National Crime Victimization Survey (Department of Justice 2017), which gathers data from a nationwide sample of households in the United States, reports that only 19.5% of rapes are committed by a stranger. Of the reports in this study, 26.43% were allegedly from strangers, in agreement with the general finding that stranger cases are more likely to be reported to law enforcement. Spohn et al. (2001) suggest that in stranger cases, prosecutors are less likely to consider victim risk-taking as a factor in their decision-making. The results of the present study found that victim risk-taking, measured here as victim credibility, was not a significant consideration in the overall models.

The presence of suspected “date rape” drugs was a significant variable in all multivariate models, with this scenario leading to a decreased likelihood of prosecution. Drug-facilitated sexual assault (DFSA) can be “perpetrated by drugs that render the victim: unconscious; very passive; powerless to resist sexual advances; unable to fight off the attacker; incapable of thinking clearly…with little or no memory of what happened” (Dinis-Oliveria and Magalhaes 2013, p. 471). In non-DFSA cases, the victims are more likely to remember the incident. It may be disorganized, but the memory of what happened is still there (Busch-Amendariz et al. 2010). Much of the time, in DFSA cases, the victims do not remember what happened. GHB is the drug of choice for most predators. This drug is popular because it can leave the system in 20 min to an hour, so it is virtually undetectable most of the time in a urine test (Grela et al. 2018). This drug quickly dissolves in liquid and is most often disguised with alcohol to enhance its effects (Kapitány-Fövény et al. 2017). Unfortunately, these drugs metabolize and leave the body quickly, making them difficult to detect, suggesting that law enforcement may be consistently underestimating the use of these drugs in reported cases of sexual assault.

Delayed reporting was a significant factor in all three multivariate analyses. When victims waited for more than 24 h to report their assault, prosecutors were significantly less likely to pursue criminal charges in their cases. Pattavina et al. (2021) found that this also affects whether law enforcement decides to consult with a prosecutor about a case. Anecdotally, the time with which a SANE nurse can collect viable DNA evidence seems to vary depending on location and the technology available. For example, the department studied here could collect evidence up to seven days after the assault, while a smaller department also studied by the author could collect evidence only up to three days after the assault. Regardless, in this study, the presence of a SANE and possible DNA evidence was not correlated with prosecutor acceptance, suggesting that it is not the presence of DNA evidence that helps a prosecutor decide to pursue a case but the other factors discussed above. However, it can be difficult for juries to assign guilt without some form of medically acquired and tested evidence (Taylor and Joudo Larsen 2005). The pressure for testing is high, as seen in numerous media stories about a “backlog”, but these often fail to consider the resources required for testing and the lack of guidelines on when a kit should be tested (EndtheBacklog 2023). The “backlog” media narrative briefly created a moral panic over the issue, implying that testing was the only thing keeping cases from moving forward. However, Campbell et al. (2021), analyzing a sample of newly tested kits, found that victim cooperation is still a very significant factor in determining if a case is moved forward through the criminal justice system.

The #MeToo movement, beginning in 2018, did appear to have some influence on the frequency of delayed reporting in the city of study, with a sharp increase in delayed reports occurring from 2017 to 2018 and remaining higher through 2019. The decision to report is often affected by the relationship between the victim and the offender. Bachman (1998) found that if a victim reported an assault, there was no significant difference in reporting time between stranger and acquaintance assaults. Decreased reporting has also been found for college women and when the victim knows the offender (Fisher et al. 2003). Current research is mixed on whether juries would be open to handing down guilty verdicts for sexual assault cases with delayed reporting. Franiuk et al. (2019) found that jurors were more likely to believe victims and have negative feelings toward the offender when victims reported immediately, compared to two months after the assault (also supported by Taylor and Joudo Larsen 2005), while Thompson et al. (2021) found that a one-year delay in reporting is more believable by a jury compared to a ten-year delay. There are myriad reasons why a victim of sexual assault may choose to delay reporting or not report at all. That was not examined here. However, it is estimated that only 2–10% of reported sexual assaults are found to be false (Lisak et al. 2010), suggesting that prosecutors should be more open to pursuing cases of sexual assault with delayed reporting, particularly if some form of evidence is still available.

Lastly, we would like to highlight the importance of suspect criminal histories. A previous arrest for sexual assault was significantly related to a prosecutor’s decision to accept, which suggests that prosecutors take repeat offenders seriously. However, a suspect having any other form of criminal history was not significantly related to the prosecutor’s decision-making. This is surprising due to the extensive criminal histories of some suspects in this study. Lisak and Miller (2002) infamously report on undetected rapists, noting that in a sample of college men, approximately 76 men (4.04% of their sample) had engaged in an average of 5.8 rapes each without any criminal justice intervention. Further, these men also admitted to an average of 13.75 other types of violent acts. The Rape, Abuse, and Incest National Network (RAINN 2023) goes further, reporting that 52% of arrested rape suspects will be released awaiting trial and approximately 14% will be arrested for committing another crime during this time. When considering the low percentage of victims that report and the high number of cases that do not result in any sort of conviction, the likelihood of suspects reoffending violently or sexually is almost terrifying.

4.1. Limitations

The present study was not without limitations. Unfortunately, we were only able to examine reported cases of sexual assault, which, as noted above, are a minority of sexual assaults perpetrated. In the United States, it is estimated that only 30% of sexual assaults are reported to the police, with differences depending on the age of the victim and whether or not they are enrolled in college (Department of Justice 2020). The findings of this study can only be understood as an analysis of reported assaults and not the likelihood of any assault being pursued by a prosecutor’s office. Further, juvenile cases were not assessed due to a lack of accessibility given to the researchers. In the instant agency, juvenile cases were those in which the victim or the offender were under the age of 17. As juvenile cases are significantly more likely to result in an approved charging decision (Spohn and Tellis 2019), child sex abuse decision-making should be investigated in additional studies.

