Race and the Criminal Justice System Response to Sexual Assault: A Systematic Review.

Prior research has consistently documented that the vast majority of sexual assault cases do not progress through the criminal justice system. However, there is less agreement in prior work on how race influences case progression, resulting in a literature frequently described as "inconsistent." This systematic review examines all prior research that has included race as an independent variable in predicting the criminal justice system response to sexual assault (N = 34) in an effort to provide insight into seemingly disparate findings. We assess each study for the degree to which race was a focal point of interest, if and what theory was used to inform the investigation of race, how samples were drawn, and how and whose race was measured. Results illustrate that findings in prior research are not inconsistent, but rather unite to tell a nuanced story of the role of race in the criminal justice system response to sexual assault. The review demonstrates how decisions made by researchers throughout the research process can have significant impacts on reported findings, and how such findings may be used to influence policy and practice.

Introduction

In 2006, Tarana Burke coined the phrase, “Me Too,” as a means to let survivors of sexual assault (SA), particularly women and girls of color, know that they are not alone; and that in working together, they may find hope, support, and inspiration (Ohlheiser, 2017). About a decade letter, #MeToo emerged as a viral awareness campaign, with millions taking to Facebook, Twitter, and other venues to share their experiences of sexual violence and to join in community with one another. The “Me Too” conversation reemerged and continued amid a backdrop of reports of SA at the hands of many prominent figures (Johnson & Hawbaker, 2019); the Larry Nassar sexual abuse investigations (Hauser & Astor, 2018); the discovery and documentation of hundreds of thousands of untested SA kits nationwide (Campbell, Feeney, Fehler‐Cabral, Shaw, & Horsford, 2017; Reilly, 2015); the election and inauguration of President Trump and the worldwide Women's March in response to it (Smith‐Spark, 2017); and the nomination, Senate Judiciary Committee's hearing on SA reports against, and subsequent swearing in of Supreme Court Justice Kavanaugh (Stolberg, 2018). While the public discourse on sexual violence surged, so did a parallel, though largely separate conversation on race and the criminal justice system (CJS). The acquittal of George Zimmerman for the killing of Trayvon Martin (L. Alvarez & Buckley, 2013); the killings of Eric Garner, Mike Brown, Philando Castile, and many others by police officers on duty (Hafner, 2018); the continued mass incarceration of African American men and women (Tucker, 2017); and the mobilization of the Black Lives Matter movement (Thomas & Zuckerman, 2018) have renewed a focus on the experiences and interactions of individuals of color and the CJS, with a particular focus on the extent to which such experiences are defined by systemic racism.

Though recent public discourse on race and the CJS largely occurs in separate, distinct spaces from discussions on sexual violence, much research has examined their intersection. A sizable body of literature has investigated the quite complex CJS response to SA, and how cases move through it (see Lonsway & Archambault, 2012; Spohn & Tellis, 2012 for reviews). For a SA case to progress in the CJS, it must first be reported to police. Police may then conduct an investigation and refer the case to the prosecutor's office for the consideration of charges against an identified suspect. In some jurisdictions, interactions between police investigators and the prosecutor's office happen earlier on in the process, as some investigators routinely screen all cases with the prosecutor's office shortly after the assault is reported. The police frequently also interface with the medical system and crime laboratory during their investigation. Police may assist the victim in accessing a medical forensic exam, and transport completed SA evidence collection kits from the medical facility where they are completed to crime laboratories where they may be analyzed and used in the course of the criminal investigation and potential future prosecution. Once a case is referred to the prosecutor's office, the prosecutor may choose to file charges. The charges may be dismissed prior to prosecution, the defendant may plead guilty, or the case may go to trial. If the case goes to trial, it may result in a guilty verdict, or an acquittal. If the defendant is convicted via a guilty plea or conviction at trial, the defendant may be incarcerated, or receive some other sentence (e.g., parole, fine). Prior research on the CJS response to SA has documented how fewer and fewer cases reach each juncture in this system. Lonsway and Archambault (2012) call this the “funnel of attrition;” for every 100 forcible rapes committed, an estimated 5–20 are reported to police, 0.4–5.4 are prosecuted, 0.2–5.2 result in a conviction, and 0.2–2.8 result in incarceration of the offender (p. 158).

Beyond documenting high rates of attrition among SA cases in the CJS generally, prior research has also attempted to understand what factors distinguish those cases that do progress from those that get left behind. For example, studies have examined what predicts some cases resulting in an arrest, while others do not (e.g., Addington & Rennison, 2008); some cases being prosecuted, while others are not (e.g., see Campbell, Patterson, Bybee, & Dworkin, 2009); some cases resulting in a conviction, while others end in an acquittal or dismissal (e.g., Maxwell, Robinson, & Post, 2003); and how convicted offenders are sentenced (e.g., Curry, 2010). Race is frequently included as one such factor. However, while there is general agreement that the vast majority of cases do not progress through the CJS, there is less agreement across studies in how race influences case progression. Prior research on the influence of race on the CJS response to SA has been described as “mixed” (Shaw, Campbell, & Cain, 2016, p. 458), “inconsistent” (Spohn & Tellis, 2012, p. 176), and containing “contradictions” (Maxwell et al., 2003, p. 524). Thus, the purpose of this review is to examine prior research on race and the CJS response to SA so as to try and resolve seeming disparities in prior work and provide empirical information that may be used to catalyze and inform public discourse focused on the intersection of race, gender, sexual violence, and the CJS response to it. Through a thorough examination of how race is conceptualized, theorized, measured, and discussed, we hope to help tell a more cohesive, nuanced narrative that can contribute to this ongoing public discussion and perhaps help inform change initiatives therein.

Method

Literature Search

The empirical studies included in this review were identified through library and online databases, as well as checking the reference lists of identified articles for additional relevant studies (see Booth, Papaioannou, & Sutton, 2012). Specifically, we searched Web of Science, PsycINFO, Social Service Abstracts, Applied Social Sciences Index and Abstracts (ASSIA), and International Bibliography of the Social Sciences (IBSS) using the key word “race” in combination with “rape” or “sexual assault,” and “police,” “prosecution,” “law enforcement,” or “criminal justice.” After removing duplicates across databases, our initial search yielded 196 records. Based on the first author's existing knowledge of the literature on the CJS response to SA, the abstracts of an additional thirteen records were added.

Selection and Data Abstraction of Included Articles

Each of the authors reviewed independently the abstracts for the 209 records to assess if they should be included in the review (see Fig. 1). To be included, the abstract must have described a study that met the following inclusion criteria: an original empirical article published in a peer‐reviewed journal through 2016, that used quantitative methods to examine factors that predict some aspect of the CJS response to SA, with actual data on the CJS response (e.g., not surveys of individuals’ perceptions, review of vignettes, mock juror reactions, etc.). The authors then met to review their independent assessments, and discussed any discrepancies until consensus was reached. We identified 39 records for potential inclusion (28 from the initial database search and 11 from the first author's supplementary list of articles).

Figure 1

Flowchart of search results

The full‐text articles for the 39 records were reviewed in their entirety to ensure each study met the above inclusion criteria. In addition, each article needed to include race (of the victim, the suspect, or both) as a predictor variable in the quantitative analysis and be written in English. Following independent assessment and group discussion, we identified 27 articles for inclusion in this review. A check of the references provided in each of these 27 articles identified an additional 66 references for potential inclusion. We followed a similar process as described above, reviewing the abstracts for each of these 66 references, then the complete articles to ensure race was included as an independent variable. We identified ten additional articles for inclusion through these means, yielding 37 articles for inclusion in this review. Once coding was underway, however, we excluded one article due to multiple errors identified in the manuscript, some pertaining directly to the purpose of this review. Two additional articles were removed as they only provided descriptive statistics; no inferential tests were performed.

The first author reviewed and coded the final sample of N = 34 studies for the extent to which race was a focal point of interest (i.e., race was a focal variable of interest, one of a battery of variables of interest, or treated as a control variable); extent to which theory was used to inform the investigation of race (i.e., the specific theory or theories used and provided explanations for theory selection); sample selection and composition (i.e., inclusion criteria, sample size, sample locales, samples’ racial composition, and data sources); how race was coded (i.e., race of the victim, perpetrator, or both; and race categories used); and outcomes of interest and research findings (i.e., the dependent variables explored and reported findings).

Results

Table 1 describes the outcome of interest; sample; if race was a focal, control, or variable of interest; if and what theory was used to guide the examination of race; race measures; and race findings for each of the included articles. Included studies were published between 1961 (Bullock, 1961) and 2016 (Shaw et al., 2016), relying upon data and administrative records collected as early as 1945 (Wolfgang & Riedel, 1975), and as recently as 2009 (Shaw et al., 2016). All studies were conducted in the United States. Unless otherwise noted, in discussing each article, we use the same language presented in the original article for terms related to race (e.g., “Victims of Color,” “non‐White victims,” “Negroes,” etc.). We maintain the specific language and terms used by the authors in their original manuscripts as the language we use to discuss race provides insight into how we consider and socially construct race.

