A Systematic Review of Sexual Assault Case Attrition in the United States from 2000 to 2020

Abstract

This article systematically reviewed research findings of five sexual assault case outcomes (founding, arrest, referral to prosecution, charging, and conviction) between 2000 and 2020. Records were collected from PsychINFO and ProQuest and had to report at least one quantitative criminal justice outcome, include data from a U.S. sample and involve original research. Thirty-six records (288,066 sexual assault cases) were analyzed. Results were stratified by reported cases, referred cases, charged cases, and victim age. Studies show that 72.09% to 92.66% of sampled cases are founded (M = 89.21%), 17.73% to 53.42% result in arrest (M = 27.25%), 13.60% to 69.57% result in referral (M = 31.36%), 3.28% to 83.16% result in charging (M = 19.58%), and only 1.86% to 40.31% result in conviction (M = 8.28%). This considerable variability in case outcomes is related to differences in cases included or excluded (e.g., age criterion, only sampling cases with sexual assault kits); thus, these studies are not reflective of all sexual assault cases. Studies were most likely to report arrest rates, followed by charging, referral, conviction, and founding. Overall, we have a good picture of what attrition looks like in adolescent-adult sexual assault cases from the 1980s to early 2010s in the Midwest and Southern CA. More updated research in the remaining parts of the country is needed, particularly from rural locales. Specific recommendations for more precise measurement and reporting methods are provided to improve our understanding of attrition and strengthen where, how, and with whom interventions are needed.

Over the past five decades, legal and policy reforms have attempted to address the failure of the criminal justice system to adequately investigate and respond to sexual crime (Caringella-Macdonald, 1988; Spohn et al., 2014). More recently, media coverage of untested sexual assault kits (SAKs) and the viral #MeToo movement spurred national-level conversations around the response to sexual assault. While some efforts have resulted in improved case outcomes, including increased referral rates to prosecutors (R.Campbell, Bybee, et al., 2012), attrition in sexual assault cases remains problematic. As a gendered crime, most sexual assaults are perpetrated against women by men (Basile et al., 2022). About one in four women will be assaulted in their lifetime, with a majority first victimized before the age of 25 (Basile et al., 2022; S. G.Smith et al., 2018). Growing research demonstrates sexual assault is also prevalent among LGBTQ+ adults (21%; Langenderfer-Magruder et al., 2016) and, to some extent, men (4%; Basile et al., 2022). Despite high prevalence rates, less than a third of victims will report the assault to the police (S. G.Smith et al., 2018), even fewer offenders will be arrested, and the majority of cases will not result in prosecution or conviction (Lonsway & Archambault, 2012).

However, the most recent review (an integrative review) of case attrition is now over 10 years old (Lonsway & Archambault, 2012). Since then, documentaries have been released (e.g., I Am Evidence; Adelsic & Gandbhir, 2017), infamous offenders have been convicted (e.g., Larry Nassar; Hoffman, 2017), and laws have been enacted (e.g., SAK tracking; End the Backlog, n.d.). Many original research and secondary analysis studies have also been published. The present paper seeks to provide the field with updated figures in sexual assault case attrition to capture any changes during this time. This work will also help identify explicit next steps for future research and where interventions at the policy and practitioner level should be targeted.

Police and Prosecutor Decision-Making in Sexual Assault Cases

Attrition can occur at each stage of the criminal justice system. After receiving a sexual assault report, police must decide whether a crime has taken place (Spohn & Tellis, 2019). Following a preliminary investigation, police will found a case if there is enough evidence to meet the definition of sexual assault as defined by the state statute (United States Department of Justice, 2017b). If police conclude a crime did not occur, the case is deemed unfounded and is closed (Alderden & Ullman, 2012). In some instances, unfounded cases may also be reclassified to a lesser offense (Alderden & Ullman, 2012).

Following founding, officers continue collecting evidence until there is enough probable cause to arrest a suspect. Arrest is one way to “clear” or close a case. Cases that are not cleared by arrest are either left open for further investigation or are cleared by exceptional means. Exceptional clearance is used when the police know the identity and location of the offender and have enough evidence to make an arrest but are unable to for reasons out of officers’ control (e.g., victim decided not to participate in case, prosecutor declined to pursue case; United States Department of Justice, 2017a). Cases cleared by arrest or exceptional means contribute to a department’s clearance rate—the proportion of “solved” crimes to reported cases × 100 (Scott et al., 2022). Clearance rates are often the main metric used to determine how successful officers or agencies are at solving crime (Scott et al., 2022).

Once a suspect is arrested, police decide whether there is sufficient evidence to refer the case to prosecution, although some cases may be referred directly without arrest. This is typically seen in jurisdictions using pre-arrest consultation. In these instances, officers may believe there is enough probable cause to arrest a suspect, but send the case to prosecutors for preliminary review to determine whether prosecutors plan to pursue the case (Pattavina et al., 2021). Cases approved by prosecutors will likely be cleared by arrest, and cases that do not will likely be cleared by exceptional means (prosecutor declined to pursue case) or left open for further investigation (Pattavina et al., 2021).

Following referral (not pre-arrest consultation), prosecutors must decide whether to charge the suspect and the severity of charges brought against the suspect. If charged, suspects may plead guilty (or possibly plead down to a lesser charge), judges may dismiss the case, or the case will go to trial. At trial, suspects will be acquitted (not guilty at trial) or found guilty (Morabito et al., 2019). Conviction occurs when suspects plead guilty to the crime or are found guilty at trial.

Methodological Challenges in Sexual Assault Case Attrition

Traditionally, two types of data sources have been used to estimate attrition rates in sexual assault cases. The first is publicly available datasets, including the Uniform Crime Report (UCR) and the National Incident-Based Reporting System (NIBRS). The UCR is a series of programs across the United States that collect and forward their data to the Federal Bureau of Investigation (FBI) using a summary reporting system (SRS; United States Department of Justice, 2019a). Police agencies (i.e., tribal, college/university, city, state, federal) are responsible for tracking data within their district, including police employment, incidence of crime, and case clearances. Agencies voluntarily submit these data to their state program (or territorial, tribal, or federal program when appropriate). Programs then submit these data to the FBI, which compiles the data and makes it public. While UCR data can provide some insight into national crime trends, the voluntary nature of the UCR means not all jurisdictions are represented. For example, in 2019 the New York City Police Department and D.C. Metropolitan Police Department did not submit arrest data for violent crime (including rape) to the UCR and have inconsistently reported data for the past two decades (United States Department of Justice, 2019b). Additional methodological aspects of the UCR limit its usability and the conclusions that can be drawn from these data. For example, the UCR only includes cases cleared by police, so unfounded cases cannot be examined. The UCR also only reports case clearance as one outcome rather than disaggregating by cases cleared by arrest versus exceptional means (and the accompanying exceptional clearance categories).

