Rape and Attrition in the Legal Process: A Comparative Analysis of Five Countries

Despite legal reforms, there has been little improvement in police, prosecutor, and court handling of rape and sexual assault. In the past 15 years in Australia, Canada, England and Wales, Scotland, and the United States, victimization surveys show that 14 percent of sexual violence victims report the offense to the police. Of these, 30 percent proceed to prosecution, 20 percent are adjudicated in court, 12.5 percent are convicted of any sexual offense, and 6.5 percent are convicted of the original offense charged. In the past 35 years, average conviction rates have declined from 18 percent to 12.5 percent, although they have not fallen in all countries. Significant country differences are evident in how cases are handled and where in the legal process attrition is most likely. There is some good news: a victim’s “good” character and credibility and stranger relations are less important than they once were in police or court outcomes. However, evidence of nonconsent (witness evidence, physical injuries to the victim, suspect’s use of a weapon) continues to be important. In 1978, the first studies written in English of police and court responses to rape were published. As of September 2007, over 90 attrition studies had reported findings for Australia, Canada, England and Kathleen Daly is professor of criminology and criminal justice at Griffith University (Brisbane). Brigitte Bouhours is research officer in the Centre of Excellence in Policing and Security, Australian National University (Canberra), formerly, senior research assistant, Griffith University, working with Daly for 9 years to October 2009. We are grateful to Michael Tonry and the reviewers for their extended and helpful comments on an earlier draft. 566 Kathleen Daly and Brigitte Bouhours Wales, Scotland, and the United States from 1970 to 2005. Some analyze the same data set, and the findings from others can be combined; thus, the unique set of cases reduces to 75. In this essay, we analyze this body of research from five countries to identify patterns in police, prosecutor, and court handling of rape and sexual assault cases. It is contextualized by victimization surveys, police statistics, and court data from the countries examined. We also explore the factors associated with conviction and attrition such as the victim’s character and credibility, prompt reporting of the offense, victim-offender relations, and evidence of nonconsent. The project began with a more simple aim: to summarize Englishlanguage studies of rape case handling by the police and courts. However, we soon discovered that there were widely varying estimates of attrition and conviction in the literature. Authors cited different studies or selected findings or focused on some jurisdictions or countries. Conviction rates were given, but often it was not clear if they pertained to any offense or to the original offense charged, and researchers calculated estimates and defined outcomes in different ways. Research on the prevalence of rape and its legal handling is highly politicized and contested. Victim advocates are criticized for providing “widely inflated estimates” of sexual victimization (Gilbert 1997, p. 101), and skeptics can be criticized for not understanding sexual violence in more fluid terms, as a continuum (Kelly 1988). Legal definitions of rape have changed in the last three decades, but the more consequential change is social and political. This was encapsulated in early feminist challenges to the “real rape” construct, that rape is carried out by a stranger, using a weapon, and with serious victim injury (Estrich 1987), when the more typical rape is by a known person, without a weapon, and without physical evidence of nonconsent. As legal 1 Since September 2007, New Zealand has completed a rape attrition study (Triggs et al. 2009). Other studies that are planned or under way include the Irish Rape Attrition Project (2010), a rape attrition study of 11 U.K. and European countries (Lovett and Kelly 2009), the Understanding Attrition in Rape Cases Project in Sussex (McMillan and Thomas 2008), and a rape attrition study planned for South Africa (Gender, Health and Justice Research Unit 2008). Johnson, Ollus, and Nevala (2008) report estimates from the International Violence Against Women Survey that we cite when relevant. 2 We initially combined England and Wales and Scotland but then decided to treat them separately because their rape law and criminal procedures differ, as do their attrition rates over time. In Scotland, unlike England and Wales, the definition of rape is still gender specific, although this is changing. The Scottish criminal justice system includes an additional stage, that of the procurator fiscal, who takes on some duties of the police and the prosecution (personal communication with Michele Burman, January 24, 2008). Rape and Attrition in the Legal Process 567 definitions and sociopolitical understandings of rape widened and as advocacy for victims grew from the 1970s onward, research on rape has been caught up in a politics of rape. Debates initially focused on rape prevalence (e.g., was it an epidemic or not?). In the past decade, debates have matured, but governments have been called on to do more. Low or declining conviction rates, faulty or questionable police investigations, and poor treatment of rape victims have put pressure on governments to review rape laws and legal procedure, not for the first time, but yet again. As our research progressed, the need to create an authoritative and comprehensive record of what is known about rape and its handling in the legal process became clear. Reviews by Bryden and Lengwick (1997), Kelly (2001), Lievore (2004), Koss (2006), and Du Mont and White (2007) consider the prevalence and contexts of sexual victimization, victims’ reporting patterns, and justice system responses. However, ours is the first study to assemble and harmonize the relevant body of research to estimate rates of conviction and of case attrition at different stages of the legal process and to identify factors associated with case attrition or retention. To provide a comparative context and understanding of the patterns that may emerge, we also review rape law reform and compare police statistics and court outcomes in the five countries studied. A note on definitions. Rape is “unwanted oral, anal, or vaginal penetration against consent through force, threat of force, or when incapacitated” (Koss 2006, p. 208). It includes sexual intercourse with children (typically at law, under 16). “Rape” differs from “sexual assault” and “all sexual offenses.” Sexual assault refers to a wider set of offenses, including penetrative (i.e., rape) and nonpenetrative (e.g., indecent assault) offenses that touch the body sexually. “All sexual offenses” include rape, sexual assault, and “no touch” offenses (e.g., indecent behavior or sexual exposure). Although most attrition research is concerned with the sociolegal response to rape (i.e., forced penetrative sex), victimization surveys, attrition studies, and official police and court data may include a broader set of offenses and victims of varied age groups. For simplicity of expression, we use victim and offender throughout the essay, without the “alleged” preface, and we use victim rather than survivor or victim-survivor. 3 Because the composition of offenses varies, depending on the study or official data source, we use the more generic terms “sexual victimization” or “sexual offenses.” 568 Kathleen Daly and Brigitte Bouhours Here are our major findings. Of sexual offenses reported to the police during the past 35 years, the overall rate of conviction to any sexual offense is 15 percent. When an early period (1970–89) is compared with a later period (1990–2005), this rate has declined significantly from 18 to 12.5 percent. Significant decreases have occurred in England and Wales, Canada, and to a lesser degree in Australia, but not in the United States or Scotland. Across the three decades, the overall rate of conviction to any sexual offense is a bit higher in samples of child or youth victims than those of mixed age or adults only. Significant decreases in conviction rates over time are evident in samples of mixed age and child or youth victims and across all types of sexual offenses. With regard to where attrition occurs in the legal process, the following are averages across countries for the more recent period. Of sexual offenses reported to the police, 30 percent proceed past the police to prosecution, 20 percent are adjudicated in court, 12.5 percent are convicted of any sexual offense, and 6.5 percent are convicted of the original offense charged. Few cases go to trial (8 percent) and are convicted at trial (4.5 percent). Attrition is greatest at the start of the sociolegal process: from victimization surveys, an average 14 percent of victims report the offense to the police. Once reported, a minority of cases proceed past the police to the prosecutor’s office, and this occurs for a variety of reasons. Suspects cannot be identified or located, victims withdraw complaints, and the police believe that there is insufficient evidence to charge a suspect or the victim’s story lacks credibility. Attrition averages do not tell the whole story and can be misleading because there are significant differences by country and time period in the police and court handling of cases. One explanation for decreasing conviction rates is that as more sexual offenses are reported to the police, they contain a higher share of known relations and rape contexts that do not accord with the real rape construct. At the same time, police, prosecutorial, and court decisions continue to operate with the real rape construct in mind. We find that this explanation applies best to England and Wales but is less evident in other countries. Of the four countries with sufficient re4 The terms “overall rate of conviction” and “overall conviction rate” refer to the proportion of cases reported to the police that are convicted of any sexual offense. 5 When we say “significantly” here and elsewhere in discussing the results, the reference is to statistically significant differences. Rape and Attrition in the Legal Process 569 search, the United States is an anomaly with no change in conviction rates over time. There is no one pattern of conviction and attrition in the countries studied. We recognize that legal reforms of the past several decades may have helped some victims, but all commentators agree that the gains have been modest. We call for a shift in the priorities of legal reform—away from the trial and toward mechanisms of encouraging admissions to offending, which includes pursuing alternative pathways of participation and support for victims, offenders, and others affected by sexual offenses. This essay has six sections. Section I reviews research and survey data on the prevalence of rape, on reporting to the police, and on case processing in the prosecutorial and judicial systems. It also discusses charges in rape law in the five countries. Section II sets out questions and related hypotheses examined in later sections. Section III describes the scope of the analysis, our strategy for locating relevant studies, problems of data comparability, and how estimates of conviction and attrition were calculated (app. B discusses these matters in more detail). Section IV presents and discusses the main findings. In Section V we examine our findings in relation to the questions and hypotheses set out in Section II. The