Table 1

Description of included articles

Author (year)	Outcome of interest	Samplea	Racial composition of sampleb	Theory to examine race	Race measurec	Race findings
Addington and Rennison (2008)d	Rape clearance: Arrest or clearance by exceptional means	22,876 female rape victims in the FBI Uniform Crime Reporting (UCR) Program's National Incident Reporting System (NIBRS) data from 2002	Victim 76.9% White 20% Non‐White 3.1% Unknown/missing Suspect 60.5% White 28.5% Non‐White 11.1% Unknown/missing	None	Victim Non‐Hispanic White or Notb Suspect White or Non‐White	Cases with Non‐Hispanic White victims had a lower odds (OR = 0.88) of being cleared by arrest or exceptional means. Cases with White suspects had a higher odds (OR = 1.13) of being cleared by arrest or exceptional means.
Alvarez and Bachman (1996)e	Sentence length: Number of years. Life and death sentences recoded to reflect 15 and 20 years above the longest sentence received	2,204 active inmates aged 16–72 years old, assumed to be representative of average Arizona state correctional population in 1990	Suspect 95% Caucasian 5% American Indian Other racial/ethnic groups excluded	ict Theory	Suspect Caucasian or American Indian Other racial/ethnic groups excluded	Caucasians had an average sentence of 26 years compared to 19 years for American Indians. This was not statistically significant.
Beichner and Spohn (2012)d	Prosecutor's charging decision: Charges filed vs. no charges filed	666 reported rape cases from police agencies in Miami, Kansas City, and Philadelphia that involve female victims at least 12 years old with a corresponding arrest and referral to the prosecutor's office from 1996 to 1998	Victim 65.6% non‐White 34.4% White Suspect 76.8% non‐White 23.3% White	None explicitly identified for race examination. Focal Concerns Theory used for overall investigation.	Victim White or non‐White Suspect White or non‐White	White victims were more likely to have charges filed in cases of aggravated rape (i.e., suspect and victim were stranger, suspect used gun or knife during assault; or victim suffered collateral injuries). No effect of victim race or suspect race in simple rapes (i.e., no aggravating factors).
Bouffard (2000)e	Founding: Founded vs. unfounded by police Case closure: Among founded cases, case closure category	326 felony SA cases drawn from the investigative files of one urban/suburban agency in 1995	Victim 76.6% African‐American 18.3% White 6.1% Other Suspect 85.1% African‐American 9.8% White 5.1% Other	Theory of Law	Victim White or Notb Suspect African‐American or Notb Dyad White victim and African‐American suspect or Notb	No effect of victim race, suspect race, or race dyad.
Bradmiller and Walters (1985)e	Severity of charge: Rape (more severe charge) vs. nonrape (less severe charge)	89 cases of charged SA being evaluated by the Court Psychiatric Center in Hamilton County, OH from January 1973‐May 1979 that involved offenders with no prior conviction for SA; included charges that were still considered crimes at the time of the study (1985); did not have a more severe charge (e.g., aggravated murder); and were not missing data on variables of interest (e.g., force used)	Suspect 60% Black 40% White	None	Suspect Black or white	Race accounted for 5% of the variance in the dependent variable—severity of charge. Given similar crimes, black offenders were more likely to be charged with more serious crimes (i.e., rape as opposed to nonrape).
Briggs and Opsal (2012)e	Arrest: Suspect arrested vs. no suspect arrested	226,496 cases reported to FBI UCR NIBRS by 1,490 police departments in 30 states for offences that occurred in 2008; SA cases are one type of offense included	Victim 68.1% white non‐Hispanic 20.6% black non‐Hispanic 11% white Hispanic 0.3% black Hispanic Other racial/ethnic groups excluded	Theory of Law Procedural Justice	Victim White or black Hispanic or Non‐Hispanic Other racial/ethnic groups excluded	Cases involving Hispanic victims were more likely to conclude in arrest than non‐Hispanic victims, though this relationship was not statistically significant.
Bullock (1961)e	Sentence length: short (less than 10 years) vs. long (ten years and longer)	3,644 inmates committed for burglary, rape, and murder to the Texas State Prison at Huntsville in 1958	Suspect 68.2% white 31.% Negro Other racial/ethnic groups excluded	None officially named. Theory of Law concepts presented.	Suspect White or Negro Other racial/ethnic groups excluded	Negro offenders are more likely to receive short sentences, even after considering geographic region, plea bargains, and number of previous felony convictions.
Campbell et al. (2009)f	Case outcome: Not referred by the police for prosecution vs. referred but not charged vs. charged but later dropped or acquitted vs. conviction at trial or guilty plea	137 adult SA cases reported to police and treated by a SA nurse examiner program in a geographically diverse county in the Midwest from 1999 to 2005	Victim 86% Caucasian 14% racial/ethnic minority	None	Victim Caucasian or racial/ethnic minority	No effect of victim race.
Chandler and Torney (1981)e	Case outcome: charged vs. released/dropped. If charged, plea bargain vs. trial	408 SA victims treated by the Sex Abuse Treatment Center of a large, urban hospital between October 1976 and September 1978	Not provided	Conflict Theory	Victim Caucasian or non‐Caucasian or Hawaiian Suspect Caucasian or non‐Caucasian or Hawaiian Dyad Both Caucasian or both non‐Caucasian or Defendant non‐Caucasian/Victim Caucasian or Defendant Caucasian/Victim non‐Caucasian	Caucasian suspects are underrepresented and Hawaiian suspects are overrepresented among those indicted. Caucasian victims are overrepresented among those seeing their assailant indicted. Cases with Caucasian victims and Non‐Caucasian suspects are more likely to result in a charge than to have the suspect released. Caucasian defendants are overrepresented in the plea bargaining process and Hawaiian defendants are overrepresented among those going to trial.
Curry (2010)e	Sentence length: Number of years	241 randomly selected male offenders who are at least 18 years old and convicted of felony SA against a female in the seven largest counties in Texas between January and September 1991	Victim 43% White 31% Black 26% Hispanic Suspect 41% Black 31% White 28% Hispanic Other racial/ethnic groups excluded	Conflict Theory Focal Concerns Theory Blameworthiness Attribution Bounded Rationality	Victim Non‐Hispanic White or non‐Hispanic Black or Hispanic Suspect Non‐Hispanic White or non‐Hispanic Black or Hispanic Dyad Offender Black/Victim White or Offender Hispanic/Victim White or Offender White/Victim White or Offender or Offender any race/Victim Black or Hispanic Other racial/ethnic groups excluded	Hispanic offenders receive shorter sentences than White offenders. No effect of victim race or race dyad.
Frazier and Haney (1996)f	Suspect interview: Suspect questioned vs. no suspect questioned Prosecutor's charging decision: Charges filed vs. no charges filed	569 criminal sexual conduct cases involving females victims over the age of 15 reported to a Midwestern metropolitan police department in 1991	Victim 55% White 31% Black 10% Native‐American 3% Other or Unknown	None	Victim Caucasian or Notb Suspect Caucasian or Notb	No effect of victim race or suspect race.
Gray‐Eurom et al. (2002)f	Conviction: Guilty verdict vs. not guilty verdict Prosecutor's charging decision: Charges filed vs. no charges filed	355 SA patients who received a forensic exam at the Adult and Adolescent SA Program in Duval County, Florida between October 1993 and September 1995; reported to police; had a suspect identified; and a known legal conclusion	Victim 51% black 47% white 2% other minorities Suspect 32% black 10% white 58% examiner did not record offender race	None	Victim White or Notb	Cases with white victims trended toward conviction, though the relationship was not statistically significant. Cases with victims who were not white trended toward being dropped, though the relationship was not statistically significant.
Holleran et al. (2010)d	Prosecutor's charging decision: Charges filed vs. no charges filed	368 reported rape cases from police agencies in Kansas City and Philadelphia involving male offenders, female victims at least 12 years old, an initial police charge of forcible rape, and a corresponding arrest and referral to the prosecutor's office from 1996 to 1998	Not provided	Focal Concerns Theory	Victim White or Black Suspect White or Black Other racial/ethnic groups excluded	No effect of victim race or suspect race.
Horney and Spohn (1996)d	Suspect identification: Suspect identified vs. no suspect identified Case outcome: closed by police vs. referred to prosecutor but declined vs. accepted by prosecutor but later dismissed vs. prosecuted and not‐guilty verdict; vs. prosecuted and guilty	259 reported criminal sexual conduct cases randomly selected from the daily complaint log of the Detroit Police Department's Sex Crimes Unit in 1989 with a female complainant at least 16 years old, initial charges of first or third‐degree criminal sexual conduct, and not unfounded.	Victim 88.8% black 11.2% white Suspect 93.8% black 6.2% white	None	Victim Black or white Suspect Black or white	Cases with black victims are more likely to have a suspect identified. No effect of victim or suspect race on case outcome.
Kerstetter (1990)f	Founding: Founded vs. unfounded by police Prosecutor's charging decision: Charges filed vs. no charges filed Victim decline prosecution: Victim declined prosecution vs. did not decline Suspect line‐up: suspect line‐up completed vs. no suspect line‐up Arrest: Suspect arrested vs. no suspect arrested	1,530 founded rape cases reported to and founded by the Chicago police department in 1979 and 671 randomly selected cases from all SA complaints made by women to the Chicago police department in 1981	Not provided	Conflict Theory	Victim Not provided Suspect Not provided Dyad Not provided	No effect of victim, suspect, or dyad race.
Kingsnorth et al. (1998)e	Prosecutor's charging decision: Case charged and moves forward to prosecution vs. rejected or charges or dismissed Plea: Went to trial vs. guilty plea Incarceration Type: Jail vs. prison Sentence length: Number of years	365 rape cases forwarded to the district attorney's office by Sacramento police and county sheriff's department from 1992 to 1994	Dyad 34.5% White suspect/White victim 32.1% Black suspect/Black victim 21.4% Black suspect/White victim 12.1% Hispanic suspect/White victim Other racial/ethnic combinations excluded	Conflict Theory Structural Contexts Sexual Stratification Hypothesis	Dyad White suspect/White victim or Black suspect/White victim or Hispanic suspect/White victim or Black suspect/Black victim Other racial/ethnic combinations excluded	Cases with White suspects/White victims are arrested, on average, on a greater number of counts than cases with Black suspects/White victims and cases with Hispanic suspects/White victims. No effect of race dyad on prosecutor's charging decision, plea, incarceration type, or sentence length.
LaFree (1980a)e	Arrest: Suspect arrested vs. no suspect arrested Seriousness of crime: Mean prison term imposed by statute if convicted Felony charge: felony charges filed vs. no felony charges filed Dismissal: Prosecutor's decision to dismiss charges vs. pursue charges Plea: Went to trial vs. guilty plea Conviction: convicted by a guilty plea or guilty verdict at trial vs. no conviction Sentence type: Executed vs. non‐executed (i.e., combination of probation, suspended sentence, and fine) Incarceration Type: For executed sentences, incarceration at state penitentiary or elsewhere (less serious) Sentence length: Number of years	870 suspects forcible sex offenses of female victims reported to police in a large Midwestern city between January 1970 and December 1975	Dyad 44.7% black suspect/black victim 32.3% white suspect/white victim 23.0% black suspect/white victim	Conflict Theory Sexual Stratification Hypothesis	Dyad black suspect/black victim or white suspect/white victim or black suspect/white victim	From case report through final sentencing, the percentage of black intraracial assaults steadily declined, the percentage of white intraracial assaults remained relatively constant, and the percentage of cases involving black suspects/white victims steadily increased. Cases with black suspects/white victims were more likely to receive more serious charges; be filed as felonies; result in executed sentences; result in incarceration in the state penitentiary; and result in longer sentences. No effect of race dyad on arrest, dismissal, plea, or conviction.
LaFree (1980b)f	Plea: Went to trial vs. guilty plea Guilty verdict: Among those who go to trial, guilty verdict vs. not guilty verdict Conviction: convicted by a guilty plea or guilty verdict at trial vs. no conviction	124 forcible rape cases involving female victims adjudicated by the courts of a large Midwestern city in 1970, 1973, and 1975	Victim 59.7% white 40.3% black Suspect 67.7% black 32.3% white Dyad 40.3% black suspect/black victim 31.5% white suspect/white victim 28.2% black suspect/white victim Two victims and one suspect were non‐white other than black. They were grouped in with black. One case involved white suspect/black victim. This case was added to black suspect/black victim.	None explicitly identified for race examination. Conflict Theory and Labeling Theory for overall investigation.	Victim white or black Suspect white or black Dyad black suspect/black victim white suspect/white victim black suspect/white victim Two victims and one suspect were non‐white other than black. They were grouped in with black. One case involved white suspect/black victim. This case was added to black suspect/black victim.	Cases with black suspects were more likely to go to trial than result in plea bargain. Cases with black victims were less likely to result in a conviction via a plea bargain or a guilty verdict at trial.