To address methodological challenges with the UCR, the FBI officially switched from the SRS reporting system to NIBRS in January 2021. NIBRS is an incident-level reporting program for local, state, and federal law enforcement agencies and is more accurate and reliable than the SRS (Lantz, 2021; Strom & Smith, 2017). Still, few researchers have made use of NIBRS to examine sexual assault case outcomes, despite the program being in effect since 1985. Those who have, have tended to study only one outcome—like arrests (e.g., Richards et al., 2020; Stacey et al., 2017; Walfield, 2015)—rather than the progression of attrition throughout the criminal justice system. Moreover, NIBRS is currently limited in its generalizability as only 46% of law enforcement agencies participate nationwide, with participation biased toward less populous jurisdictions (McCormack et al., 2017).

The second source of data on sexual assault case attrition is scientific studies. These can provide rich insight into factors influencing police and prosecutor decisions in sexual assault cases and accurate estimates of attrition, especially when they cover more than one locale (e.g., R.Campbell et al., 2014; Morabito et al., 2019). However, most published studies focus on one jurisdiction (e.g., Alderden & Ullman, 2012; Spohn & Tellis, 2012a, 2019; Wentz, 2020), which limits their generalizability. Others do not disclose the study locale(s) (e.g., R.Campbell et al., 2014; Morabito et al., 2019), which enhances confidentiality, but bars assessment of geographic region. Many also make use of the same data source (e.g., Meeker et al., 2019; O’Neal & Spohn, 2017; Spohn & Tellis, 2012a, 2019). Similarly, highly cited studies have typically examined populous metropolitan areas in the Midwest or West (Alderden & Ullman, 2012; Spohn & Tellis, 2012b), with less known about smaller jurisdictions in other parts of the country. Measurement strategies have also varied widely. For example, some studies include cases with victims aged 12 and up (e.g., Walfield, 2015) while others include lifespan (Shaw, 2014). Previous research points to the importance of stratifying outcomes by age as young victims may have better outcomes than older victims (R.Campbell, Greeson, et al., 2012; Spohn & Tellis, 2019). However, there have been no systematic attempts to synthesize across these studies. Moreover, while local studies alone are helpful for informing policy and practice in a particular place, they limit our understanding, conclusions, and next steps at the national level.

In 2012, Lonsway and Archambault published an integrative review of sexual assault case attrition using both publicly available data and some scientific studies. The authors examined reporting, arrest, conviction, and incarceration rates from approximately 1960 through the mid-2000s in the United States. Their aim was to determine whether news reports about the “good” progress the criminal justice system had been making regarding sexual assault were founded. They concluded these claims were generally unsubstantiated and provided a robust discussion considering why attrition looked the way it did. They also concluded more rigorous research was needed to provide a fuller picture of case attrition, including work that examined case reporting, founding, charging, and conviction. The present review responds to this call by examining five sexual assault case outcomes (founding, arrest, referral to prosecution, charging, and conviction) and adds methodological rigor by systematically examining studies using publicly available data and scientific studies published from 2000 to 2020.

The Current Study

Overall, researchers, practitioners, and policymakers frequently report sexual assault attrition rates across research studies to demonstrate the need for increased attention to this issue. However, these numbers often rely on a small, unsystematic sample of studies; studies with significant methodological challenges; or outdated research. Thus, there is a need to systematically examine studies across the nation to provide a more comprehensive understanding of how police and prosecutors process sexual assault cases through every step of the criminal justice system. Studies using NIBRS can offer some picture of what case attrition looks like at the national level but are limited as they tend to focus on only one criminal justice outcome, and NIBRS data come from less populous jurisdictions. Scientific studies can offer some picture of what case attrition looks like at a local level but are limited as they do not stratify outcomes by age and tend to focus on more populous jurisdictions. By combining these complementary sources, we can construct a more accurate, current picture of studies of attrition in the United States. Further, the last review only captured data up to the early 2000s. Since then, there has been a great deal of attention to sexual assault (e.g., the #MeToo movement going viral) and the criminal justice response to sexual assault (e.g., sexual assault kit initiatives, trauma-informed interviewing by police, promotion of sexual assault response teams, etc.). These many changes suggest re-evaluation of the attrition of sexual assault cases is warranted.

In addition to providing a new look at attrition numbers, this systematic review will also critique the evidence base on sexual assault attrition and provide recommendations to guide the field. An up-to-date systematic review will help shape future priorities for research directions, methodologies, and reporting practices related to sexual assault attrition. For example, in 2012, Lonsway and Archambault called for research emphasis on prosecution and conviction outcomes. Is this call still warranted or has there been an increase in the number of studies examining these outcomes? Relatedly, this work can improve practice by identifying in which stage of the criminal justice system interventions are needed and with whom (police, prosecutors, or both). To date, no studies have systematically reviewed sexual assault case attrition in the United States. The present study addresses this gap through a systematic review of sexual assault case outcomes published from January 2000 through December 2020 across the United States. The following research questions were asked: what percent of sexual assault cases result in (a) founding, (b) arrest, (c) referral to prosecution, (d) charging, and (e) conviction? This review also distinguishes between case outcomes by age group to highlight the need for future research.

Method

Search Strategy and Inclusion Criteria

First, the authors identified five records they knew would meet the final inclusion criteria for this review. These records were then searched within PsychINFO and ProQuest to identify their subject headings. Subject headings are standardized terms under which publications are indexed by a subject specialist (Nova Southeastern University, n.d.). Then, frequencies were run on the subject headings to identify the most common index terms. These index terms then became the keywords used in this review. These index terms also contained the key sexual assault outcomes we were interested in.

To complete the search process, PsychINFO and three databases in ProQuest were searched: Criminology Collection, Social Science Database, and Sociology Collection. Two sets of keywords identified during the search strategy were searched simultaneously: (a) “sex* assault” OR “rape” OR “sex* violence” OR “sex* crime” OR “sex* offenses” OR “sex* abuse” and (b) “case processing” OR “criminal proceedings” OR “decision making” OR “decisionmaking” OR “prosecutorial discretion” OR “case dismissal.” The asterisk in sex* allowed for search results beginning with “sex,” including “sexual.”

Searches were limited to empirical peer-reviewed articles, dissertations, and reports published in English between January 1, 2000 and December 31, 2020. The initial search returned 1,412 records. To be included in this review, sources had to meet the following inclusion criteria: (a) report quantitative findings on sexual assault cases that resulted in at least one criminal justice outcome (e.g., arrest and referral to prosecution); (b) include data from a U.S. sample; and (c) involve original research. Of the original 1,412 records identified, 79 were selected for full-text review after removing 63 duplicates. Of those, 43 were excluded because they did not report at least one criminal justice outcome (k = 21), include data from a U.S.-based sample (k = 2), or involve original research (k = 7). Twelve additional records were excluded because they came from the same dataset and provided redundant information. One additional record was excluded as inaccessible because the author of this dissertation requested it not be made public. The final number of eligible records included in this review was K = 36. See Figure 1 for a PRISMA diagram detailing the screening and extraction process (Page et al., 2021).

