The furor over the study that attributes falling crime partly to abortion may tell us as much about America as about crime. If you missed it, the study -- done by economist Steven Levitt of the University of Chicago and law professor John Donohue III of Stanford University -- concluded that half the drop in crime since 1991 might reflect the Supreme Court's 1973 Roe v. Wade decision legalizing abortion. The argument is that some potential criminals weren't born, because their mothers had abortions. If true, the country got a slightly better crop of teenagers by the 1990s.

But is it true? We may never know. Crime's decline is one of the great mysteries of the 1990s. Between 1991 and 1997, the murder rate fell 31 percent; the rates for all violent crime (murder, rape, assault, robbery) and property crime (burglary, auto theft) dropped 19 percent and 16 percent. No one predicted this, and the usual theories -- better policing, tougher sentences, lower unemployment -- don't seem to explain it fully. Frustrated, Levitt and Donohue (who have both studied crime extensively) wondered whether abortion might matter.

They convincingly say that their aim is to understand crime, not promote abortion. What impressed them were the sheer numbers. "I don't think that people realize that [about] one in four pregnancies ends in abortion," says Levitt. Indeed. In 1992 there were 4.1 million births and 1.5 million abortions. But their inquiry is the social-science equivalent of "don't ask, don't tell." Few politicians want to cede credit for crime's decline to a distant Supreme Court ruling. The press is also uninterested. The Chicago Tribune broke the study with a front-page story, but most coverage has been minimal. The Washington Post ran a short story on Page A9; the New York Times story (on A14) didn't appear for almost two weeks.

This reticence, of course, reflects the bitter abortion debate. Interestingly, both sides deplore the study. "Fraught with stupidity," says one antiabortion group. If you believe abortion is murder, the idea that it's an anti-crime device is outrageous. Consider the grim arithmetic: Between 1991 and 1997, the annual number of homicides dropped by 6,500 (from 24,700 to 18,200), while abortions regularly exceeded a million a year. But pro-abortion groups also dislike the study. They promote abortion as a woman's right, not as a covert means of social control -- weeding out criminals and incompetents -- with racial overtones. Blacks and other minorities account for about 40 percent of abortions.

The temptation is to embrace self-censorship: Let's drop the issue. But will the country be better off if we do? Not really. The probable result is that the study's conclusion will quietly infiltrate popular wisdom -- be accepted as accurate, even if it isn't openly discussed -- when it may or may not be true. And our understanding of crime will suffer if we can't debate one plausible cause of its decline.

What's the debate, then? Well, even Donohue and Levitt don't claim that abortion fully explains lower crime. A previous study by Levitt, for example, concluded that keeping criminals in prison has a huge effect on crime; by his estimate, the premature release of one prisoner results in 15 crimes over the next year. The new study attributes half the drop in crime to more prisoners (between 1987 and 1997, the prison and jail population doubled to 1.7 million). But the study also presents powerful, if circumstantial, evidence for abortion's role.

First, the decline in crime began in 1992, nearly two decades after Roe v. Wade. This is just when youths who would have been born in the mid-1970s would have hit their peak crime years (roughly between 18 and 24).

Second, five states legalized abortion before Roe v. Wade (Alaska, California, New York, Hawaii and Washington). These states had some crime declines before other states that didn't allow abortion until after the court's ruling.

Third, states with higher abortion rates in the mid-1970s have now experienced steeper drops in crime after controlling for other factors (police, prison populations, poverty, unemployment).

Unwanted children may suffer most from parental neglect, say Levitt and Donohue. Or the children of poorer, less-educated parents may be more crime-prone. If abortion reduced crime, it could have other social consequences. Perhaps more people are employable, because fewer unemployables were born. This may have aided welfare reform. But any effects will be doubtful if abortion's connection to crime is mostly a statistical coincidence.

And it might be. Everyone agrees that the crack-cocaine epidemic ignited a firestorm of violence in inner-city neighborhoods. By the mid-1990s, this subsided as a result (perhaps) of more arrests, murders and the settling of turf wars. Donohue and Levitt don't think this completely explains lower crime. But it could. Mobilizing against crack may have made police more effective. Their theory also requires that youth crime rates fall in the early 1990s. But the data may not fit this. Social commentator Steve Sailer -- in an analysis for -- notes that homicide arrest rates for teens rose until the mid-1990s. Drug arrests also increased.

Indeed, it's possible that legalized abortion increased crime by contributing to family breakdown. In a 1996 study, economists George Akerlof of the Brookings Institution and his wife, Janet Yellen, until recently the chairman of the Council of Economic Advisers, argued that Roe v. Wade and contraception had helped cause the explosion of single-parent families. Men felt less responsible for the children they fathered, because women could avoid or abort pregnancies. "Shotgun marriages" virtually vanished.

The truth is that we don't know the truth. Even if Donohue and Levitt are correct, the abortion debate should remain one of moral values. There are other ways to avoid unwanted children: abstinence, birth control. But it's delusional to pretend that something as common as abortion is without social consequences. We cannot find them unless we look. The trouble is that in modern America people often won't search for what they fear they might find.