LaFree (1981)f	Arrest: Suspect arrested vs. no suspect arrested Seriousness of crime: Mean prison term imposed by statute if convicted Felony charge: felony charges filed vs. no felony charges filed	905 forcible sex offenses with female victims reported to police in a large Midwestern city in 1970, 1973, and 1975	Victim 54.6% white 45.4% black Suspect 67% black 33% white Dyad 44.1% black suspect/black victim 33% white suspect/white victim 22.9% black suspect/white victim Non‐white suspects and victims who were not black (less than 1%) were grouped in with black. Cases with white suspects/black victims (1.2%) excluded.	None	Victim white or black Suspect white or black Dyad black suspect/black victim or white suspect/white victim or black suspect/white victim Non‐white suspects and victims who were not black (less than 1%) were grouped in with black. Cases with white suspects/black victims (1.2%) excluded.	Once a SA unit was implemented, black intraracial cases were significantly less likely to have an arrest than White intraracial cases. Cases with black suspects/white victims had more serious charges and were more likely to be charged as felonies.
Maxwell et al. (2003)e	Pretrial release: Defendant released vs. detained Prosecutor's charging decision: Arrest charge maintained or elevated by prosecutor vs. downgraded to a lesser offense Conviction: By a guilty plea or guilty verdict at trial vs. not guilty at trial or dismissal Incarceration: Incarceration vs. no incarceration Incarceration Type: Prison vs. jail, probation, or other type of sentence Sentence Length: Expected maximum number of months	4,050 arrested defendants associated with filed felony SA cases selected from a stratified sample of the Nation's 75 most populous counties in the National Pretrial Reporting Program (nPRP), also known as the State Court Processing Statistics (SCPS), in 1990, 1992, 1994, and 1996	Suspects 49% African‐American 38% White 10% Hispanic 3% Asian	Consensus Theory Conflict Theory Lotz & Hewitt's Five Models	Suspect White or African‐American or Hispanic or Asian and other races	No effect of suspect race on pretrial release, prosecutor's charging decision, incarceration. African‐American and Hispanic suspects were significantly less likely to be found guilty than White suspects. Asian suspects are somewhat more likely to be found guilty than White suspects, though this does not reach statistical significance. African‐American and Hispanic suspects were significantly less likely to be sentenced to prison than White suspects. White suspects convicted of SA receive significantly longer sentences than all other groups, regardless of their prior criminal history or their current criminal justice status.
Roberts (2008)f	Clearance by arrest: Case cleared by arrest vs. not cleared by arrest	11,215 forcible rape incidents reported in the 2000 FBI UCR NIBRS from 10 cities with populations between 25,000 and 660,000 across seventeen states	Victim 61.87% White 34.28% non‐White 3.85% Missing	Conflict Theory Theory of Law	Victim White or non‐White	Cases with non‐White victims were more likely to be cleared by arrest.
Shaw and Campbell (2013)f	Rape Kit Submission: Rape kit was submitted to the crime lab vs. rape kit not submitted	393 reported SA cases involving 13–17 year old victims treated by two SA nurse examiner programs in two Midwestern communities from 1998 to 2007	Victim 80.3% White 14.9% African‐ American 3.0% Bi/Multiracial 1.0% Latino/a 0.8% Other	None	Victim White or non‐White	Cases with non‐White victims were nearly twice as likely to have their kit submitted.
Shaw et al. (2016)e	Investigative steps: Number of investigative steps completed Arrest: Suspect arrested vs. no suspect arrested Referral: Case referred to the prosecutor for the consideration of charges vs. no referral to the prosecutor	248 reported SA cases corresponding to randomly selected, largely unsubmitted SA kits that accumulated in police property in a large, urban, predominately Black/African‐American, Midwestern City from 1980 to 2009	Victim 86.3% Black 13.3% White 0.4% Other Suspect 91.9% Black 5.6% White 0.4% Other 2.0% Missing	Social Dominance Theory	Victim White or of color Suspect White or of color All but one victim and one suspect of color were Black/African‐American.	Victims of color were more likely to be deemed uncooperative, without a phone/means of contact, without information, or weak, which then meant a case was likely to have fewer investigate steps completed and was less likely to have an arrest and referral to the prosecutor's office. No effect of suspect race.
Spears and Spohn (1997)d	Prosecutor's charging decision: Charges filed vs. no charges filed	321 SA cases presented to the warrant section of the county prosecutor's office in Detroit in 1989	Victim 82.5% black 17.5% white Suspect 85.6% black 14.4% white	None	Victim white or black Suspect white or black	No effect of victim or suspect race.
Spohn and Cederblom (1991)e	Sentence Type: Sent to prison vs. not sent to prison Sentence length: Of those sentenced to prison, number of days for expected minimum sentence	363 rape and 299 other sexual offense cases with a male offender that resulted in a conviction and sentencing of the offender from 1976 to 1978 based on Detroit's Recorder's Court records	Not provided for subsample of rape and other sexual offense cases. Entire sample, which included cases with offenders convicted of 11 different felony crimes, was 84.2% black and 15.8% white.	Liberation Hypothesis	Suspect white or black	Cases with black suspects were more likely to result in incarceration than cases with white suspects when it was a stranger assault. Race did not have an effect on incarceration for acquaintance assault. No effect of suspect race on sentencing length.
Spohn and Holleran (2001)f	Prosecutor's charging decision: Charges filed vs. no charges filed	526 reported SA cases involving female victims 12 years and older, male offenders, and that were referred to the prosecutor in Kansas City and Philadelphia from 1996 to 1998	Stranger Rapes 41% white victims 21% white suspects Acquaintance Rapes 31% white victims 24% white suspects Partner Rapes 32% white victims 23% white suspects	None	Victim white or Notb Suspect white or Notb	Cases with white victims were 4.5 times more likely to be charged when it was a stranger assault. Victim race did not have an effect on charging in non‐stranger cases.
Spohn and Horney (1993)ffor victim; C for suspect	Dismissal: Prosecutor's decision to dismiss charges vs. pursue charges Plea: Went to trial vs. plea to a lesser charge Conviction: By a guilty plea or guilty verdict at trial vs. not guilty at trial or dismissal Incarceration: Incarceration vs. no incarceration	812 rape cases charged by the prosecutor in Detroit from 1970 to 1984	Victim 73.0% black 27.0% white Suspect 87.9% black 11.2% white	None	Victim white or black Suspect white or black	Following rape law reform in 1975, cases with black victims were less likely to have charges dismissed or result in a plea bargain. No effect of suspect race on dismissal or plea. Following rape law reform in 1975, cases with black suspects were less likely to result in a conviction. No effect of victim race on conviction. No effect of victim or suspect race on incarceration.
Spohn and Spears (1996)e	Dismissal: Prosecutor's decision to dismiss charges vs. pursue charges Conviction: By a guilty plea or guilty verdict at trial vs. not guilty at trial or dismissal Incarceration: Incarceration vs. no incarceration Sentence Length: Among those incarcerated, maximum sentence imposed	1,152 SA cases involving female victims and male perpetrators charged by the prosecutor in Detroit from 1970 to 1984	Dyad 74% black suspect/black victim 13.3% black suspect/white victim 12.7% white suspect/white victim	Sexual Stratification Hypothesis	Dyad black suspect/black victim or black suspect/white victim or white suspect/white victim	Cases with black suspects/white victims were significantly more likely to be dismissed, though when convicted, resulted in longer sentences than other dyads. Cases with black suspects/white victims were less likely to be convicted than cases with white suspects/white victims. No effect of dyad race on incarceration. Among intraracial black dyads, stranger cases were significantly less likely to have charges dismissed, more likely to result in incarceration, and resulted in longer sentences than cases of acquaintance assault.
Spohn et al. (2001)f	Prosecutor's charging decision: Charges filed vs. no charges filed	140 rape cases involving victims 12 years old and up that were cleared by arrest in 1997 from the sex crimes bureau of the Miami‐Dade police department	Victim 58.1% black 31.3% white 10.9% Hispanic/other Suspect 64.9% black 18.9% white 16.2% Hispanic	Focal Concerns Theory	Victim white or Notb Suspect white or Notb	Charges were more frequently rejected with non‐white victims and suspects, though this was not statistically significant.
Spohn et al. (2014)f	Founding: Founded vs. unfounded by police	393 SA cases involving female complainants 12 years and up reported to the Los Angeles Police Department in 2008	Victim 47.6% Hispanic 28% white 19.1% black	None explicitly identified for race examination. Theory of Law and Focal Concerns Theory for overall investigation.	Victim white or black or Hispanic or Other	No effect of victim race.
Tellis and Spohn (2008)e	Founding: Founded vs. unfounded by police Prosecutor's charging decision: Charges filed vs. no charges filed Victim decline prosecution: Victim declined prosecution vs. did not decline	1,452 SA cases with victims 14 years old and up reported to the Sex Crimes Unit of the San Diego Police Department from 1991 to 2002	Dyad 35% White suspect/White victim 18% Black suspect/Black victim 15% Black suspect/White victim 15% Hispanic suspect/Hispanic victim 10% Hispanic suspect/White victim 4% White suspect/Hispanic victim 3% Black suspect/Hispanic victim Other racial/ethnic combinations excluded	Sexual Stratification Hypothesis Liberation Hypothesis	Dyad Black suspect/Black victim or Black suspect/White victim or Black suspect/Hispanic victim or Hispanic suspect/Hispanic victim or Hispanic suspect/White victim or White suspect/White victim or White suspect/Hispanic victim Other racial/ethnic combinations excluded	Cases with intraracial White dyads were more likely to be unfounded than cases with Hispanic suspects/White victims. No dyad race effect on prosecutor's charging decision or victim declining prosecution.
Walsh (1987)e	Sentence Type: Probation vs. prison Sentence severity: Scale with score determined based on points allocated for each year of probation (10), day in county jail (1), $25 fine (1), two days in work release (1), and day in state prison (1.1).	417 defendants sentenced on felony SA charges in a metropolitan Ohio county from 1978 to 1983	Suspects 58% white 42% black	Sexual Stratification Hypothesis	Victim white or black Suspect white or black Dyad white suspect/white victim or black suspect/black victim or black suspect/white victim Nine cases of white suspect/black victim were excluded.	Black suspects/white victims are four times more likely to be imprisoned as black suspects/black victims and twice as likely as white suspects/white victims. black intraracial dyads received significantly more lenient sentences than white intraracial dyads even though black intraracial dyads committed more serious crimes and had more significant prior records. black suspects who assault white victims, regardless of offender/victim relationship, received significantly harsher penalties than black suspects who assault black victims. Penalties are even more lenient for black suspects/black victims when the suspect and victim are acquainted.
Wolfgang and Riedel (1973)e	Sentence type: death sentence vs. other sentence	1,265 rape convictions from 1945 to 1965 in 230 counties in eleven southern and border states in which rape is a capital offense	Suspect 65.1% black 34.9% white	None	Suspect black or white Dyad black suspect/white victim or Notb	After accounting for two dozen possible nonracial variables, black defendants were sentenced to death at a rate seven times that of white defendants; black defendants with white victims were sentenced to death eighteen times more frequently than any other race dyad.
Wolfgang and Riedel (1975)e	Sentence type: death sentence vs. other sentence	361 rape convictions from 1945 to 1965 in Georgia	Not provided	None	Victim Not provided Suspect Not Provided Dyad black suspect/white victim or Notb	Black suspect/white victim cases are most likely to result in a death sentence. No effect of victim or suspect race alone.