Figure 1.

PRISMA flow diagram of search strategy.

Screening, Extraction, and Synthesis

Title and abstract screening were completed by the first and second authors using Covidence version 2.0. Following full-text screening, the first and second authors independently read and extracted data from each of the 36 records. When available, extracted information included years study data were collected, data source (e.g., prosecutor case files), study location, sample size, victim age range, sample characteristics (e.g., whether victims knew the offender), and case outcomes (i.e., number of cases resulting in founding, arrest, referral to prosecution, charging, and conviction). Discrepancies during the extraction process were discussed among all authors to reach a consensus.

Then, records were sorted into groups that could provide data on reported cases, referred cases, and/or charged cases. Within each group, records were sorted into four age groups: child-adolescent (≤17), adolescent-adult (≥12), child-adult (2–81), and unreported age (meaning the study did not report the victim age). Next, the number and percentage of sexual assault cases resulting in each outcome (when available) were extracted and/or calculated for each record. For example, if a study of 100 reported cases indicated that 40% were referred and 20% were charged, we would enter the referral rate (40%) and the charging rate (20%) out of reported cases. Then, we would calculate the charging rate out of the number of referred cases as (20% × 100)/(40% × 100) = 50%. Percentages by age group were calculated as the number of cases with that outcome divided by the number of cases within that age group. Finally, percentages were calculated across all studies, as the total number of cases with that outcome across studies divided by the number of cases across studies. All calculations were performed by the first and second authors independently, and any discrepancies were discussed with the third author.

Results

Study Characteristics

Across all studies (K = 36), sample sizes ranged from 20 to 215,602 (Mdn = 364, M = 8,002, SD = 35,907). Several studies used records from police (k = 17) and/or prosecutors (k = 5). Others used records from state or federal databases (k = 9), sexual assault nurse examiner (SANE) programs/sexual assault response teams (SART; k = 3), multidisciplinary teams (MDT; k = 1), child advocacy centers (CAC; k = 4), and/or victim reports (k = 1). These numbers are not mutually exclusive as studies sometimes used a combination of sources. Sample characteristics ranged widely across studies. Many sampled only cases with one female victim and one male offender. Others sampled cases with only SAKs or an identified offender. Victim ages ranged widely. Some studies sampled cases with specific age groups (e.g., 10–15 years old), while others were broader (e.g., 2–81). Many studies sampled cases at age 12 or above, while others did not specify age or only reported mean or median sample age. Of note, most studies examined various outcomes and predictors in sexual assault cases, while a few examined interventions (e.g., R.Campbell, Patterson, et al., 2012; R.Campbell et al., 2014). Overall, 17% of studies (k = 6) reported founding, 47% (k = 17) reported arrest, 28% (k = 10) reported referral, 36% (k = 13) reported charging, and 33% (k = 12) reported conviction outcomes. Fifty percent of studies (k = 18) reported more than one outcome. The percentage of cases reaching each outcome is detailed out of reported cases (Table 1), out of referred cases (Table 2), and out of charged cases (Table 3).

Table 1.

% Founded, Arrested, Referred, Charged, and Convicted for Reported Cases by Age Group (k = 29).

Authors (Year)	Years Data Collected	Data Source	Study Location	# Cases	Age Range (Years)	Sample Characteristics	% Founded	% Arrested	% Referred	% Charged	% Convicted
Cases with child and/or adolescent victims (no adults)
R. Campbell, Greeson et al. (2012)	Feb 1998–Nov 31, 2007	SANE program	Un reported	392	13–17	Cases with SAK and police report	n/a	n/a	65.56	83.16	40.31
D. W. Smith et al. (2006)	Aug-Oct [Year Unreported]	CAC, MDT, and prosecutor case files	Mid-south rural county	61	3–17	Male and female victims	n/a	n/a	n/a	3.28	3.28
Cases with adolescent and/or adult victims
Alderden and Ullman (2012)	Jan 2003–Aug 2003	Police case files	Midwest	465	≥18	Female victims with one male offender	90.54	23.49	24.95	7.14	n/a
R. Campbell et al. (2014)	June 1995–Sept 2009	SANE program	Rural, midsized, and urban locations	1696	≥18	Cases with SAK; SAK and police report not anonymous	n/a	n/a	13.60	n/a	9.43
R Campbell, Patterson et al. (2012)	1994–2005	SANE program and police case files	Geographically diverse county in the Midwest	293	≥18	Cases with SAKs	n/a	n/a	53.92	38.91	26.28
Durfee and Fetzer (20l6)	2010	NIBRS	33 statesa and DC	2530	Adult	Female victims and male offenders; repeat offenses excluded	n/a	26.21	n/a	n/a	n/a
Johnson et al. (2012)	2003	NiJ	Los Angeles, CA; Indianapolis, Evansville, Fort Wayne, and South Bend IN	602	Adult	Female victims and male offenders; half the cases had SAKs	n/a	45.02	n/a	n/a	11.13
Kaiser et al. (2015)	2008	Police and sheriff case files	Los Angeles, CA	944	≥12	Female victims	n/a	39.62	n/a	n/a	n/a
Kelley and Campbell (2013)	1995–2005	Police case files	Geographically diverse county in the Midwest	266	Adult	Offender identified	83.83	n/a	48.87	n/a	n/a
Morabito et al. (2019)	2008–2010	NIBRS and Police case files	Six jurisdictions across the United States	2887	>12	Female victims	92.66	18.84	n/a	12.57	6.55
Nugent-Borakove et al. (2006)	1997–2001	SANE/SART case files	Monmouth County, NJ; Sedgwick County, KS; Suffolk County, MA	530	≥18	Female victims	n/a	39.25	n/a	n/a	n/a
O’Neal (2015)	2008 and 2010	Police and detective case files	Los Angeles, CA	401	≥12	Female victims	78.05	34.66	n/a	n/a	n/a
O’Neal and Spohn (2017)	2008	Police and sheriff case files	Los Angeles, CA	161	≥12	Female victim and male offenders; intimate partner sexual assault	n/a	53.42	69.57	21.74	n/a
Patterson (2011)	1999–2007	Victim reports	Unreported	20	18–53	Female victims; cases with SAKs	n/a	n/a	n/a	45.00	n/a
Richards et al. (2020)	2001	NIBRS	36 states	21,977	≥15	Single victim and offender	n/a	23.77	n/a	n/a	n/a
Siller (2018)	2005–2011	Police case files	Northeastern metropolitan	189	≥12	Female victims and male offenders; offender identified; relationship between victim/offender known	n/a	35.98	n/a	37.57	n/a
Spohn and Tellis (2019)	2008	Police and sheriff case files	Los Angeles, CA	491	≥12	Female victims; offenders identified	n/a	n/a	n/a	23.42	n/a
Stacey et al. (2017)	2002–2009	NIBRS	35 states	215,602	≥12	Female victim and one male offender	n/a	25.88	n/a	n/a	n/a
Walfield (2015)	2006–2011	NIBRS (2006–2011) and LEMAS (2007)	238 police departments across the United States	16,231	≥12	Female victims and one male offender	n/a	52.71	n/a	22.32	n/a
Wentz (2020)	Jan 1, 2000–Dec 31, 2010	Police and prosecutor case files	Midwestern city	570	18–80	Female victim and one male offender	n/a	n/a	46.14	13.86	10.00
Ylang and Holtfreter (2020)	1982–2012	NACJD	Los Angeles, CA	310	≥12	Offender identified; cases with SAKs	n/a	45.16	n/a	n/a	n/a
Cases with chi Id-adult victims
Coulborn et al. (2000)	1988–1989	Prosecutor case files	County in southwest, MI	323	3–21	Male and female victims	n/a	n/a	n/a	n/a	1.86
Lovell et al. (2020)	1993–2011	Prosecutor case files	Cuyahoga County, OH	717	Mdn = 23 (Q1 = 17, Q3 = 32)	Cases with SAKs	90.52	n/a	41.28	6.69	n/a
Shaw (2014)	2008	Police case files	Detroit, MI	248	2–81 (M = 23.36)	Cases with SAKs	n/a	32.26	35.08	n/a	n/a
Cases with unreported victim age
Bouffard (2000)	1995	Police case files	Urban/suburban county	326	M = 24		72.09	18.10	n/a	n/a	n/a
Stolzenberg and D’Alessio (2004)	2000	NIBRS	3,070 reporting jurisdictions across 19 statesb and DC	12,342	M = 20.8		n/a	24.09	n/a	n/a	n/a
Tasca et al. (2013)	2003	Police case files	Large city in AZ	220	Mdn = 31.2		n/a	17.73	n/a	n/a	n/a
Lovell et al. (2018)	1980–2013	Prosecutor case files	Cuyahoga County, OH	4966	Unreported	Cases with SAKs	n/a	n/a	n/a	12.55	4.31
Snodgrass (2006)	Jan 2000–Dec 2003	State law database	Anchorage, AK	1052	Unreported	One victim, one offender	n/a	n/a	17.87	12.07	10.55

