Final Assignment

Donahue and Levitt present a relatively convincing argument for a link between abortion and crime rate. However, before I go into detail about their article, it is pertinent that I discuss how Freakonomics treats the work. Obviously, with one of the authors of the article also being an author of Freakonomics, the chapter pays a close attention to the article as written by Donahue and Levitt and little about the criticisms from the comment by Foote and Goetz. That is far from saying that the chapter in the book devoted to this topic is not with merit.

Dubner's style of writing is able to put a particular spin on something that would be of a boring nature to the average reader. He is able to make the topic of the link between abortion and crime rate an accessible one. To begin the chapter with an anecdote to the story of Romania makes for understanding the reverse of what happened in America in the past few decades; ie criminalized abortion and high crime rate. The chapter treats the topic of Donahue and Levitt's paper in full following this anecdote by providing reasons for why other suggestions of why the crime rate has dropped so significantly in the 90's just doesn't pan out. It then of course poses a decent summary in well versed prose of Donahue and Levitt's findings.

Donahue and Levitt's article of course posits that there is a relationship between abortion and crime rate, potentially a causal one, and that the legalization of abortion in the early 70's can explain at most 50% of the drop in the crime rate in the 90's. If true, this is indeed a significant find. Donahue and Levitt achieve this by analyzing differences in states that legalized abortion before Roe v. Wade, against states with abortion illegal. Likewise, they compare trends before the historic 1973 Supreme Court case and after, finding that the cohort born after the legalization of abortion was less likely to contribute to violent crime compared to the pre-legalization cohort. Comparisons within states also seemed to reveal a negative relationship between abortion and violent crime rate. The most interesting part of this study is actually how it analyzes the different cohorts, and is able to find that the cohort affected by Roe v. Wade had less crime on average than the cohort born before it.

Foote and Goetz's comments are of course concerning. The most startling realization I had from reading their comments was to consider the effect the cross sectional data would have on Donahue and Levitt's results. It doesn't entirely make sense to compare cohorts across states, many of which have several different demographic make ups and of course different laws which might affect crime rates outside of the model that was considered in Donahue and Levitt. The comment that the in-state regression is flawed from a coding error, which after being corrected Foote and Goetz found no relationship between abortion and crime, is also concerning as it raises the issue of the validity of Donahue and Levitt's results.

It is difficult to reconcile Foote and Goetz's comments with the chapter four of Freakonomics. The book never truly analyzes the problems with cross sectional data, nor does it address the concern posited by Foote and Goetz that there is a coding error in the regressions presented in Donahue and Levitt. That said, the comments by Foote and Goetz do not make Donahue and Levitt's model entirely invalid. Foote and Goetz were unable to prove false the cross sectional regressions done by Donahue and Levitt, other than positing that the effect might by due to the nature of the cross sectional data rather than any actual link between abortion and crime. However, they were unable to show this. Therefore, Donahue and Levitt's model does indeed have some validity, as Foote and Goetz cannot fully reconcile the weakness of cross sectional data and the results in Donahue and Levitt, nor do the posit other possible explanations for the reason as to why crime rates have dropped so significantly in the past few decades.