SA, sexual assault.

Cases, victims, and suspects from crimes other than rape or SA were included in some study samples. The sample descriptions provided here are specific to those samples/subsamples for rape and SA. If a study included cases, victims, or suspects from crimes other than rape or SA and did not provide specific information on the subsample for rape and SA only, it is noted.

These variables were dichotomous. “Or Not” is not language used by the authors/researchers of each manuscript; it is our language to indicate a dichotomous variable.

In some studies, the racial composition information and race measures provided do not include exhaustive categories for all races (e.g., “black” and “white” are the only race categories reported or measured). When authors/researchers explained how other races were handled, it is noted. If this is not noted, that means it was not clear if other races did not appear in the sample, or if other races were intentionally excluded from the sample.

Race was a control variable.

Race was a focal point of interest.

Race was one of many variables of interest.

Race as a Focal Point of Interest

Of the thirty‐four articles reviewed, half (n = 17) included race as a focal point of interest in the investigation. For example, articles examined the potential discriminatory application of the law for “American Indian” compared to “Caucasian” perpetrators (Alvarez & Bachman, 1996); the main effect of victim “race/ethnicity” on sentencing outcomes (Curry, 2010); and the impact of “the racial composition of the victim‐defendant dyad” on police and prosecutor decision‐making (LaFree, 1980a). An additional twelve articles included race as one of many variables of interest, though not a focal point of the empirical investigation. The remaining five articles included race as a control. Some articles provided an explanation for this decision. For example, in examining SA case progression through the CJS, Horney and Spohn (1996) explain that while “the interaction of victim and offender race has been shown by some studies to be a significant determinant of case outcomes (LaFree, 1989; Walsh, 1987), [they] were not able to explore this interaction because 82% of the cases in this Detroit sample (90% of cases with suspects) were black‐on‐black offenses” (p. 141); thus, race was entered as a control. The majority of studies that included race as a focal point found at least one statistically significant race finding (n = 13; 76.5%), compared to about sixty percent of studies including race as a variable of interest (n = 7; 58.3%) or control (n = 3; 60%). In four studies, the original authors called attention to patterns in their data that suggested an influence of race, but that did not reach statistical significance. If these patterns are considered, 88.2% of studies that included race as a focal point (n = 15) report a relationship between race and the outcome variables; 75% of studies with race as a variable of interest (n = 9); and 60% of studies that included race as a control (n = 3).

Use of Theory to Explore Race

In half of the included articles (n = 17), the authors explicitly identified theory or conceptual models informing their empirical investigation of race. Another three articles discussed theory informing their overall investigation, though not specific to race; one article presented theoretical concepts that informed their investigation of race, but did not specifically name the theory as the theory was not yet developed; and the remaining thirteen articles provided no mention of theory in what guided their investigation. A total of thirteen different theories were used by authors to inform their empirical investigations. Regardless of if and how theory was used, about two‐thirds of the included articles had a least one statistically significant race finding (n = 11; 64.5% of articles with theory to guide race investigation; n = 2; 66.6% of articles with theory, though not explicitly tied to race investigation; and n = 9; 69.2% of articles with no mention of theory).

Conflict Theory

The theory most frequently used to inform empirical investigations of race was Conflict Theory (n = 8). Three articles used Conflict Theory alone, and five articles used Conflict Theory in combination with other theories or models (discussed below). One additional article used Conflict Theory to inform the overall investigation, though did not explicitly tie it to the inclusion of race in the study. In applying Conflict Theory to investigations of the CJS response to SA, authors explain how the very definition of what constitutes a crime, the extent to which a specific activity is treated as a serious crime, and how the CJS responds is determined by elites—powerful subgroups who occupy and want to maintain their favorable position in society (Maxwell et al., 2003). Crime and criminal justice processing is defined and designed in such a way to act as a means of control, help maintain existing power differentials, and ensure powerful subgroups retain their access to scarce resources while denying access to others (Chandler & Torney, 1981; Kerstetter, 1990; LaFree, 1980a). Race is one key determinant of the societal subgroup to which an individual belongs (Maxwell et al., 2003). In examining race and the CJS response to SA, we would expect to see “Black” crime victims devalued relative to “White” victims, leading to harsher penalties for those who harm “Whites” and more lenient sentences for those who harm “Blacks” (Curry, 2010). Not only are less powerful groups expected to be afforded less protection when they are victimized, but also to be more severely punished when suspected of perpetrating crimes (Maxwell et al., 2003).

Sexual Stratification Hypothesis

The second most frequently used theory was the Sexual Stratification Hypothesis (n = 5). Two articles used the Sexual Stratification Hypothesis on its own; one article used it in combination with Conflict Theory; one article in combination with the Liberation Hypothesis (discussed below); and another article used it in combination with both Conflict Theory and Structural Contexts (discussed below). The Sexual Stratification Hypothesis builds from Conflict Theory by conceptualizing sexual access as one of the scarce resources powerful subgroups intend to control (LaFree, 1980a). Individuals are classified into subgroups, or strata, based primarily on race and gender. Power differentials among these subgroups then determine sexual access to whom, by whom. The Sexual Stratification Hypothesis consists of a series of assumptions: (a) women are the valued and scarce property of the men of their own race; (b) “white” women are more valuable than “black” women; (c) the SA of a “white” woman by a “black” man is a dual threat, threatening both the “white” man's property rights and his dominant social position; (d) the SA of a “black” woman by any race does not threaten the status quo, and is thus less serious; and (e) “white” men have the power to sanction differently based on their social position and perceived threats to it (Walsh, 1987, p. 155). The Sexual Stratification Hypothesis requires an examination of the racial composition of the victim/perpetrator dyad, rather than examining victim or perpetrator race alone (Chandler & Torney, 1981; LaFree, 1980a; Tellis & Spohn, 2008; Walsh, 1987). In examining the CJS response to SA, Sexual Stratification Hypothesis suggests that “blacks who sexually assault whites” will receive the most robust CJS response (e.g., harsher sentences), “followed by whites who assault whites, blacks who assault blacks, and white (sic) who assault blacks” (Walsh, 1987, p. 155).

Black's Theory of Law

Three articles used Black's Theory of Law to inform the investigation of race in the CJS response to SA. One article relied solely on Black's Theory of Law; the second article used Black's Theory of Law in combination with Conflict Theory; and the third used Black's Theory of Law paired with Procedural Justice (see below). Spohn, White, and Tellis (2014) also used Black's Theory of Law, though for the overall investigation, not specifically for the examination of race. Interestingly, Bullock (1961) also referenced concepts and ideas from Black's Theory of Law, but did not explicitly name the theory, as the Theory of Law was not articulated by Black until 1976. Black's Theory of Law posits that the “application of law varies in its quantity,” with an arrest representing more law than no arrest or investigation at all, a criminal charge representing more law than no charges filed, and so on (Bouffard, 2000, p. 528). How much law is applied in a given situation is dependent on the social status of those involved (Bouffard, 2000; Spohn et al., 2014). “Upward” crimes are those committed by someone of lower social status against those of higher status; “downward” crimes are the opposite. Upward crimes threaten the status quo. Thus, they are considered more serious, and result in a greater quantity of law, whereas victims of downward crimes are devalued and receive less law (Bouffard, 2000; Briggs & Opsal, 2012; Roberts, 2008). In applying this theory to the CJS response to SA, we would expect a generally low level of law to be applied for all SAs, as the majority of SAs are “downward” crimes, with a male perpetrator and female victim (Bouffard, 2000). An even lower quantity of law would be expected for “ethnic minority victims” (Roberts, 2008).

