All crime data have flaws, but sexual assault data are notoriously inaccurate. Why are these data so problematic? And what are the consequences for how we address sexual violence in the United States?

Data on rape and sexual assault suffer from inconsistent estimates and underreporting, leading to misunderstandings about the extent of the problem and adequate policy solutions. Let’s look at two major sources of information on the topic: survey-based studies that estimate prevalence of sexual assaults and criminal justice system data.

In this post, we look at data on female victims of sexual violence, since most existing reports and statistics focus on women. Data on sexual assault against men are especially sparse; we know even less about the experiences of male victims.

Two different surveys, two different stories

There are multiple national surveys that study sexual assault prevalence in the United States, but the estimates produced by these surveys vary immensely. In 2011, the National Crime Victimization Survey (NCVS) estimated an annual rate of rape and sexual assault victimization of 1.6 in 1,000 women. That same year, the National Intimate Partner and Sexual Violence Survey (NISVS) estimated that 16 in 1,000 women were raped—about 10 times higher than the NCVS estimate.

These surveys yield vastly different estimates in part because of the way they ask their questions.  For instance, it is widely believed that the NCVS underestimates prevalence of sexual violence because it focuses on crimes and criminal behavior. Respondents may not always think about experiences of sexual violence as criminal incidents, or be willing to label themselves as victims of rape or sexual assault.

The NISVS, on the other hand, uses explicitly worded questions about specific behaviors to ask respondents about incidents they have experienced, a technique which results in higher estimates of prevalence and is considered a best practice in measuring sexual victimization. The differences between these two surveys reveal that the way sexual assault research is conducted has a massive impact on how much of the problem is detected.

Criminal justice data can be misleading

When data from criminal justice agencies, like reported crimes, are used to estimate sexual assault prevalence, measurement problems become even more pronounced. Alarmingly low percentages of victims report incidents of sexual violence to the police, especially when compared with reporting rates for other severe violent crimes.

The NCVS estimates that only 36 percent of rape or sexual assault incidents between 2005 and 2010 were reported to the police. Women provided a variety of reasons for not reporting incidents to police, including fear of reprisal, a feeling that police would not do anything to help, and the view that the incident was a personal matter. Research suggests that many victims have experienced incidents that qualify as rape or sexual assault, but do not think about the incident in those terms.

The potential for underreporting increases as sexual assault cases advance through the justice system. According to the NCVS, out of the 36 percent of incidents reported to police from 2005 to 2010, police responded in only 84 percent of cases. Of those   incidents that had a police response, only 86 percent involved the police taking a report. Ultimately, a police report was generated for only about a quarter of the actual rape or sexual assault victimizations reported by the entire survey sample.

Things only get worse at court. One national study found that only about 8 percent of rape victimizations resulted in the perpetrator being criminally prosecuted. Those who committed rape were convicted in 3 percent of cases, and incarcerated in only 2 percent of cases.

Defining and measuring rape and sexual assault involves many measurement challenges, not to mention additional emotional, cultural, and practical challenges. Thus, any one single measure is unlikely to paint a complete picture. The persistent issue of underreporting and the lack of data consistency can make it hard to determine the full extent of the problem, to understand how best to help victims, or to know if anything we do is making a difference.

When rates vary across studies, policymakers are able to choose which statistics they want to lean on, even if the evidence suggests those statistics may be underestimating the true scope of the problem.  Understanding these measurement problems is an important step toward a more complete understanding of sexual assault and more effective policy responses.