S ince abortion was legalized in the early 1970s, roughly I to 1.5 million abortions are performed each year. Recently, some have conjectured that the magnitude of these numbers may be altering political outcomes. If those pregnancies that were aborted were more likely to have been democratic voters, the reasoning goes, then there are far fewer democratic voters now than there would have been had the Supreme Court ruled differently in Roe v. Wade. This effect has been labeled the "Roe Effect." To date, the Roe Effect remains pure conjecture. Yet over the past decade or so, a considerable amount of academic research has been conducted, and continues to be conducted, on the impact of abortion policy on a range of social outcomes. The knowledge gained from this research can be brought to bear on the merits of the argument. What was the impact of abortion legalization on fertility? Did the reduction in fertility represent a permanent one, or just a shift in the timing of childbearing? Did abortion legalization alter the attributes of those children who were born? All of these issues speak directly to the feasibility of the Roe Effect. Although it is impossible to determine credibly the validity of the hypothesis without a counterfactual, it is possible to address whether the argument is at all consistent with what we know about the effect of abortion legalization. That is the goal of this essay. The first point that should be made is that the Roe Effect is based upon a faulty premise. James Taranto, who first proposed the Roe Effect, states "it is a statemerit of fact, not a moral judgment, to observe that every pregnancy aborted today results in one fewer eligible voter 18 years from now." He then points out that 40 million abortions have been performed since 1973, when the Roe decision was handed down. Larry Eastland similarly expresses that "there were 12,274,368 in the Voting Age Population of 205,815,000 missing from the 2000 Presidential election, because of abortions from 1973-1982." These statements are simply incorrect. They are based upon a common, but faulty, assumption that women have no control over their likelihood of becoming pregnant. If people engage in sexual activity (or not), or choose to use birth control (or not), independent of outside influences, then their statistical statements would be valid, In other words, if pregnancies were independent of outside influences, then every additional abortion would reflect an additional birth (well, almost we still would need to ignore the fact that some of the aborted pregnancies would have resulted in miscarriage, and women who abort can get pregnant again sooner). One could plausibly argue the contrary. Couples make decisions about their sexual activity and/or contraceptive use, taking into consideration relevant factors in their environment. This is the fundamental premise of an economic model of fertility. In such a model, women/ couples make decisions regarding pregnancy and childbirth by weighing the relevant benefits and costs, monetary or otherwise, The legal status of abortion is clearly an important component of the cost of a pregnancy. Even if childbirth is not the goal, women would be more reluctant to get pregnant if they knew that ending that pregnancy would be very expensive, as it would be if that resolution was an illegal abortion. Therefore, there is no obvious relationship between the number of abortions performed and the number of children who would have been born in the absence of those abortions, An additional abortion may represent a child who would have been born otherwise, or a pregnancy that would not have occurred otherwise. Yet the credibility of the Roe Effect has little to do with the rhetorical nature of the claimed magnitude of missing democratic voters, All that is required is that abortion [egatization needs to have reduced the number of children born. In fact, the impact on the number of children born does not have to be large at all for the effect to work. Although it sounds good to claim that 12 million additional voters could have swayed the 2000 election, it turns out that only a few hundred would have done ~he 'Lrick. This leads us to the first question thai can be addressed by academic research: By how much did abor-