Focal Concerns Theory

Three studies used Focal Concerns Theory to inform their empirical investigation of race. Two studies used Focal Concerns Theory alone, while the third used it in combination with Conflict Theory, Blameworthiness Attribution (see below), and Bounded Rationality (see below). Two additional articles used Focal Concerns Theory for the overall examination of the CJS response to SA, but did not link the theory explicitly to the investigation of race. Focal Concerns Theory states that judges’ sentencing decisions reflect a set of focal concerns; specifically, the blameworthiness or culpability of offenders, judges’ desire to protect communities from potentially dangerous individuals, and practical considerations (Holleran, Beichner, & Spohn, 2010; Spohn, Beichner, & Davis‐Frenzel, 2001). However, judges are not able to accurately determine just how dangerous an offender is. Thus, they rely on stereotypes and other attributions linked to an individual's group‐based identities, with some groups believed to be more dangerous or crime‐prone (Holleran et al., 2010; Spohn et al., 2001). This “perceptual shorthand” used by judges to make sentencing decisions has a far‐reaching impact (Steffensmeier, Ulmer, & Kramer, 1998, as cited in Spohn et al., 2001). CJS personnel who encounter offenders earlier on in the process develop a “downstream orientation,” in which police and prosecutors share judges' concerns, but also consider how a judge and jury will view the case, and the likelihood of a conviction (Frohmann, 1997, as cited in Spohn et al., 2001). Because “African American” individuals are frequently stereotyped as more dangerous and crime prone, CJS personnel are expected to be more likely to arrest, charge, and provide a harsher sentence for “African American” offenders as compared to “Whites” (Curry, 2010; see also Holleran et al., 2010; Spohn et al., 2001).

Liberation Hypothesis

Two studies used the Liberation Hypothesis to inform their investigation of race, with one relying on the Liberation Hypothesis alone and the other pairing it with the Sexual Stratification Hypothesis. According to the Liberation Hypothesis (Kalven & Zeisel, 1966, as cited in Spohn & Cederblom, 1991), when jurors are presented with weak, ambiguous, or contradictory evidence, they must find other sources of information to guide their decision‐making. Thus, they are ‘‘liberated’’ from the constraints imposed by the law and a pure fact‐finding mission, and instead turn to their own sentiments, values, and biases to inform their decisions (Spohn & Cederblom, 1991; Tellis & Spohn, 2008). Racial bias, then may influence how a juror decides a case, and even how CJS personnel respond to a case. Researchers who have used this theory explain that legally irrelevant characteristics, like the race of the suspect and victim, would be expected to influence CJS processing in “more ambiguous cases of simple rape” that involve a single known offender, no weapons and no injuries, as opposed to “aggravated rapes” that involve a stranger, multiple perpetrators, the use of a weapon, or injuries to the victim (Tellis & Spohn, 2008, p. 253; see also Spohn & Cederblom, 1991).

Social Dominance Theory

One article relied on Social Dominance Theory to inform its empirical investigation of race. Social Dominance Theory (Sidanius & Pratto, 2001, 2011) posits that all societies organize themselves into group‐based social hierarchies in which a few dominant groups possess greater social value (e.g., power, access to resources) than a many subordinate groups. Groups are defined by age, gender and other salient aspects of group‐based identity in a given context (e.g., race). Hierarchy‐enhancing and hierarchy‐attenuating forces at the individual, interpersonal, and systems level determine the extent to which groups are organized into hierarchies in a given a society at a given time. Legitimizing myths are one mechanism by which group‐based social hierarchies are created and reinforced. Legitimizing myths are shared ideologies that appeal to morality or intellect, and are used to legitimize or justify the disproportionate allocation of social value. In Social Dominance Theory, the CJS is considered “one of the most important hierarchy‐enhancing social institutions,” as individuals are treated differently within the CJS based on their group identity, with the intention of reinforcing hierarchical groups and power differentials (Sidanius, Liu, Shaw, & Pratto, 1994, p. 340, as cited by Shaw et al., 2016). In applying this theory to race and the CJS response to SA, “victims of color” would be expected to receive a less‐than‐thorough CJS response, with legitimizing myths operating to justify the disparity (Shaw et al., 2016).

Structural Contexts

One article drew upon the concept of Structural Contexts to complement the researchers’ primary applications of Conflict Theory and the Sexual Stratification Hypothesis. Kingsnorth, Lopez, Wentworth, and Cummings (1998) explain that there may not be systematic discrimination based on race and ethnicity in the CJS. Rather, Structural Contexts determine the extent of discrimination, with discrimination being limited to certain times, places, and offenses types within the CJS. One role of research is to identify the “structural and contextual conditions that are most likely to result in racial discrimination” (Hagan & Bumiller, 1983, p. 21, as cited in Kingsnorth et al., 1983).

Blameworthiness Attribution Theory

One article drew upon Blameworthiness Attribution Theory, as well as Bounded Rationality Theory (see below) to supplement the authors’ primary application of Conflict Theory and Focal Concerns Theory. In citing Baumer, Messner, and Felson (2000), Curry (2010) briefly explains that “non‐White” victims of violent crime may be seen as more responsible, or to blame, for their victimization due to stereotypes that “non‐White” individuals are more likely to be involved in violent crimes. As a result, offenders who target “non‐Whites” are seen as less blameworthy and receive more lenient punishments as compared to those who victimize “Whites.”

Bounded Rationality Theory

Curry (2010) also briefly mentioned Bounded Rationality Theory. In citing Albonetti (1991), Curry explains that there is prior support that judges seek to reduce uncertainty by relying on stereotypes that “Black” offenders have a greater criminal propensity and pose a greater risk to the community, thus they receive longer sentences. Curry (2010) points out that Bounded Rationality Theory, Blameworthiness Attribution Theory, Focal Concerns Theory, and Conflict Theory are all complementary to one another, thus their collective inclusion in Curry's investigation of race and the CJS response to SA. No other articles referenced Bounded Rationality Theory.

Consensus Theory

One article presented Consensus Theory alongside its presentation of Conflict Theory, and before introducing Lotz and Hewitt's Five Models (see below). Consensus Theory posits that society is stable and unified, law is neutral, there is consensus among most members of society as to how criminal behavior is defined, and the CJS responds equally and fairly to all groups and people, regardless of their social location, value, or power (Hunt, 1980, as cited in Maxwell et al., 2003). Maxwell et al. (2003) discuss how Consensus Theory and Conflict Theory present two competing perspectives, that research supporting both perspectives exists, and how Lotz and Hewitt's Five Models (see below) provide a menu of options to explain patterns observed.

Lotz & Hewitt's Five Models

After discussing the competing Consensus and Conflict Theories, Maxwell et al. (2003) introduce Lotz and Hewitt's Five Models (1977). Lotz and Hewitt (1977, as discussed in Maxwell et al., 2003) provide five models concerning the impact of legal and extra‐legal factors on CJS case processing. The first model aligns with Consensus Theory and contends that extra‐legal factors, like race, are not empirically related to outcomes; only legal factors, like a defendant's prior record, explain outcomes. The second model argues that extra‐legal factors may be significant, but are based on how closely tied the defendant is to their community and their family, not the defendant's race (or class or sex). The third model, more consistent with Conflict Theory, argues that class and racial biases play a role early on in the CJS response—at arrest—and go on to impact final case outcomes; harsher treatment of “minorities and those in lower classes” results from systemic racial and class bias, not from the severity of the offense or prior criminal record. The fourth model, consistent with Conflict Theory and Labeling Theory (discussed below), describes the sentencing process as a power play between CJS agents who do the labeling, and defendants who are labeled. How much power a particular defendant has depends on their race (as well as sex and class), thus influencing case processing. The fifth and final model, also consistent with Conflict Theory, contends that race, and other irrelevant factors, impact sentencing, yet in the opposite direction. At the time of sentencing, “minorities and lower class defendants” are treated more leniently compared to “Whites and upper class defendants” as the courts attempt to correct for the disproportionately high rates of arrest by police of “minorities and lower class defendants.” Maxwell et al. (2003) explain that these five models help illuminate how the same theory—Conflict Theory—may produce seemingly contradictory findings. No other articles referenced these Five Models.

Procedural Justice

One article drew upon concepts of Procedural Justice to complement their primary application of Black's Theory of Law. Briggs and Opsal (2012) argue that CJS case processing is dependent on both police action, and victim cooperation. Tyler's Procedural Justice framework (2004, as referenced in Briggs & Opsal, 2012) argues that when community members perceive CJS procedures as fair, they are more likely to view the system as legitimate, and to ultimately cooperate with police and other system personnel. Briggs and Opsal (2012) explain that existing research has documented that “blacks,” and has suggested that “Hispanics,” view these systems and actors as having decreased legitimacy, which leads to such victims being less cooperative, and to their cases not progressing in the CJS.

Labeling Theory

One article employed Labeling Theory to support their primary use of Conflict Theory in examining the CJS response to SA, though the authors’ use of these two theories was not explicitly tied to race. LaFree (1980b) briefly explains that labeling theory suggests criminal justice agents’ stereotypes for crimes influence their decision‐making in responding to crime. Thus, “the more similar the characteristics of victims, offenders, and offenses are to the typifications of rape held by processing agents, the more likely an incidence is to result in the conviction of an accused offender” (LaFree, 1980b, p. 835).

