"The Secret Bench of Knowledge," sculpture by Czech-born Canadian sculptor Lea Vivot. Photograph: teachandlearn

“There is nothing more horrible than the murder of a beautiful theory by a brutal gang of facts.” So wrote the French nobleman and essayist François de La Rochefoucauld. But what happens if that gang turns out to be ineffectual, if the theory and the numbers that support it refuse to die? The world becomes inhabited by zombie theories and zombie facts.

Thankfully, there are intrepid fact-checkers out there to set us straight. When it comes to feminism, Christina Hoff Sommers, former philosophy professor and resident scholar at the right-wing think tank the American Enterprise Institute, is on the case. In print and in a weekly video blog called “Factual Feminist,” Sommers assures that most of what we hear about “the plight of American women” is false, that zombie facts have led to ill-advised legislation, and that women’s advocates would do better if they corrected their data and worked toward change based on true numbers.

Let us consider one of Sommers’s “feminist myths”: that one in five in college women is sexually assaulted. This is an important number. It shows up constantly in the media, not to mention on demonstration placards, on the walls of campus buildings, and in the stalls of college bathrooms. But is it true? It sounds high and a little, well, crafted, doesn’t it? Like a compelling sound bite based on manipulated data. It has the hallmarks of a zombie statistic.

The one-in-five factoid comes from a 2007 survey conducted at two large, public, four-year institutions (one in the Midwest and one in the South) using a random, self-reporting sample of 5,446 women. One-in-five is a lumpy statistic, combining the 12.6 percent of women who experienced attempted sexual assault and 13.7 percent who suffered actual sexual assault, meaning unwanted oral, anal, or vaginal penetration with any body part or foreign object. Accounting for overlap between attempted and actual assaults yields a total of 19 percent, which rounds up to that neat one-in-five stat.

So what is wrong with this number? For one thing, the demographics are narrow. The sample is limited to only two colleges, and students who had experienced attempted or actual assault may have been more likely to respond to the survey. More than half of the 14,000 women invited to participate didn’t respond. The data also don’t distinguish between important institutional categories—private versus public, large versus small, religious versus secular, etc. And while combining attempted and successful assaults may be reasonable for estimating possible danger, uncritical use of the resulting bigger number muddies the rhetorical waters. If we only count actual assaults, the figure changes to one in seven, which would still, it seems to me, warrant significant action. One in five sounds stronger, but its statistical deficits open the door to anti-feminist attacks.

Feminism has changed ideas of where knowledge comes from, who has authority to know.

Indeed, Sommers calls the one-in-five number a “baseless canard” and laments that it has become “the inspiration for new legislation and the focus of college programs.” Advocates of safe campuses should note this backlash and consider how a zombie statistic can steal the spotlight from real dangers that women, primarily, face. The pesky statistic becomes a distraction from important questions about how colleges and universities handle sexual assault: Should such institutions even adjudicate assault complaints, or are allegations better handled by the criminal justice system? How can the need to protect an assaulted person from retribution be balanced against the right of the accused to know what he is accused of and by whom? My own feeling is that universities should educate their constituents about the problem and enforce regulation of campus alcohol and drug use, as these often play a role in assaults. But criminal charges are best handled by the justice system. These issues, not the precise proportion of women who face assault, warrant debate.

The backlash actually does more than target a single questionable statistic. It also shores up conventional ideas about knowledge that feminists have sought to undermine. Since the 1980s feminist philosophers have pushed back at the dominant theory that the primary, and perhaps only, way of knowing something is by direct observation. Informally, philosophers refer to this as S-knows-that-P epistemology, in which S is an observer (the knowing subject) and P is a propositional claim about an observed thing (e.g., the pencil is on the table). S-knows-that-P epistemology assumes the most reliable source of measurable knowledge involves direct observation under controlled conditions. But feminist philosophers have argued that knowledge that matters to people’s lives is more complicated, something our discussion of rape illustrates.

Unlike, say, a pencil lying on a table, sexual assault is not a static object. Cultural tradition and the law define what counts as rape. During the 1970s feminists transformed rape. Consider the laws in New York State in 1971. To prosecute an assailant, a woman had to prove vaginal penetration, provide evidence of force, identify the perpetrator, and secure corroboration from a witness. Across the country variants of these laws ensured that there were relatively few complaints and even fewer arrests and convictions. To improve women’s lives by giving them better tools with which to defend themselves, feminists vigorously and successfully pursued reform. Definitions of rape were broadened to include oral and anal penetration, thereby encompassing a wider range of assault, including assault against men. New laws recognized that there might be degrees of sexual assault, from penetration to groping. Those assaulted no longer needed to resist or have third-party corroboration in order to make a complaint, and their identities could be shielded from public exposure. Finally, husbands could be prosecuted for raping their wives.

So, before 1970, the incidence of rape was lower in part because the definition of rape was different. The object of knowledge then was not the object of knowledge today. And the subject has changed, too. Then, the assaulted person did not count as a possible knowing subject, hence the law demanded a third-party observer.

Sommers and others who decry the one-in-five zombie statistic are picking at a deep philosophical difference: the difference between mainstream epistemology, which pictures a knower and object abstracted from society and power differentials, and a feminist philosophy that understands knowledge production to be a social project incorporating knowing subjects from many cultures and social strata.

It goes without saying that it is best to have the right facts, to not use zombie stats uncritically. The stronger the information, the better the chance to fix the problem at hand. But contemporary feminism is complex in terms of both ends and means. It is a political movement and, as such, strives to change lives—to make women safer and economically self-supporting and to improve men’s lives by lifting from their shoulders some of the weights of constructed masculinity. Like any political movement, compelling rhetoric is part of its success. But it is also an academic endeavor, and in this mode, feminism aims to democratize knowledge itself. It has changed our ideas of where knowledge comes from, who has the authority to know, and how we gather information.

Zombie statistics inhabit a messy space where politics and knowledge meet. In a sense they are hybrids. The headline survey uses feminist-inspired definitions of rape and treats the respondents as knowing subjects. But it uses a limited method to gather data, producing a striking but also vulnerable sound bite. A study that uses feminist approaches to gathering knowledge, such as narrative interviews and a more diverse sample of knowing subjects, might produce more robust knowledge. It would also challenge feminist activists to develop a different rhetorical approach less susceptible to right-wing counterattacks.