Note. CAC = child advocacy center; LEMAS = law enforcement management and administrative statistics; MDT = multidisciplinary team; NACJD = National Archive of Criminal Justice Data; NIBRS = National Incident-Based Reporting

System; NIJ = National Institute of Justice; SANE = sexual assault nurse examiner; SART = sexual assault responses team; UCR = uniform crime report.

^aStates include AL, AK, AZ, CO, CT, DE, GE, ID, IL, IO, KS, LS, ME, MA, MI, MO, MT, ND, NE, NH, OH, OR, RI, SC, SD, TE, TX, UT, VA, VT, WA, WS, WV.

^bStates include AZ, CO, CT, ID, IA, KS, KT, MA, MI, NE, ND, OH, SC, TN, TX, UT, VT, VA, WV.

Table 2.

% Charged and Convicted for Referred Cases by Age Group (k = 14).

Authors (Year)	Years Data Collected	Data Source	Study Location	# ReferredCases	Age Range(Years)	Sample Characteristics	% Charged	% Convicted
Cases with child and/or adolescent victims (no adults)
R. Campbell, Greeson et al. (2012)	Feb 1998–Nov 31, 2007	SANE program	Unreported	255	13–17	Cases with SAK and police report	74.90	61.96
Edelson Goldberg and Joa (2010)	Jan 2002–Dec 2004	Prosecutor case files	Moderate-sized city in pacific northwest	172	1–17	Male and female victims	63.95	n/a
Edinburgh et al. (2008)	Jan 1998–Dec 2003	Hospital-based and community CAC case files	Unreported	256	10–15	Extrafamilial sexual abuse	57.03	n/a
D. W. Smith et al. (2006)	Aug–Oct [Year Unreported]	CAC, MDT, and prosecutor case files	Mid-south rural county	11	3–17	Male and female victims	18.18	18.18
Walsh et al. (2008)	Dec 2001–Dec 2003	CAC case files	Dallas, TX	160	2–17		n/a	43.75
Cases with adolescent and/or adult victims
Alderden and Ullman (2012)	Jan 2003–Aug 2003	Police case files	Midwest	465	≥18	Female victims with one male offender	38.79	n/a
R. Campbell, Patterson et al. (2012)	1994–2005	SANE program and police case files	Geographically diverse county in the Midwest	156	≥18	Cases with SAKs	73.08	49.04
O’Neal and Spohn (2017)	2008	Police and sheriff case files	Los Angeles, CA	112	≥12	Female victim and male offenders; intimate partner sexual assault	31.25	n/a
Patterson (2011)	1999–2007	Victim reports	Unreported	20	18–53	Female victims; cases with SAKs	45.00	n/a
St. George and Spohn (2018)	2008	Police and sheriff case files	Los Angeles, CA	476	≥12		37.82	n/a
Spohn et al. (2001)	1996–1998	Police and prosecutor case files	Miami, FL; Kansas City, MO; Philadelphia, PA	666	≥12	Female victims (Miami and Kansas City)	54.50	44.74
Wentz (2020)	Jan 1, 2000–Dec 31, 2010	Police and prosecutor case files	Midwestern city	570	18–80	Female victim and one male offender	30.04	21.67
Wood et al. (2010)	2003–2004	Police trooper case files	AK	239	≥14	Cases founded; adult offender	72.38	n/a
Cases with child-adult victims
Lovell et al. (2020)	1993–2011	Prosecutor case files	Cuyahoga County, OH	296	Mdn = 23 (Q1 = 17, Q3 = 32)	Cases with SAKs	16.22	n/a
Cases with unreported victim age
Snodgrass (2006)	Jan 2000–Dec 2003	State law database	Anchorage, AK	1052	Unreported	Sexual assaults committed against one victim by one offender	67.55	59.00

Note. CAC = child advocacy center; MDT = multidisciplinary team; SANE = sexual assault nurse examiner.

Table 3.

% Convicted for Charged Cases by Age Group (k = 8).