Sample Selection and Composition

Sample Sizes

Sample sizes ranged from n = 89 to n = 226,496, with an average of n = 8,398 and median of n = 413. However, the largest sample included cases of SA, as well as robbery, aggravated assault, and simple assault. Of the study samples that only included SA cases, the largest sample was n = 22,876. As can be expected, studies with the largest samples came from national databases, including the National Incident‐Based Reporting System (NIBRS) of the Federal Bureau of Investigation and National Pretrial Reporting Program (NPRP), funded by the Bureau of Justice Statistics. The largest single state and city samples came from Texas at n = 3,644, and Chicago at n = 1,530, respectively.

Sample Locales

Using the U.S. Census Bureau's four census regions (https://www.census.gov/geo/reference/gtc/gtc_census_divreg.html), fourteen studies (41.2%) used samples from the Midwest: ten studies drew samples from Midwestern cities, such as Chicago or Detroit; three studies drew samples from Midwestern counties, such as Hamilton County, Ohio; and one study from “two Midwestern communities” (Shaw & Campbell, 2013). Ten of the fourteen studies conducted with samples from the Midwest (71.4%) found at least one statistically significant race finding while the other four reported no statistically significant race findings or trends. Six studies’ samples (17.6%) came from the South, sampling from a single Southern city—Miami; multiple counties in a single Southern state—Texas; multiple counties across multiple Southern states; or single states, such as Texas and Georgia. Four studies with samples from the South found at least one statistically significant race finding, with the remaining two articles reporting a trend or pattern that suggested a race influence. Four studies’ samples (11.8%) came from the West: two studies sampled from Western cities, including Los Angeles and San Diego; one from a Western county—Sacramento County; and one from a Western state—Arizona. Half of the studies with samples from the West reported at least one statistically significant race finding, with the other half reporting trends or patterns that suggested a race influence. The remaining studies (n = 7; 20.1%) drew samples from several cities spanning several states and U.S. regions. This included a series of studies that drew samples from Kansas City (Midwest region) and Philadelphia (northeast region), with one of these studies also including cases from Miami (Southern region). This also included four studies with samples that included cities or counties from all regions of the United States, or were national in scope. Five of these studies (71.4%) reported at least one statistically significant race finding. Locale information was not provided for the remaining three studies (8.8%).

Cases Included in Sample

Studies varied in terms of which cases were, and were not, included in study samples. One study included all cases for which victims presented to a medical facility for post‐assault care, regardless of if they reported the assault to the CJS or not; this study found at least one statistically significant race finding. Fifteen studies only included SA cases that had been reported to police in their study sample, with nine of these requiring an additional related criterion, such that the case be reported and have an associated forensic exam or rape kit for the victim (n = 3); be reported and founded by police (n = 2); be reported and included in NIBRS data provided by police agencies (n= 3); or be reported, have an associated forensic exam, and have a suspect identified (n= 1). Sixty percent of the studies that limited their sample to reported cases (n = 9) reported at least one statistically significant race finding, with an additional study reporting trends. Six studies only included cases that had an arrest (n= 1), referral to the prosecutor (n= 2), or both (n= 3). One third of these studies reported at least one statistically significant race finding (n= 2), with an additional third of these studies reporting a race trend that did not reach statistical significance (n= 2). Four studies required that cases not only be reported and referred to the prosecutor, but also charged by the prosecutor. Two of these studies included an additional criterion paired with charging, such that the charges not being dismissed, or that a suspect also be arrested. Six studies’ samples only included cases that resulted in a conviction. All studies with samples limited to cases that had been charged, and studies limited to cases that resulted in a conviction, reported at least one statistically significant race finding. Finally, two studies’ samples were limited to cases in which the offender was incarcerated; one of these studies reported a statistically significant race finding, while the other reported a race trend that did not reach statistical significance.

Samples’ Racial Composition

Studies varied in how samples’ racial compositions were reported, with some reporting only victim race, only offender race, only the victim/offender dyad, or a combination of these. Eighteen studies reported on the race of the victims associated with cases in their samples. In nine of these studies, “white,” “Caucasian,” and “white nonHispanic” victims made up the largest proportion of the samples (43%–86%); “Black” or “African‐American” victims made up the largest proportion of the samples for seven studies (51%–89%); “Hispanic” victims made up the largest proportion of one study's sample (47%); and “non‐White” victims made up the largest proportion of one sample's study (66%). Eighteen studies reported on the race of the suspects associated with the cases in their samples. “Black” offenders made up the largest proportion of twelve studies’ samples (41%–94%). “White” offenders made up the largest proportion of the samples for four studies (58%–95%); and “non‐White” offenders made up the largest proportion of one study's sample (77%). The final study that reported offender race indicated that offender race was not recorded in the majority of cases (58%), as race was coded from medical examiner records (Gray‐Eurom, Seaberg, & Wears, 2002). Of the cases in which offender race was recorded, the majority (32%) were “black.” Six studies reported on the victim/offender racial dyad. In four of these studies, the largest proportion of the sample consisted of “black intraracial dyads” (40%–74%); “white suspect/white victim” dyads made up the largest proportion of the samples in the remaining two studies (35%).

Data Sources

All studies relied on CJS records or national databases. Four studies used national databases, including NIBRS and NPRP. The remaining studies relied on police records (n= 14); county records (n= 2); corrections records (n= 1); court records (n= 1); judge's referral paperwork (n= 1); prison records (n= 1); crime lab records (n= 1); some combination of police, prosecutor, and court records (n=6); or an unspecified form of CJS records (n= 3). Four studies also used medical records.

Race Coding

Victim Race, Suspect Race, and Racial Dyads

In eleven of the studies, victim and suspect race were coded and analyzed as separate variables. Eight of the studies coded for and included variables in analysis for victim/suspect racial dyad, in addition to victim race and suspect race. Six studies coded and analyzed only victim race; five studies only suspect race; and four studies only the victim/suspect dyad. All of the studies that coded and analyzed dyads alone (n= 4) or suspects alone (n= 5) reported at least one statistically significant or trending race finding. About three‐quarters of the studies that coded and analyzed victim and suspect race as separate variables (n= 8; 73%), or that coded and analyzed victim race, suspect race, and racial dyad as separate variables (n= 6; 75%) reported at least one statistically significant or trending race finding. One‐third of studies that coded and analyzed victim race alone reported at least one statistically significant race finding (n= 2; 33%).

Race Categories

Fifteen studies coded race into one of two categories: “Black”/“African American”/“Negro” versus “White.” Some of these studies explained that only two categories were provided because “there were very few complainants and suspects from other racial or ethnic categories” (Holleran et al., 2010, p. 397). Eighty percent (n= 12) of studies that used a black/white dichotomy found at least one statistically significant race finding. Ten studies coded race into “white” versus “non‐white”/“victims of color”, though two of these studies noted that “African‐Americans made up most of the non‐White category” (Roberts, 2008, p. 65) or that “all but one victim of color were Black/African‐American” (Shaw et al., 2016, p. 453). Sixty percent (n= 6) of studies that used a white/non‐white or white/person of color dichotomy found at least one statistically significant race finding, with an additional two studies finding at least one trending race finding that did not reach statistical significance. Three studies used three race categories: “white” versus “black” versus “Hispanic.” These three categories were treated as being mutually exclusive. Two of these studies found at least one statistically significant race finding, with the third reporting a trending race finding. Two studies used “black” versus “white” versus “Hispanic,” and provided a fourth “other” category. Both of these studies reported at least one statistically significant finding. One study coded race and ethnicity separately, including a code for “white” versus “black,” and a code for “Hispanic” versus “non‐Hispanic,” resulting in four categories: white non‐Hispanic, black non‐Hispanic, white Hispanic, and black Hispanic. This study found no significant race finding. One study coded race as either “Caucasian” or “American Indian,” noting that “studies including multiple minority groups are important, however, exclusive focus on American Indians allows for more specific analyses and discussions of this neglected group” (Alvarez & Bachman, 1996, p. 551). This study reported a trending race finding, though it did not reach statistical significance. One study coded race into categories of “Caucasian” versus “non‐Caucasian,” versus “Hawaiian,” and reported at least one statistically significant race finding. The remaining study did not report how it coded race.

Outcomes of Interest and Findings

Studies examined a range of outcomes related to all stages of the CJS process, from the specific investigate steps completed by law enforcement, to prosecutor charging decisions, to final case dispositions and sentencing outcomes.

Police Investigations

Investigative steps

Five studies examined the investigative steps taken by police during the course of an investigation, prior to an arrest being made or the case being referred to the prosecutor's office for the consideration of charges. These studies found that cases with “black” or “Non‐White” victims were more likely to have their rape kit submitted by police to a crime lab for analysis and to have a suspect identified (Horney & Spohn, 1996; Shaw & Campbell, 2013), but that the race of the victim, suspect, and dyad had no impact on if there was a suspect line‐up or suspect interview (Frazier & Haney, 1996; Kerstetter, 1990). Shaw et al. (2016), in looking for mechanisms that explained how race had its impact, found that race did not directly predict the overall number of investigative steps completed on a given SA case. Rather, the race of the victim predicted if police were likely to deem the victim uncooperative or otherwise blame the victim for the less‐than‐thorough police investigation, which then predicted the number of investigative steps completed. In this study, Shaw et al. found that “black” victims were more likely to be deemed uncooperative, which then meant fewer investigative steps were completed. Taken together, these findings align, as it is possible that “non‐white”/“black” victims are more likely to have their rape kits submitted and a suspect identified initially, but then are deemed uncooperative during the investigation, which halts the investigation and precludes a suspect line‐up or interview.