final section puts forward ideas for more innovative and effective responses to rape and sexual assault, including cases that are not reported to the police and those that are reported, but subsequently withdrawn in the criminal process. I. Attrition Research in Comparative Context To understand patterns of rape case attrition over time and across different jurisdictions requires attention to a wider social and legal context. Since the 1960s and 1970s, with the rise of second-wave feminism, the legal definitions and social meanings of rape began to change. Major challenges were raised by feminist scholars about the veracity of official data and the harsh treatment of victims in the legal process. A. Was It Rape? Reporting Rape to the Police and Legal Responses Estrich’s (1987) review of the literature up to the mid-1980s analyzed the interrelationships among victims’ experiences of rape, sample survey estimates of victimization, victims’ reports to the police, and how cases were handled by the police and courts. She coined the term 570 Kathleen Daly and Brigitte Bouhours “real rape,” drawing from Kalven and Zeisel’s (1966, p. 252) term “aggravated rape.” Real rape has one or more of these elements: stranger relations, multiple assailants, weapon use, and evidence of serious physical injury. Or as Estrich says, the real rape image is of an “armed man jumping from the bushes” (p. 8). By contrast “not real” rape (also termed “simple” rape) has none of these aggravating elements: the offender is a lone man, whom the woman knows (a neighbor, an acquaintance, a date), using no weapon, and leaving no physical injuries or bruises on a victim. Estrich argued that in simple rape contexts, questions about a woman’s character, credibility, and believability were especially likely to be raised. Although real rape was reported to the police by victims and treated seriously by the criminal justice system, simple rape was “far more common, vastly underreported, and dramatically ignored” by the police and courts (p. 10). From early victimization surveys in the United States (the National Crime Surveys [NCS] conducted during 1973–82), the Bureau of Justice Statistics (1985) estimated that 52 percent of rapes (both attempted and completed) were reported to the police. It soon became apparent that the NCS had grossly underestimated the prevalence of rape and overestimated the likelihood of victim reports to the police because of the way in which the survey questions were asked. Other studies in the United States appeared at the time (e.g., Russell 1975; Williams 1984; see Estrich 1987, pp. 11–14) showing that women who were sexually assaulted by persons they knew (acquaintances, friends, neighbors, or relatives) were far less likely to report the offense to the police and to victim survey interviewers compared to women who were assaulted by strangers. The Sexual Experiences Survey, developed by Koss and colleagues in the mid-1980s (Koss and Gidycz 1985; Koss, Gidycz, and Wisniewski 1987), broadened the behaviors associated with rape. Using this instrument, the authors found that the rate of rape was at least 10 times greater than that estimated from the NCS (see Johnson, Ollus, and Nevala 2008, pp. 10–14). In 1992, the NCS sexual and domestic 6 Although Estrich (1987) draws from Kalven and Zeisel’s (1966) work in defining real rape, her analysis focuses mainly on different responses to stranger and known relationship cases. 7 Although Koss and colleagues were criticized by Gilbert (1997) for overestimating prevalence, a more recent study of victimization on U.S. college campuses vindicates Koss’s point that estimates change dramatically, by 10 times, when questions are asked differently (see Fisher, Cullen, and Turner 2000, pp. 11–14). Rape and Attrition in the Legal Process 571 violence victimization questions were redesigned, and special modules were introduced in the British Crime Survey in 1994 to elicit respondents’ experiences with rape and sexual assault. Both led to significant increases in the estimated rates of sexual victimization. In 1993, Statistics Canada fielded the first national survey of women’s experiences of sexual and physical victimization, which used more sensitive approaches to elicit information and conduct the interviews (see a review of methods used in Johnson and Sacco [1995] and Johnson [1996]); their approach was adopted in other countries during the 1990s. The International Crime Victim Survey (ICVS) began to gather victimization data in 1989 from 60 countries (now numbering 70 countries), but it was not designed to elicit an understanding of female experiences of sexual or physical victimization. To address this problem, the International Violence Against Women Survey (IVAWS) project was established as an international comparative survey of women’s experience of sexual and physical violence and of criminal justice responses. After pilot studies in 2001 and 2002, the first survey was carried out during 2003–5 in 11 countries of the developed and developing world. Publications are now emerging (see, e.g., Mouzos and Makkai [2004] for Australia and Johnson, Ollus, and Nevala [2008] for the entire sample) that present a complex picture of victimization and reporting patterns, with an analytical emphasis on partner and nonpartner violence. Several observations can be made. Reliable estimates of female sexual victimization and associated estimates of reports to the police are recent. Such estimates depend on how the questions are asked and the degree to which the interview context is supportive of participants. As methodologies improve to elicit the frequency, types, and contexts of rape and sexual assault and the estimated incidence of sexual victimization increases, the rate at which women tell survey researchers that they reported the offense to the police decreases. Thus, when victimization surveys capture a larger share of “nontraditional” rapes (simple rapes and those involving known relations), the rate at which victims say to interviewers that they reported the offense to the police goes down. Likewise, the composition of reports received by the police may have an increasingly larger share of nontraditional rapes over time because there are more supports and services for victims, coupled with a 8 Where relevant, we draw on the IVAWS findings. However, just one of the five countries in our study (Australia) participated in the IVAWS. 