Joseph Wright, Assistant Professor of Political Science Penn State University.

The main findings of my research show that the abortion rate is responsive to socioeconomic factors in the 1990s, when the overall abortion rate in the United States fell dramatically. Higher male employment, lower poverty rates, and more generous economic assistance are all correlated with lower abortion rates

It is becoming increasingly clear that Professor New has either: (1) not taken the time to read the full working paper that contains the supporting analysis of the data; or (2) he simply does not understand the statistical methodology I employ. For example, in his latest attack on my research, New argues that “most decisions about sexual activity and pregnancy resolution are based on short-term factors.”

While this is certainly possible, I have employed a statistical method that allows me to test whether this is true instead of simply assuming it, as New does. The data unambiguously show that many socioeconomic factors have a long-term effect, as is clearly demonstrated in the graphs in the working paper. This is not something that I assume, but by using the proper statistical analysis, I test using the actual data.

While New dismisses this technique as “faulty” and “questionable,” there are important reasons for this choice. This type of statistical technique is particularly important if we think that some factors — such as state laws regulating abortion — may have a teaching effect that is not confined to a particular year. Consider a parental involvement law. In addition to having a possible direct effect on the teenage abortion rate, one would hope that these laws also have a teaching effect in which pro-life legislation sends a signal that obtaining an abortion (regardless of whether the mother is a teenager) is immoral and socially unacceptable behavior.  If the teaching effect is operative, then a parental involvement law might not only affect the overall abortion rate (for women ages 15-44) but passing and enforcing this type of law could have an effect over the long-term. For example, it is possible (and indeed all pro-life advocates would

hope) that passing a parental involvement law in 1999 when a teenage woman is not pregnant could still affect her decision over whether to obtain an abortion in 2005 when she is in her 20s and pregnant — regardless of whether the parental involvement law is on the books in 2005. Rather than assuming away the long-term effects of socioeconomic factors and the potential teaching effect of these abortions laws, I decided to use a statistical technique that tests whether these are present in the data.

In his most recent attack, New also appears quite invested in running a tally totaling the number of published articles with a particular finding. My approach has been to defer to the experts on where the state of the academic literature on this subject resides. One expert, the economist Phillip Levine, recently (2004) wrote an excellent book about public policy and abortion. In this book he summarizes the current literature on the subject (including most of the studies New cites) and makes the following conclusion about the empirical studies on abortion and economic assistance: “our knowledge regarding the impact of welfare reform and welfare generosity is limited”… and “[f]urther research in this area would be very helpful.” (p. 180) After reading Levine’s book, I took this as a useful starting place for my research on the socioeconomic determinants of abortion.

New insinuates on more than one occasion that I was trying to hide the updated analysis. In fact, I did just the opposite. Soon after I discovered that the data I used mixed two different time series, I posted an updated analysis for CACG as well as a 30-plus page working paper that discusses the statistical techniques I employ and reports the results of numerous robustness tests. Again contrary to New’s claim, any change in the reported results was due entirely to using the corrected data in the updated analysis. While New was quick to criticize the original study, I took these criticisms seriously and addressed them in the working paper. In the end, New’s concerns about the sample, the statistical technique, and the particular weighting of the data were all unfounded. In the working paper, I show that the main results of this study remain robust when addressing all of these criticisms: The research suggests that the abortion rate is correlated with socioeconomic factors and that economic assistance lowers abortion rates.

Personal attacks on academic honesty and the integrity of another person’s research in an effort to score political points have no place in a responsible discussion of important moral issues.

Joseph Wright is Assistant Professor of Political Science Penn State University.

Notes:

The academic working paper was posted online for public scrutiny in November 2008 when I posted the updated CACG study; the version I posted in November is intended as supporting evidence for the CACG analysis and only has my name on it.

Levine, Phillip B. 2004. Sex and Consequences: Abortion, Public Policy, and the Economics of Fertility. Princeton: Princeton University Press.