Authors (Year)	Years Data Collected	Data Source	Study Location	# Charged Cases	Age Range (Years)	Sample Characteristics	% Convicted
Cases with child and/or adolescent victims (no adults)
R. Campbell, Greeson et al. (2012)	Feb 1998–Nov 31,2007	SANE program	Unreported	191	13–17	Cases with SAK and police report	82.72
Duron (2018)	Jul 2009–Oct 2013	CAC case files	Suburban county in Southern U.S.	50	3–16	Adult offender	72.00
D. W. Smith et al. (2006)	Aug–Oct [Year unreported]	CAC, MDT, and prosecutor case files	Mid-south rural county	2	3–17	Male and female victims	100
Cases with adolescent and/or adult victims
R. Campbell, Patterson et al. (2012)	1994–2005	SANE program and police case files	Geographically diverse county in the Midwest	114	≥18	Cases with SAKs	67.54
Morabito et al. (2019)	2008–2010	NIBRS and Police case files	Six jurisdictions across the United States	354	>12	Female victims	53.39
Spohn et al. (2001)	1996–1998	Police and prosecutor case files	Miami, FL (1996–1998); Kansas City, MO (1996–1998); Philadelphia, PA (1997)	363	≥12	Female victims (Miami and Kansas City)	82.09
Wentz (2020)	Jan 1, 2000–Dec 31, 2010	Police and court records	Midwestern city	79	18–80 (M = 28.05)		72.15
Cases with unreported victim age
Snodgrass (2006)	Jan 2000–Dec 2003	State law database	Anchorage, AK	127	Unreported	Sexual assaults committed against one victim by one offender	87.40

Note. There were no studies with child-adult victims. CAC = Child Advocacy Center, MDT = Multidisciplinary Team, SANE = Sexual Assault Nurse Examiner.

Outcomes by Reported Cases

Twenty-nine studies examined outcomes out of reported cases (Table 1). Most were in the Midwest, and almost all pulled data from police and/or prosecutor case files. Across all age groups for reported cases (Table 4), 89.21% of cases were founded, 27.25% resulted in arrest, 31.36% were referred, 19.58% were charged, and 8.28% yielded a conviction.

Table 4.

Critical Findings: Total % Founded, Arrested, Referred, Charged, and Convicted by Case Type and Victim Age.

	Mean % Founded	Mean % Arrested	Mean % Referred	Mean % Charged	Mean % Convicted
Age Group	# studies (Total N)	# studies (Total N)	# studies (Total N)	# studies (Total N)	# studies (Total N)
All reported cases
Child and/or Adolescent	n/a	n/a	65.56% k = 1 (N = 392)	72.41% k = 2 (N = 453)	35.32% k = 2 (N = 453)
Adolescent and/or Adult	90.37%	27.47%	29.26%	20.90%	9.04%
	k = 4 (N = 4,019)	k = 13 (N = 262,829)	k = 6 (N = 3,451)	k = 9 (N = 21,307)	k = 7 (N = 6,532)
Child-Adult	90.51%	32.26%	39.69%	6.69%	10.55%
	k = 1 (N = 717)	k = 1 (N = 248)	k = 2 (N = 965)	k = 1 (N = 717)	k = 1 (N = 1,052)
Unreported	72.09%	22.59%	17.87%	11.64%	4.31%
	k = 1 (N = 717)	k = 3 (N = 12,888)	k = 1 (N = 1,052)	k = 2 (N = 3,506)	k = 1 (N = 4,966)
TOTAL	89.21%	27.25%	31.36%	19.58%	8.28%
	k = 6 (N = 5,062)	k = 17 (N = 275,965)	k = 10 (N = 5,860)	k = 13 (N = 28,495)	k = 10 (N = 13,003)
Referred cases
Child and/or Adolescent	n/a	n/a	n/a	64.70%	53.99%
				k = 4 (N = 694)	k = 3 (N = 426)
Adolescent and/or Adult	n/a	n/a	n/a	48.73%	39.78%
				k = 8 (N = 2,048)	k = 3 (N = 1,086)
Child-Adult	n/a	n/a	n/a	16.22%	n/a
				k = 1 (N = 296)	n/a
Unreported	n/a	n/a	n/a	67.55%	59.04%
				k = 1 (N = 188)	k = 1 (N = 188)
TOTAL	n/a	n/a	n/a	50.27%	45.47%
				k = 14 (N = 3,226)	k = 7 (N = 1,700)
Charged cases
Child and/or Adolescent	n/a	n/a	n/a	n/a	80.67%
					k = 3 (N = 243)
Adolescent and/or Adult	n/a	n/a	n/a	n/a	68.24%
					k = 4 (N = 910)
Child-Adult	n/a	n/a	n/a	n/a	n/a
Unreported	n/a	n/a	n/a	n/a	87.40%
					k = 1 (N = 127)
TOTAL	n/a	n/a	n/a	n/a	72.50%
					k = 8 (N = 1,280)

Founding.

Few studies reported on founding rates (k = 6). As shown in Table 1, founding rates across reported cases were similar, ranging from 72.09% (Bouffard, 2000) to 92.66% (Morabito et al., 2019). For example, Morabito et al. (2019)’s replication study of Spohn (2012a)’s research examined a sample of n = 2,887 sexual assaults among female victims aged 12 and up. Data were collected from six jurisdictions across the United States and the authors reported a 92.66% founding rate. Similarly, Lovell et al. (2020) examined n = 717 child through adult sexual assault cases with untested SAKs in populous Cuyhagoa County, OH, and reported a 90.52% founding rate. Overall, these data suggest most cases move past founding to be investigated by police.

Arrest and Referral.

Seventeen studies reported arrest rates and k = 10 studies reported referral rates. Data were most often sampled from police and/or prosecutor case files or NIBRS, and studies tended to include SAKs and cases with one female victim and one male offender (who may or may not be identified by police). As shown in Table 1, arrest rates across reported cases ranged from 17.73% (Tasca et al., 2013) to 53.42% of reported cases (O’Neal & Spohn, 2017); referrals ranged from 13.60% (R.Campbell et al., 2014) to 69.57% of reported cases (O’Neal & Spohn, 2017), with most below 50%. Of note, higher and lower arrest and referral rates appeared to be related to sampling criteria. For example, the lowest arrest rate (17.73%) found in Tasca et al. (2013) was established using police case files from a large city in AZ (victim age unreported). Only half the sample knew the offender, while all adolescent-adult victims in O’Neal and Spohn’s study (2017), which had the highest arrest rate of 53.42%, knew the offender. Similarly, O’Neal and Spohn (2017) had the highest referral rate (69.57%) while R.Campbell et al. (2014)’s study of adults who completed non-anonymous SAKs had the lowest (13.60%). Overall, these studies suggest police make an arrest and/or refer to prosecution in about one-third of cases.

Charging and Conviction.

With regard to charging and conviction, k = 13 studies reported charging rates and k = 10 studies reported conviction rates. Records were typically pulled from police and/or prosecutor case files and many included only cases with female victims and male offenders. As shown in Table 1, charging rates ranged from 3.28% (D. W.Smith et al., 2006) to 83.16% of reported cases (R.Campbell, Greeson, et al., 2012). Conviction rates were generally low (minimum = 1.86%; Coulborn Faller & Henry, 2000), although like charging, R.Campbell, Greeson, et al. (2012)’s study had the highest conviction rate of 40.31%. Of note, R.Campbell, Greeson, et al. (2012) examined cases from a SANE program with an SAK and police report; thus, this may be why they report the highest charging and conviction rates. Overall, these studies indicate less than a quarter of cases are charged and even fewer result in conviction.