572 Kathleen Daly and Brigitte Bouhours changed consciousness about rape, to bring these incidents to police attention. Table 1 itemizes the major surveys of sexual victimization conducted since 1992 in four countries that are the subject of this study. It shows that across all the surveys, the rates of report to the police range from 6 percent (the first victim-friendly survey in Canada) to a high of 32 percent (a standard victimization survey in the United States, albeit with redesigned questions). When the latter unusually high estimate is excluded, the average rate of victim report is 14 percent. By country, rates of report are 15–19 percent (United States), 14–18 percent (England and Wales), 12–20 percent (Australia), 6–19 percent (Canada), and 12 percent (New Zealand). For age, surveyed victims are typically 16 or older, and the youngest age group (16–24 years) has the highest rate of sexual victimization. Where data are available, however, the highest rate of sexual victimization is found for those 10–14 years (Snyder 2000; AIC 2008). In research on child and youth victims (those 12–17 years), the rate of reporting sexual victimization to the police ranges from a high of 30 percent for “violent sexual assault” from the National Crime Victimization Survey (NCVS; Finkelhor and Ormrod 1999) to 13 percent for “sexual assault” (Kilpatrick and Saunders 1997) and to a low of 3 percent for “sexual victimization” (Finkelhor and Dziuba-Leatherman 1994; cited in Finkelhor, Wolak, and Berliner 2001, p. 18). Although the number of studies is small (and all are from the United States), they suggest a somewhat lower likelihood that child and youth victims report sexual victimization to the police than adult victims. Why, on average, do 86 percent of victims not report rape and sexual assault to the police? The reasons given by victims, often in combination, are not viewing the assault as rape or not thinking that others will view it as rape; fearing that others will disbelieve or blame her, including family members or friends; fearing or distrusting the police and court processes; fearing threats or further attacks by the offender or his family and friends; and having divided loyalties when reporting a family member or ex-partner (Kelly 2001, pp. 9–10; Lievore 2003). Because many victims are unsure about what to do or may blame them9 By comparison, estimates of reports to the police for nonpartner sexual violence, drawing from the IVAWS project, are somewhat lower: ranging from 4 percent (Poland), to 6–8 percent (Costa Rica, Denmark, Australia, and the Czech Republic), to 13 percent (Mozambique; Johnson, Ollus, and Nevala 2008, fig. 6.12, p. 147; table III.12, p. 210). Rape and Attrition in the Legal Process 573 selves, they may delay their report to the police. The police, in turn, may interpret delay as a sign that the assault was not serious or that a victim is not being fully truthful. Young women’s potential for victimization is high in social occasions in which alcohol or illegal drugs are used. These contexts heighten a risk of what is termed “acquaintance” rape, although the assaults may be viewed by victims, general members of society, and legal authorities as nontraditional forms, not as real rape. B. Real Rape and Victim-Offender Relations An important question for attrition research is whether the relative composition of aggravated and simple rapes, or of stranger rapes, reported to the police has changed over time. In England and Wales, attention has been drawn to the growing gap between women’s reports to the police and a diminished rate of court convictions (Kelly, Lovett, and Regan 2005, p. 25). One explanation is that an increasing share of nontraditional rapes, which are more difficult to prove in court, are being reported to the police. Evidence from varied jurisdictions over time is lacking on victimoffender relations and reporting patterns, but two studies are relevant. Baumer’s (2004) analysis of the NCS for 1973–91 finds that of the 51 percent of rapes that victims said they reported to the police, half involved known persons. Data from the redesigned NCVS for 1992– 2002 show that of the 30 percent of rapes that victims said they reported to the police, 84 percent involved known persons. From multivariate analyses, Baumer finds that in the 1970s and 1980s, women were less likely to report being raped by a known person than by a stranger, but by the early 1990s, victim-offender relationships had no effect on the likelihood that victims reported rape offenses to the police. He suggests that increases in victims’ reports of nonstranger rapes can be attributed to changing perceptions of and broadened definitions of rape that include acquaintances and intimates. The second study, by Harris and Grace (1999), compares British Home Office data for 1985 and 1996 on the stranger share of cases reported to the police. They find a drop in the stranger share from 30 percent to 12 percent (see also Grace, Lloyd, and Smith 1992). To pursue this question further, we identified a subsample of cases from our attrition sample that examined victim-offender relations as an attrition factor and then reread each to determine the stranger re-