Unfounding

During an investigation, police may decide to “unfound” a case, meaning they determine no crime occurred and close the case. Four studies examined unfounding. Three reported no significant effect of victim, suspect, or dyad race (Bouffard, 2000; Kerstetter, 1990; Spohn et al., 2014). Tellis and Spohn (2008), however, found that among “White” victims, cases with “White” suspects were more likely to be unfounded than cases with “Hispanic” suspects. This finding may have been due to Tellis & Spohn's nuanced approach to examining race. In examining unfounding decisions, Bouffard (2000) examined racial dyads, but only had categories for “White” and “African‐American” victims and suspects; Spohn et al. (2014) provided a third category—”Hispanic”—but did not examine dyads. Tellis and Spohn (2008) examined dyads, and included a category for “Hispanic” victims and suspects. If a dyad approach were not taken, and a category for “Hispanic” not provided, they would not have been able to learn that cases involving “White” victims with “White” suspects were more likely to be unfounded than those with “Hispanic” suspects. It is not possible to evaluate where Kerstetter's study (2000) falls in this, as information on how race was categorized is not provided.

Clearance by arrest

Nine articles examined predictors of case closure by arrest or exceptional means, meaning a suspect has been identified and arrested, or that the police were prepared to make an arrest, but were unable to do so due to exceptional means (e.g., the suspect is deceased). Cases involving “White” victims, “white” suspects, and “White intraracial” dyads were more likely to result in arrest, and for suspects to be arrested on a greater number of charges (Addington & Rennison, 2008; Kingsnorth et al., 1998; LaFree, 1981; Shaw et al., 2016). Addington and Rennison (2008) and Roberts (2008) found the opposite, with (“Non‐Hispanic”) “white” victims less likely to have their cases cleared by arrest. However, both of these studies were conducted with large samples from NIBRS data (n = 22,876; n = 11,215, respectively). Addington and Rennison note that “because of the large sample size, almost all of the predictors and controls are statistically significant. As such, it is important to consider the “substantive” (or clinical) significance and examine the effect size” (p. 216). Both studies reported relatively small odds ratios for the race effect (OR = 0.88; OR = 1.113, respectively). An additional four studies reported no statistically significant effect of victim, suspect, or dyad race on arrest (Bouffard, 2000; Briggs & Opsal, 2012; Kerstetter, 1990; LaFree, 1980a). Similar to patterns observed in unfounding (see discussion above), studies that reported no significant race findings on arrest had less nuance in their measurement and analysis. For example, LaFree's studies in 1980a and 1981 relied on the same data. The earlier study reported no significant race finding on arrest. However, when the data were revisited in 1981, LaFree included an examination of the influence of race on arrest before and after an SA unit was implemented in the focal jurisdiction. Accounting for this additional influencing feature revealed a significant relationship between race and arrest.

Referral to the prosecutor

In order for a case to proceed to prosecution, it must be referred to the prosecutor by police for the consideration of charges. Only one study examined the influence of race on the likelihood of a referral to the prosecutor. Shaw et al. (2016) found that race indirectly predicted referral to the prosecutor's office. “Black” victims were more likely to be deemed uncooperative by police, or otherwise disruptive to the investigative process, which then decreased the likelihood that the case would be referred to the prosecutor's office.

Summary

Looking at the police investigation in totality, prior research finds that cases involving folks of color were more likely to have a suspect identified initially and the rape kit submitted to the crime lab for analysis. However, victims of color—more specifically, Black victims—were more likely to be deemed uncooperative or otherwise disruptive to the investigation by police. This results in fewer investigative steps being completed throughout the investigation, including a suspect lineup, suspect interview, arrest, and referral to the prosecutor. That victims of color are met with a less thorough CJS response lends support to several of the theories identified herein, such as Black's Theory of Law, Social Dominance Theory, and Blameworthiness Attribution Theory.

Prosecutor Action

Charging decisions

Once a case is received by the prosecutor's office, the prosecutor determines if they will file and pursue charges against the suspect. Eighteen studies examined the decision to file charges initially, the decision to pursue charges (i.e., not drop or dismiss the charges), or the severity of the charges filed. Cases involving “White” victims were more likely to have charges filed as compared to “non‐White” and “Black” victims (Beichner & Spohn, 2012; Spohn & Holleran, 2001), and “Caucasian” victims were overrepresented among those with their assailants indicted, as compared to “Hawaiian” victims (Chandler & Torney, 1981). “Black” suspects, in general, as well as “black” suspects with “white” victims were more likely to be charged with more serious crimes and for the charges to be filed as felonies (Bradmiller & Walters, 1985; LaFree, 1980a, 1981). “Caucasian” suspects were also underrepresented among those indicted, as compared to “Hawaiian” suspects (Chandler & Torney, 1981). However, a couple of studies have found that “white” victims, generally, and “white” victims with “black” suspects, specifically, are more likely to have charges dismissed as compared to “black” victims, and other racial dyads, respectively (Spohn & Horney, 1993; Spohn & Spears, 1996). These findings are not at all disparate and suggest that prosecutors initially issue charges on more cases involving white victims and suspects of color, particularly black and Hawaiian suspects. However, after initially charging these cases, it may become clear that there is not a strong enough case for it to proceed to prosecution, and the charges are dropped. An additional ten studies reported no statistically significant findings of victim or suspect race, or racial dyad on charging decisions, or later dismissal of charges (Campbell et al., 2009; Frazier & Haney, 1996; Gray‐Eurom et al., 2002; Holleran et al., 2010; Kerstetter, 1990; Kingsnorth et al., 1998; Maxwell et al., 2003; Spears & Spohn, 1997; Spohn et al., 2001; Tellis & Spohn, 2008). Similar to patterns observed in comparing measurement and analytic decisions between studies with significant and non‐significant findings above, studies with non‐significant findings frequently (though not always) had less nuance. For example, in 2010, Holleran, Beichner, & Spohn reported no effect of race on charging decisions. In 2012, Beichner & Spohn used the same data and found that “White” victims were more likely to have charges filed than “non‐White” victims. The difference was that in their new analysis, they distinguished between aggravated and simple rape. In cases of aggravated rape, meaning the assault involved a stranger perpetrator, a gun or knife, or collateral injuries to the victim, the race effect described above emerged. Race continued to have no effect on charging decisions in cases of simple rape (i.e., none of the factors required for aggravated rape). Beichner and Spohn (2012) explained that this “suggests prosecutors believe that the seriousness of the crime is enhanced when the victim is White…[and] that the race of the victim may itself be an aggravating factor” (p. 20). In other studies with null findings, study samples did not have a great deal of racial variation, thus “there may not have been sufficient variability to detect an effect, if there was one to be found” (Campbell et al., 2009, p. 722).

Victims declining prosecution

If a prosecutor chooses to charge a suspect, the case moves forward as the charging entity (e.g., county or state) versus the defendant. It is not the individual victim who chooses if charges will be pursued or not. Nonetheless, victims may inform the prosecutor that they do not wish to pursue prosecution of the suspect. Two studies examined the influence of race on victims’ decisions to pursue or participate in prosecution. Both Tellis and Spohn (2008) and Kerstetter (1990) found no effect of victim race, suspect race, or the racial dyad on victims declining prosecution.

Pretrial release

After a suspect has been charged, the suspect may be detained as they await trial, or the suspect may be released. Maxwell et al. (2003) examined the influence of race on pretrial release and found no effect of suspect race.

Many studies have examined prosecutor action and decision‐making in the context of SA cases. About half of the studies that have examined the influence of race in charging decisions found no significant effect. Studies that do find a significant effect of race find that cases involving white victims and suspects of color—more specifically, Black and Hawaiian suspects—are more likely to be charged, and to be charged with more serious crimes. Following initial charging, though, these cases are more likely to be dropped. The very few studies that have examined the influence of race on victims’ decisions to participate in prosecution and pretrial release have found no effect. These findings lend support to several theories highlighted, including Conflict Theory, the Sexual Stratification Hypothesis, and Black's Theory of Law.

Final Case Outcomes

Plea bargains

Before a case goes to trial, the prosecutor and the defendant may agree to a plea bargain. Six studies examined the influence of race on the likelihood of a plea bargain. Cases involving “Hawaiian” and “black” defendants were more likely to go to trial, and less likely to result in a plea bargain, as compared to “Caucasian” or “white” defendants (Chandler & Torney, 1981; LaFree, 1980b). Cases involving “black” victims were also less likely to result in a plea bargain compared to cases with “white” victims (Spohn & Horney, 1993). The other three studies found no effect of victim race or the racial dyad on plea bargaining (Campbell et al., 2009; Kingsnorth et al., 1998; LaFree, 1980a). The non‐significant findings from Campbell et al. (2009) and LaFree (1980a) may be in part due to a lack of variation in the sample, and attention given to other contextual factors (i.e., the implementation of an SA unit), respectively. The study from Kingsnorth et al. (1998), however, introduces another critically important consideration. The Kingsnorth study only included in its sample cases that were referred to the prosecutor, not all cases that were reported to police. These cases may have already been subjected to different treatment based on the race of those involved, leaving less variation among the cases included in the sample.

Convictions

Six studies examined the influence of race on convictions. In some studies, convictions refer only to guilty verdicts at trial, while other studies also include plea bargains. All of the studies that examined suspect race found an influence of suspect race on convictions, with “Black,” “African‐American,” and “Hispanic” defendants less likely to be convicted by either a guilty verdict at trial, or a plea bargain, as compared to “White” defendants (Maxwell et al., 2003; Spohn & Horney, 1993; Spohn & Spears, 1996). LaFree (1980b) also found that cases with “black” victims were less likely to result in conviction through a guilty verdict or plea bargain as compared to “white” victims (1980b). Two studies reported no effect of victim race or dyad on convictions (Campbell et al., 2009; LaFree, 1980a), and have been previously discussed.