Outcomes by Referred Cases

Fourteen studies provided charging and conviction outcomes for referred cases (Table 2). Most were in the Midwest or West, and data were typically pulled from police and/or prosecutor case files or CACs. Cases with younger victims most often included male and female victims, while cases with older victims included female victims and cases with an SAK.

Charging.

Across all age groups among referred cases (Table 4), 50.27% were charged and 45.47% were convicted. As shown in Table 2, charging rates across studies were variable, ranging from 18.18% (D. W.Smith et al., 2006) to 74.90% (R.Campbell, Greeson, et al., 2012). For example, D. W.Smith et al. (2006) examined a sample of n = 61 male and female child-adolescent sexual abuse cases during an unreported time period in the South. Of referred cases (n = 11), 18.18% were charged. Their small sample size of referred cases may have affected charging rates. In contrast, R.Campbell, Greeson, et al. (2012) examined a sample of n = 392 adolescent sexual assault cases with SAKs in the Midwest. Of referred cases (n = 255), 74.90% were charged. As with the above, it may be that the presence of a SANE program in R.Campbell, Greeson, et al. (2012)’s study impacted charging rates so significantly.

Conviction.

As shown in Table 2, conviction rates also varied, ranging from 18.18% (D. W.Smith et al., 2006) to 61.96% (R.Campbell, Greeson, et al., 2012). As with charges, D. W.Smith et al. (2006) had the lowest conviction rate, finding that of referred cases (n = 11), 18.18% were convicted. Again, it may be their small sample size that affected conviction rates. On the other hand, R.Campbell, Greeson, et al. (2012) had the highest conviction rate, with 61.96% of referred cases (n = 255) resulting in a conviction of some kind. Overall, these studies suggest about half of the referred cases result in charges and convictions.

Outcomes by Charged Cases

Eight studies provided conviction outcomes out of charged cases (Table 3). Most records were pulled from police and/or prosecutor case files or CAC case files. Studies were located across the country in rural to urban locations and included both male and female victims. Across all age groups among charged cases (Table 4), 70.05% were convicted of some charge. As shown in Table 3, all conviction rates were over 50% and ranged from 53.39% (Morabito et al., 2019) to 100% (D. W.Smith et al., 2006) of charged cases. In the case of D. W.Smith et al. (2006), only two cases were charged, and both resulted in convictions. Overall, these data indicate most charged cases result in some conviction.

Outcomes by Age Group

Unsurprisingly, studies with child and/or adolescent cases (no adults) had less attrition compared to all other age groups. Among reported cases (Table 4), studies with child-adolescent had the highest referral (65.56%), charging (72.41%), and conviction (35.32%) rates. This is particularly noteworthy when compared to referral (29.26%), charging (20.90%), and conviction (9.04%) rates of studies of adolescent-adults. Similarly, among referred cases (Table 4), studies with childadolescents had the highest charging (64.70%) and conviction (53.99%) rates. This pattern was also generally seen among charged cases: child and/or adolescent cases had an 80.67% conviction rate while adolescent and/or adult cases had a 68.24% conviction rate. However, the one study representing the unreported age group had the highest conviction rate (87.40%), although it is difficult to draw conclusions from this. Overall, these results suggest charging rates for studies with adolescent-adult victims are lower than cases with child victims.

Discussion

This systematic review presents estimates of sexual assault case attrition in the criminal justice system published over a 20-year period from January 2000 and December 2020. We examined the findings of 288,066 sexual assault cases from K = 36 studies across the United States on five outcomes: founding, arrest, referral to prosecution, charging, and conviction. Across studies, 72.09% to 92.66% of sampled cases were founded (M = 89.21%), 17.73% to 53.42% of sampled cases resulted in an arrest (M = 27.25%), 13.60% to 69.57% of sampled cases resulted in referral (M = 31.36%), 3.28% to 83.16% of sampled cases resulted in charging (M = 19.58%), and only 1.86% to 40.31% of sampled cases resulted in conviction (M = 8.28%). Overall, we see a very similar pattern of attrition to that demonstrated by Lonsway and Archambault (2012), suggesting little has changed since their review was conducted with data from the 1960s through the early 2000s. However, despite our intentions to capture cultural shifts during the past two decades, most of the studies included in our review examined data from the 1980s to early 2010s (more on this below). So, (a) there is some overlap in our timeframes, and (b) we do not have data recent enough to comment on the impact of things like #MeToo going viral. Still, the impact of these figures can be seen when comparing the conviction rate of sexual assault to that of other violent offenses. In 2004, the conviction rate for felony defendants charged with a violent offense was 61%—almost nine times higher than sexual assault (Bureau of Justice Statistics, 2008). More recently, homicide conviction rates across the United States were cited at 34%—almost five times higher than sexual assault (Baskin & Sommers, 2010). This is unsurprising given previous research has determined sexual assault is the least likely violent crime to be founded (Hoffman et al., 2023) and cleared by arrest (Walfield, 2015).

It is critically important that readers note the wide variability across studies of the same case outcome. Some of the variability may be due to legitimate contextual differences (e.g., organizational culture, state policy) causing meaningful variability in attrition (Shaw et al., 2020). On the other hand, studies varied widely in the types of cases they included and excluded. For example, studies varied in inclusion based on victim age, when age is a known predictor of sexual assault case attrition (R.Campbell, Greeson, et al., 2012; Spohn & Tellis, 2019). Further, many studies only included cases with an identified offender, or a non-anonymous police report, both of which are likely to decrease case attrition (Morabito et al., 2019; Ylang & Holtfreter, 2020). Moreover, some studies only focused on cases with a completed SANE exam/SAK or communities with notable collaboration among sexual assault responders (e.g., SARTs), which again, are factors known to improve case outcomes (R.Campbell, Bybee, et al., 2012; Greeson & R.Campbell, 2015). Thus, it is important to remember that these case outcomes are representative of the studies published during this period but are not representative of reported sexual assault cases, referred cases, or charged cases. In fact, studies with the highest outcome rates tended to sample cases with specific types of interventions (e.g., SANE exams and SAKs) that are either not universally available or accessed by survivors. This suggests that the means and maximums we present are likely inflated, compared to outcome rates for more typical sexual assault cases.

Another interesting note is that while inclusion criteria allowed for studies published from 2000 through 2020, the bulk of sexual assault cases were collected from the early 2010s or prior. During that time period, the FBI’s definition of rape was narrow and only included male penile penetration of a female vagina (United States Department of Justice, 2012). This definition was expanded in 2012, but it was only in 2016 that the then director indicated UCR programs should discontinue using the legacy definition when reporting rape (United States Department of Justice, 2012). As a result, we might expect even greater variability in outcomes when comparing data from before and after 2012, especially among studies using national-level data.

