Fighting Abortion With Crime?

The Donohue and Levitt article builds upon the chapter we read from Freakonomics last week. They begin by discussing how the common explanations for the drop in crime in the 1990s did not actually contribute that much. Rather it was “Roe v. Wade legalizing abortion nationwide potentially fits the criteria for explaining a large, abrupt, and continuing decrease in crime” (380). They argue that abortions reduce the cohort size of the group most likely to commit crimes. In this article we are introduced to the model used by the authors to build their case for abortion as the main contributor to decreasing crime rates. From the data they gathered, Donohue and Levitt find that “legalized abortion may account for as much as one-half of the overall crime reduction” (414). In short, this paper provides the statistical analysis behind the chapter in Freakonomics.
Foote and Goetz’s response discusses three criticisms of the Donohue and Levitt article. First, Donohue and Levitt “are missing a key set of regressors because of a coding error” (408). This affected the coefficients they calculated in their regressions. Second, they do not use per capita arrests but total arrests. Per capita arrests, Foote and Getz argue, are needed to truly determine the effect of abortions on crime. The third criticism is that Donohue and Levitt run regressions that do not allow for “differential state trends based on statewide crime rates that predate the period when abortion could have had a causal affect” (407). Basically, there is an error in the functional form of Donohue and Levitt’s early tests. Foote and Goetz address concerns raised by Donohue and Levitt such as measurement error and omitted variables, and create their own model to test abortion’s effect on crime; they find “no compelling evidence that abortion has a selective effect on crime” (421).
After comparing these two papers, the results of chapter 6 in Freakonomics must be taken with a grain of salt. Two completely different results can be obtained depending on the model being used to obtain them. Donohue and Levitt believe that measuring total arrests by cohort will show the impact of abortion on crime. Foote and Goetz argue that per capita arrests by cohort is a more accurate measurement. They also believe that there are several limitations to Donohue and Levitt’s analysis, and that correcting for them yields much different results. These papers thus show that there is always at least one perspective for each analysis.


About this entry