Determining the true number of rape victims across the country is a surprisingly slippery task, a new report by the Centers for Disease Control (CDC) makes clear. For years, the gold standard for rape statistics has been the National Crime Victimization Survey (NCVS), conducted by the Department of Justice via phone calls. But a CDC report released Friday suggests that the NCVS is failing to capture a large portion of victims—as much as 88 percent.

The recent CDC report, based on surveys conducted in 2011, found that almost one in five women (and 1.7 percent of men) have been raped in their lifetimes. In a single year, 1.6 percent of women reported experiences that are considered rape—almost two million cases. But the NCVS report recorded just 243,800 cases of rape or sexual assault in that year, 12 percent of the CDC findings. Meanwhile, a report compiled by the FBI, which only documents cases that were brought to police, shows only 83,425 rapes that year.

So why the massive disparities between these numbers? Partly, it’s because the CDC and Justice Department reports have different goals. While the NCVS simply seeks to record the incidence of crimes across the country, the CDC approaches sexual assault as a public health issue. That affected the kinds of questions the surveys used to determine which respondents were rape victims.

The NCVS’s questions required the victims to identify the crime as a rape or sexual assault, asking questions like “How were you attacked?” and providing answers like “raped,” “tried to rape” and “sexual assault other than rape and attempted rape.” The CDC, meanwhile, did not mention any legal terms in the survey queries, instead asking questions like, “How many people have ever used physical force or threats to physically harm you to make you have vaginal sex?” Because of this, the CDC included cases in which the victim might not have been aware or willing to identify her experiences as rape.

This difference made the CDC’s survey broader, especially in the case of victims who were under the influence during the attack. The CDC counted alcohol- and drug-facilitated rape, asking if the respondents had ever experienced various sex acts while “drunk, high, drugged, or passed out and unable to consent.” But, as Scott Berkowitz at RAINN, the Rape and Incest Abuse National Network, pointed out, not all of those 1.2 million cases in 2011 would be considered rape by the Department of Justice. Due to the survey question's phrasing, a person who had been drunk—but still considered herself capable of giving consent—might have answered yes to that question. A CDC spokesperson clarified that being unable to consent is key to the CDC's definition of rape. 

Still, the CDC numbers are a reminder of how many sexual assaults and rapes go unreported. The total number of rapes reported to police in 2011 was 83,425—far lower than either the NCVS or CDC numbers. If the 2011 CDC estimate—almost two million people cases—all fit the legal definition of rape, that would mean only 4 percent were reported to the police. Even excluding alcohol- and drug-facilitated rapes, the 716,000 counts of completed or attempted penetration recorded by the CDC still add up to more than eight times the cases recorded by the FBI and almost three times as many as the Department of Justice. While finding an indisputable number of rape victims seems to be a Holy Grail, the CDC report certainly reveals that the most widely accepted estimates aren’t high enough.

This article has been updated with a response from the CDC.