Strengths of the Existing Literature

A second goal of our systematic review was to critique the evidence base. The body of research we reviewed has several notable strengths. First, many studies cover several phases of case processing, which helps researchers and policymakers understand what attrition looks like in the locality where data were collected. In particular, studies that cover reported cases and follow them through the entire process (e.g., Alderden & Ullman, 2012; Lovell et al., 2020; Morabito et al., 2019), compared to studies that only focus on later decision points, are especially useful. This is because there is a great deal of variability in how jurisdictions handle decisions like referrals. For example, some jurisdictions are expected to refer all cases to prosecutors, while in others, police have more discretion when determining whether to send a case forward or not. As a result, jurisdictions where all cases were referred might have lower charging rates (e.g., due to cases being perceived as less likely to succeed) than those where police were more selective about referring cases. Ultimately, charging rates look worse in the first jurisdictions, but they might have the same conviction rate as the second jurisdiction because cases dropped out at different phases. Thus, whenever possible, it is important to examine data across all phases of the criminal justice system as well as report if prosecutors are consulted during referral decisions.

A second strength is that researchers were clear about their inclusion and exclusion criteria (e.g., whether cases had SAKs). Relatedly, it was easy to identify where studies pulled their data from (e.g., police and/or prosecutor case files) and what they coded for. In particular, some studies demonstrated methodological rigor by double coding case records and establishing interrater reliability (e.g., Kelley & Campbell, 2013; Shaw, 2014). While not directly the focus of this paper, many studies also used multiple data sources, which allows researchers to understand factors related to attrition (e.g., Walfield, 2015; Wentz, 2020). Finally, researchers have examined attrition across multiple jurisdictions in the country. In sum, this work provides a good picture of what attrition looks like in adolescent-adult sexual assault cases from the 1980s to early 2010s, especially in the Midwest and Southern CA.

Critiques of the Literature and Suggestions for Future Research

Correspondingly, this body of research has several notable limitations with respect to both methodology and reporting. We have several suggestions (see Table 5). First, there were some studies excluded from this review because they combined multiple outcomes into a single variable (e.g., “charges and conviction” versus “charges” and “conviction”). In this example, we could not determine the number of people who had been charged but not convicted. This prevents researchers from pinpointing the direct source of the attrition (e.g., decision to charge versus ability to obtain a conviction), and different factors may influence disparate parts of the process. This choice may be an artifact of how data are collected or reported prior to when researchers obtain them, however, we encourage researchers to separate outcomes when possible. Second, arrest was the most common outcome reported, followed by charging, referral, conviction, and founding. More research examining founding and conviction, which represent the first and last stage of case progression through the criminal justice system, is needed. Without these pieces, our understanding of attrition remains limited.

Table 5.

Key Recommendations for Research, Practice, and Policy.

Recommendations
For Research
Collect updated data (post early 2010s), especially in rural locales and outside of the Midwest and Southern CA	When possible, be transparent with sampling procedures and reporting: include n in addition to % for each outcome, victim age range, time period data were collected, and study location
Report each outcome as a single variable (e.g., charged, convicted vs. “charged and convicted”), and disaggregate data by age of victim	Consider examining cases outside of “one female offender and one male offender,” include multiple offenders, female offenders, and male victims, and sexually minoritized individuals
Report what offender was charged with and sentenced to; include whether charges were for sex crimes and/or charged down to lower offense	Conduct more research on founding and conviction outcomes, as we know the least about these
For Practice and Policy
Given that multiple stakeholders make up the criminal justice process, we need to focus on multi-pronged interventions. Consider implementing SARTs. Consider training police and prosecutors together to allow for conversations about “downstreaming” in real time. Consider policy change that helps shift norms, so prosecutors are not afraid to move cases forward due to jury or voter rape myths	We need to reconsider federal data collection and reporting expectations. For example, when accounting for sex of victim or offender, NIBRS lists three options: male, female, unknown. This needs to be expanded, especially as officers can note whether an offense was motivated by an offender’s bias (“bias motivation”). This category includes bias directed toward transgender individuals, yet officers do not have a place to capture whether the victim is transgender

Note. SART = sexual assault reponses team.

Relatedly, founding is the first decision the police must make. A case is founded if preliminary evidence meets the definition of a crime. While some cases should be unfounded because they do not meet the definition of the legal statute, other cases may be incorrectly unfounded as police founding decisions are known to be influenced by extralegal factors (Hoffman et al., 2023). We to need monitor the number of cases that are unfounded as these cases do not proceed to a full investigation and are instead closed. We recognize gaining access to unfounded data can be difficult as police do not always include these data in their reports. Moreover, NIBRS and UCR do not collect or report founding. The authors recently published a study on founding and obtained these data by asking the study’s police department. We were informed these data were public and did not require a data agreement, but that the department does not publish unfounded data on its website. Relatedly, the end goal for survivors pursuing their case in the criminal justice system is conviction. To increase the possibility of survivors reaching their desired outcome, we need to focus more on the last stage of conviction. We encourage researchers to reach out to police departments and prosecutor offices to identify whether founding and conviction data are available for use.

Third, researchers are encouraged to go beyond simply reporting whether charges were brought, and a conviction was obtained. This review originally intended to examine what offenders were charged with (e.g., misdemeanor vs. felony) and sentenced to (e.g., prison, probation, jail), including when charging and conviction were for sex crimes and if crimes were charged down to a lower offense. However, we had to keep charging and conviction broad because so few studies published these details. Expanding charging decisions and sentencing outcomes in future research will help clarify how the criminal justice system responds to sexual assault and where interventions may be needed. See Spohn et al. (2001) and Wentz (2020) as helpful examples that go more in-depth about charging and sentencing decisions.

Fourth, it is clear new, updated data sources are needed to improve our estimates of sexual assault case outcomes. More research in other parts of the country (besides the Midwest and Southern CA) is needed, particularly from rural locales. Case attrition may look exceptionally different across diverse communities due to variations in police and prosecutor office size and specialization (e.g., detectives and prosecutors dedicated to sex crimes or not), community norms and beliefs, and other community-level factors. For example, some sexual assault stakeholders have described their rural communities as posing additional barriers to responding to sexual assault because residents believe sexual assault does not happen there (Greeson et al., 2018). Given the potential for differences in rural and urban communities to affect the response to sexual assault cases, it is important that more research is done on case attrition in rural communities.

Moreover, most research in this review included one female victim and one male offender. While this makes sense given the gendered nature of sexual assault, this is still a significant limitation. For example, we know people who identify as gay, lesbian, or bisexual are assaulted at high rates and may be at increased risk for sexual violence victimization compared to heterosexual individuals (Rothman et al., 2011), but we do not know how the criminal justice system responds to these cases. LGBTQ+ victims may be particularly reluctant to report to police due to fears their cases will not be taken seriously because of cis-heterosexism (Murphy-Oikonen & Egan, 2022). Might research on case outcomes suggest attrition is worse for these groups? We are unable to draw conclusions about questions like these given the cisheteronormativity of current research. For example, almost half (k = 14) of the studies in Table 1 (reported cases) sampled and/or analyzed cases in which the victim was female and the offender was male. We need more inclusive sampling, recruitment, measures, and data analysis practices. Echoing extant calls (McCauley et al., 2019), we encourage scholars, including those conducting ongoing federal projects, to report other types of sexual assault cases outside the male-female binary when possible. This also includes cases with multiple victims and offenders, female offenders, male victims, and sexual minoritized individuals.