There were also limitations with demographic variables. First, the database measured sex only and did not have an indication for gender. In this dataset, non-binary and transgender victims were only noted in an additional report. These were few, and it is possible that what we viewed was an underestimate of reports for this population, particularly if responding officers were not trained to ask about this, were uncomfortable asking, or were just unfamiliar with the differences between sex and gender. Second, as noted above, race was considered “white”, “black”, “Asian”, or “other”. Latinx/Hispanic was noted in a separate area as “ethnicity” with an “H” or “N” (not Hispanic/Latinx). This was often left blank in the dataset, suggesting an underestimate of Hispanic/Latinx individuals in the dataset and miscoding of these individuals as “white”.

Reports were also based on recollections from the first responding officer. Most reports were taken by generalized patrol officers, not specialized sex crime detectives. Reports varied in their level of detail, which led to a high amount of missing data as researchers did not want to make assumptions about variables of interest. For example, if the reporting officer did not specifically note the absence/presence of physical injuries, this variable was left as “no injuries reported”. Researchers could not be sure if the missing information meant that the variable was not present or if the responding officer did not include it in their report. For ambiguous statements in a report, researchers could refer to the victim’s interview with a detective, but if these interviews were not written down but only recorded with audio/video, researchers did not have access.

Lastly, criminal history was only available for offenses committed within the jurisdiction of the police department of study. As with most metropolitan areas, several suburbs surround the city, each with its own separate police department and jurisdiction. It is likely that criminal histories were underestimated, particularly for those individuals who did not reside in or frequent the city limits. Further, the database did not keep records of convictions, only of being a “suspect” and being “arrested”. A measure of previous convictions, not just involvement in the criminal justice system, would be beneficial for future analyses of this variable.

4.2. Policy Implications and Future Research

Deciding to report a rape to law enforcement can be a big step for many victims, but scrutiny from law enforcement and prosecutors regarding whether they are a “true victim” disrupts the process and could traumatize victims even further (Ahrens 2006; Jamel 2010; Maddox et al. 2011). Trauma-informed training is an essential requirement for those involved in the criminal justice system, as it can better prepare criminal justice officials in proper victim etiquette and correct long-held myths and stereotypes about the “ideal victim” of sexual assault. Several forms of trauma-informed courses have been created over the years to help spread awareness about multiple issues within the trainee’s community and promote self-awareness for the individuals taking the course (Campbell et al. 2020; Darwinkel et al. 2013; Lathan et al. 2019; Tidmarsh et al. 2020). In the United States, these courses may be a requirement in the training process, but that is not a guarantee. Worden et al. (2020) indicated that police officers in the New York Police Department (NYPD) held many biases (ex. gender, religion, race/ethnicity, disability, etc.) that could be impacting their behavior toward victims. When made aware of these implicit biases through training, officers were able to identify how their biases could be affecting their beliefs, attitudes, and behaviors and consciously reject those former stereotypes. Making criminal justice officials aware of their own biases toward victims and recognizing that victims come in different forms will help create a safe environment for victims to report their traumatic experiences and ensure that criminal justice personnel are doing everything in their power to help, not hinder the situation (Campbell et al. 2020; Darwinkel et al. 2013; Lathan et al. 2019; Tidmarsh et al. 2020). Further, trauma-informed training can improve an officer’s victim interviewing skills, resulting in greater recall, stronger statements, and a greater likelihood of remaining cooperative with the case (Patterson et al. 2009).

While many variables play a role in the decision-making process for prosecutors, drug-facilitated sexual assault (DFSA) cases appear to be a serious concern. The use of substances to inhibit victims during a sexual assault has become a significant problem (Beynon et al. 2008; Dinis-Oliveria and Magalhaes 2013; Soni et al. 2021). This decreases a victim’s ability to recall the events and provide a strong statement for their case. When there is suspicion of involuntary intoxication, DFSA testing should occur as quickly as possible, before the substances have been metabolized and are no longer in the victim’s system (Dinis-Oliveria and Magalhaes 2013). Normal toxicology screening procedures do not detect DFSA substances; therefore, they do not aid the victim in their claims (Vrees 2017). It is imperative that healthcare providers and first responders recognize signs/symptoms of DFSA to properly obtain samples (urinal and blood) and run the appropriate toxicology reports (Dinis-Oliveria and Magalhaes 2013). Having healthcare providers able to readily recognize the symptoms of DFSA, have the resources available, and take the appropriate actions will help gather the evidence before it deteriorates and aid law enforcement in their investigation and therefore prosecutors in their cases.

In the research above, victim characteristics were suggested to be a significant factor for prosecutors in the decision-making process. The characteristics are constantly weighed by prosecutors for them to achieve what they deem the “ideal victim” (Christie 1986). The “ideal victim” is a myth that harms victims during and before court. To grow past this illusion of what a victim should be, increased public awareness is needed to decrease myths surrounding sexual assault and sexual assault victims. This should start in public schools to provide the greatest reach to future jury members. Prior research has shown positive short-term (Frazier et al. 1994) and long-term effects of using school-based intervention programs to decrease rape myth acceptance (De La Rue et al. 2014). Arguably, acceptance of rape myths would contribute to an individual’s perception regarding the “ideal victim” of sexual assault.

Historically, research on sexual assault cases has heavily focused on victim characteristics as predictors, such as credibility, age, moral character, use of substances, etc. However, the present study found several offender-based characteristics to be predictive of prosecutor decision-making. Future research should focus more on offender characteristics compared to the traditional focus on the victim. By focusing on the victim alone, we will continue to evaluate victims for blameworthiness instead of spotlighting the offenders who perpetrate these horrific crimes.