Sentencing

Twelve articles examined the influence of race on sentencing following conviction. This included examinations of sentence length; sentence severity (i.e., based on a scoring system); if the sentence resulted in incarceration or not (e.g., probation, suspected sentence, fine) and the type of facility in which the offender was incarcerated (e.g., state penitentiary, prison, jail); and if the defendant received a death sentence or not. “White” offenders were more likely to receive harsher sentences, longer sentences, and to be sentenced to prison than “Asian,” “African‐American,” “Negro,” and “Hispanic,” offenders (Bullock, 1961; Curry, 2010; Maxwell et al., 2003; Walsh, 1987). The same pattern was observed between “Caucasian” and “American Indian” offenders, though this did not reach statistical significance (Alvarez & Bachman, 1996). However, several studies found exceptions to this general pattern. In examining dyads, LaFree (1980a), Spohn and Spears (1996), and Walsh (1987) found that “black” suspects who had “white” victims had longer sentences, were more likely to be incarcerated, and more likely to be sent to a state penitentiary as compared to other racial dyads. In examining stranger and acquaintance assaults separately, Spohn and Cederblom (1991) found “black” suspects were more likely to be incarcerated in cases of stranger SA, with no effect of race in acquaintance assaults. Additionally, in examining the use of the death penalty for rape convictions in eleven southern and border states from 1945 to 1965, Wolfgang and Riedel (1973, 1975) found that “black” defendants were sentenced to death at a rate seven times that of “white” defendants, and that “black” defendants with “white” victims were sentenced to death at a rate eighteen times that of any other racial dyad. Two additional studies reported no influence of the victim race, suspect race, or racial dyad on the sentence length, if the suspect was incarcerated or not, and if they were incarcerated in jail or prison (Kingsnorth et al., 1998; Spohn & Horney, 1993). These null findings may be due in part to limited variance in racial composition of the sample, as Spohn and Horney's (1993) sample consisted of nearly 90% “black” suspects; indeed the authors treated suspect race as a control variable. This may also be due in part to differences across study samples. Kingsnorth et al. turn to the theory informing their exploration—Structural Contexts—to explain that there is not systematic racial discrimination based on race and ethnicity in the CJS. Rather, such discrimination is bound to certain times, places, and offense types within the CJS, and that this discrimination simply does not appear in their sample of rape cases that were referred to the prosecutor in Sacramento County in the early 1990s.

In reviewing final case outcomes, relatively consistent patterns emerge. White suspects were more likely to take a plea than to go to trial. When white suspects do go to trial, the trial was more likely to end with a guilty verdict. Convicted white defendants were then more likely to receive harsher and longer sentences as compared to defendants of color. This pattern largely holds, unless a black defendant was convicted of assaulting a white victim, a stranger, or in contexts in which rape is a capital offense. In such situations, black defendants received harsher sentences, and far more frequent use of the death penalty. That white suspects convicted at trial received harsher and longer sentences as compared to defendants of color aligns well with Lotz & Hewitt's Five Models, particularly the fifth model that explains how “minorities and lower class defendants” are treated more leniently at sentencing as the courts attempt to correct for the overrepresentation and targeting of these individuals earlier on in the process. That this pattern differs in certain contexts supports the concept of Structural Contexts and the Sexual Stratification Hypothesis.

Discussion and Implications for Future Research

This systematic review examined prior research on race and the CJS response to SA in an effort to highlight and explore what appeared to be discrepant findings. Through this review, it becomes evident that the findings are not discrepant, but instead come together to tell a complex story of how SA cases move through the CJS in relation to the race of those involved, and how researchers shape and influence the story (intentionally or not) through the series of methodological decisions they make in their investigations. Several key considerations for future research on this subject matter emerge.

First, researchers should be more deliberate in choosing if and how to include race in their empirical investigations. Too often, race is included as one of a battery of variables, added into a study without much consideration of why and how race is being examined in relation to the phenomenon under investigation. In this review, we learned that fewer studies that included race as one of many control variables reported significant race findings, as compared to those studies that included race as a specific variable of interest. The level of detail and nuance in categorizing race also matters. Researchers frequently collapse data and treat race as a binary variable (e.g., “white” vs. “nonwhite”). They provide statistical justification for this decision; for example, that so few folks of color were included in the sample that it warranted treating them as a single category for purposes of analysis. This is problematic in that it perpetuates “whiteness” as the norm to which all others are and should be judged (see Hamby, 2015). Beyond that, though, it assumes that all “nonwhite” individuals have similar enough experiences that they can be combined with one another, and that such a choice will not hinder our ability to detect significant differences between (sub)groups. In this particular context, this would assume that all individuals of all racial and ethnic backgrounds, other than white individuals, have the same or similar relationships, interactions, and histories with the CJS. As has been argued elsewhere (e.g., see Briggs & Opsal, 2012; Hamby, 2015), and is demonstrated by our review, this assumption is erroneous. In our review, we found that far fewer studies that used “white” and “nonwhite” categorizations reported significant findings, as compared to studies that maintained the racial or ethnic variation represented in their sample when conducting analysis (e.g., “Black,” “Hispanic,” “American Indian,” “Hawaiian”). In maintaining distinct categories, Maxwell et al. (2003) was able to find that “African‐American” and “Hispanic” suspects were less likely to be found guilty than “White” suspects whereas “Asian” suspects were more likely to be found guilty (though this didn't reach statistical significance). This would not have been revealed were these groups collapsed into a single “nonwhite” category. Whose race we examine matters, too. Our review revealed that studies that included suspect race reported significant findings much more frequently than studies that only included victim race. Taken together, this suggests that the myriad decisions made by researchers regarding race, from if they include it at all, to how it is coded, really matter. Researchers should not add race to their studies for the sake of it, but be thoughtful in deciding if and how to examine race, and intentional in explaining their decision‐making. Such intentionality is likely to yield more nuanced findings, and a greater understanding of how race interfaces with the phenomenon of interest.

Second, researchers should seek samples that are appropriate for answering their research questions, and revisit the many different elements of their sample's composition when discussing findings. If we are interested in understanding how race influences the CJS response to SA, as well as other phenomena, we must select samples with adequate variation in the independent variable (i.e., race). We must also seek samples that provide enough cases within each category of our independent variable to allow for intentional analysis that does not revert to and rely on statistical justification for collapsing categories into one another. In our review, we found several studies that focused on understanding patterns in the CJS response to SA for very specific racial or ethnic groups. In some cases, the researchers were able to attain an adequate sample size to reach statistical significance (e.g., Chandler & Torney's, 1981 examination of “Hawaiian” suspects and victims). Other times, the proportion of the sample from the specific racial or ethnic group was too small; trends and patterns in the anticipated direction were reported, but did not reach statistical significance (e.g., Maxwell et al.'s, 2003 inclusion of “Asian” suspects, with only 3% of the sample reported as “Asian;” Alvarez & Bachman's, 1996 examination of “American Indian” inmates with only 5% of the sample reported as “American Indian”). Thus, researchers should make concerted effort to develop samples that are responsive to their research questions, and that include adequate representation from groups traditionally underrepresented in social science research.

Researchers should also take care when reporting their findings, and revisit how sample selection may impact them. This includes attention to the geographic region from which the sample was drawn. In our review, we found that all samples from the South and West reported at least one statistically significant or trending finding; the same was not true for samples from the Midwest. Beyond geographic region, scholars who focus on identifying structural contexts that may contribute to racial and ethnic discrimination also recommend examining the proportion of the population that is “Black,” as well as unemployment rates (Chiricos & Crawford, 1995; Hagan & Bumiller, 1983; Kingsnorth et al., 1998). A particularly important consideration in discussing findings vis‐à‐vis the sample is attending to the shifting denominator that we observe across studies. In our review, we found that some studies examined case progress out of all reported SA cases, while others limited their sample to only cases that are charged, only cases that resulted in a conviction, or only cases in which the defendant was incarcerated. Because the denominator is shifting across these different studies, it is difficult to draw clear comparisons. For example, in examining racial disparities in sentencing severity among only those who are convicted, we must ask if there are racial disparities among convictions, and thus who enters into our sample in the first place. In discussing rates of attrition in the CJS overall, Frazier and Haney (1996) explain that “less attrition in the prosecutorial phase does not necessarily reflect less bias among prosecutors; rather, it may mean that only the “strongest” cases are referred” (p. 622). This same logic can be applied to differential attrition, as a function of race. Researchers should not overlook the problem of the shifting denominator and what it means for interpreting the findings of a single study, as well as comparisons across a body of literature.

Third, race‐based discrimination and oppression, like all forms of discrimination and oppression, are systemic and cumulative. It does not exist in only a single instance or decision‐point. Rather, it is in the accumulation of instances and decision‐points that we may witness its full scope. In our review, some studies examined only one or two processing outcomes, and reported no significant race effect. As LaFree (1980a) explains, single “studies examining only one or two processing outcomes…may or may not support a view of discriminatory processing, depending on independent variables included, methods used, and specific outcomes examined” (p. 852). It is in placing each outcome, instance, and decision‐point in its broader context that we can see the “cumulative effect of race” as “consistent and substantial” (p. 852). We cannot examine every outcome in every study, but researchers can and should take care to contextualize each study in the broader literature, as well as in the CJS process.

Finally, it is time that researchers move away from examining race as a cause of system outcomes. As Hamby (2015) explains it, “race does not have a causal effect; rather, it is a marker for some unknown set of processes that have the actual cause impact on violence, risk, resilience, and other constructs” (p. 3). Documenting that individuals of certain races are more likely to have their SA cases advanced in the CJS as compared to individuals of other races does nothing to assist in identifying leverage points for changing how the system responds. Thus, it is imperative that researchers move beyond only documenting relationships between race and CJS outcomes, to examining the mechanisms by which race is used by CJS agents to catalyze or justify a defined response. The intentional use of theory in examining race and the CJS response to SA can help researchers focus on identifying mechanisms and making known the “unknown set of processes.” For example, Shaw et al.'s (2016) use of social dominance theory led the research team to identify legitimizing myths as the mechanism by which race had its impact on case outcomes, and a specific target for change. To move from understanding to action, it is critical that researchers take this next step in their work when examining race, as well as other aspects of individuals’ intersectional group‐based identities.

Researchers, regardless of their substantive focus, are charged with producing knowledge through systematic investigation. Community psychologists who engage in research also commit to using this knowledge toward action. This raises the stakes in that if the processes used and decisions made in the knowledge‐production process lack intention and deliberate forethought, the actions proposed and implemented thereafter will be ill‐informed, and may reinforce a problematic status quo. The series of decisions we make throughout the research process impact the findings we generate, shape the collective stories we tell, and define the action we take. We yield a great deal of power, and with great power comes great responsibility.

Conflict of Interest

We have no conflicts of interest to disclose. The work described in this manuscript complies with ethical standards.