Additionally, except for R.Campbell, Greeson, et al. (2012), studies tended to combine cases involving adolescents with cases involving minors or with cases involving adults. Age and development are critical in sexual assault. They affect whether the victim can consent in that state and how the medical and legal system responds (e.g., forensic interview at a child advocacy center versus not, type of SANE exam, and how they testify). Further, adolescence is a unique developmental period that may affect how survivors are perceived by criminal justice personnel (e.g., Greeson et al., 2016). We encourage researchers to do more reporting of disaggregated data based on victim age and to provide justification for their decision based on contextually relevant factors (e.g., above vs. below the state’s age of consent) whenever possible. Finally, we noticed gaps in reporting that limited our ability to include studies and interpret findings. We strongly encourage researchers to expand reporting practices. Report n, in addition to the percent for each outcome; report victim age range; report the time period data were collected; and when possible, report the location where the study took place. This information will improve our understanding of the context in which attrition data are collected and factors relating to attrition rates.

Implications for Practice and Policy

While this review does not provide sufficient data to demonstrate how to increase attrition, it does suggest the need for continued attention to the problem. We reiterate other calls in the literature for trauma-informed training of criminal justice stakeholders, as data suggest these are promising practices in certain circumstances (B. A.Campbell et al., 2022). In addition, we echo others’ calls for further multidisciplinary coordination that educates and encourages stakeholders to adopt best practices in the cross-system response to sexual assault (e.g., linking survivors to other systems seamlessly, and conducting a thorough investigation of all cases). Again, some data suggest a multidisciplinary approach can be helpful (Greeson & R.Campbell, 2015). However—although the current review cannot and does not test individual strategies for reform—the persistence of high attrition after such reforms have become common suggests that they may not have had sufficient impact (yet).

The criminal justice process is a complex multistakeholder process and is affected by “downstreaming” such that stakeholders earlier in the process may not move cases forward based on what they think stakeholders will do in later stages of the process (Pattavina et al., 2021). For example, prosecutors may not take cases they think will be perceived negatively by juries, thus handing cases back to police to be exceptionally clear. Correspondingly, police may not refer cases they think the prosecutor is unlikely to charge (Pattavina et al., 2021). In part, this may be due to the perception, or reality, that there are not sufficient resources to investigate and process all cases. Focus groups with both police and prosecutors would help illuminate their experiences with formal and informal consultation. It may be that reformists need to target specific jurisdictions holistically through multi-layered intervention that involves reallocating resources toward gender-based violence cases, trauma-informed training, developing new procedures that require thorough evidencebased investigation of all cases before making referral decisions, checks and balances to ensure procedures are followed, multidisciplinary coordination to enhance victim support and access to resources (e.g., medical forensic evidence collection), and broader community-wide norm shifts such that prosecutors are not afraid to move cases forward due to juries or voters exhibiting rape myths. State policies may also provide support for such efforts. For example, the Illinois Sexual Assault Incident Procedure Act (2017) requires police to file a report for all sexual assault crimes. Given the complex nature of the system, it may be that a combination of multilevel, multistakeholder interventions aimed at shifting knowledge and values, as well as resources and procedures, is needed to create bigger changes in attrition.

The review also suggests the need for policy officials to reconsider federal data collection and reporting expectations. For example, when accounting for sex of the victim or offender, NIBRS lists three options: male, female, and unknown. The field should expand this, especially as officers can note whether an offense was motivated by an offender’s bias (“bias motivation”). This category includes bias directed toward transgender individuals, yet officers do not have a place to capture whether the victim is transgender. More broadly, we encourage federal funders (e.g., NIJ, OVW) to expect researchers to adhere to the practices we suggested (Table 5). Finally, more funding may be needed to be able to fully implement and assess multilevel interventions aimed at reducing attrition.

Limitations

This review has several limitations. First, we accepted all descriptions of sexual assault reported in the 36 studies. Doing so allowed us to collect and examine a broad inventory of sexual assault; however, our outcomes are unable to capture more nuanced definitions of sexual crime. Second, there was a great deal of variability in sampling, reporting, and analysis for studies included in this review. This variability speaks to the many ways researchers can study and interpret findings on attrition. For example, our reported cases category includes some studies that had additional inclusion criteria (e.g., completed SAK and offender identified) while others were truly all reported cases. This allowed us to catalog a wide array of sample characteristics but affected the interpretation of our findings.

Third, our stratification by age was intended to advance our understanding of how police and prosecutors respond to sexual assault. However, very few child-adolescent studies were included in this review. This was, in part, due to a dearth of literature. However, we were also unable to use all outcomes for several studies because the initial investigative stage was performed by individuals other than police (e.g., Child Advocacy Centers [CACs] and Department of Children and Family Services [DCFS]) and are not comparable to cases in which police investigate alone. Relatedly, we were unable to capture attrition as a result of transferring cases to the juvenile criminal justice system. Fourth, while we screened over 1,000 records, we may have missed some studies given the range of disciplines, language, and keywords germane to this research area. Finally, because this review only focused on case outcomes, it is limited in explaining why some studies had higher versus lower rates of various outcomes. Future research is encouraged to examine predictors of attrition at different stages of the criminal justice system to complement these findings.

Conclusion

Sexual assault case attrition continues to be problematic in the U.S. criminal justice system. Results from the present review indicate the majority of cases drop off between founding and arrest, and very few cases reported to police result in conviction. While these systematically obtained rates are helpful in giving us a more accurate picture of attrition, they are limited by the methodologies and information reported in extant studies. Researchers are encouraged to report their sampling strategies in more detail and pursue understanding why cases are dropped. Still, it is clear that police and prosecutors are not always responding to sexual assault cases in meaningful ways. Much more work is needed to ensure practitioners, policymakers, and the criminal justice system as a whole are effective in meeting victim’s needs and—ultimately—obtaining justice for this crime.

Biographies

Erin E. Hoffman, MA is a PhD candidate in the clinicalcommunity psychology program at DePaul University. She studies community response to sexual violence within the medical and criminal legal systems. Her current work examines police and prosecutor response to sexual assault with particular emphasis on case processing decisions.

Emily Patton, BA. is a student in the clinical psychology Ph.D. program at Georgia State University. She studies sexual assault prevention and brief, scalable interventions targeting sexual assault sequala. Her current focus is on maladaptive coping mechanisms post-sexual assault, such as alcohol, to cope with distress.

Megan R. Greeson, Ph.D. is an associate professor of community psychology at DePaul University. She conducts collaborative community-based research and evaluation to understand and improve how individuals, organizations, and communities address gender-based violence. Her current projects focus on interdisciplinary collaboration in responding to sexual assault.