By Michael J. New, University of Alabama

I want to thank Professor Wright for his response to my latest article about his abortion research. Before I respond to the specific points he raises, I want to make it clear that I have not accused Professor Wright of being dishonest. In my most recent posting on MoralAccountability.com, I publicly commend him for admitting that a mistake was made in his original data collection. This is something that many researchers in his situation would not have done. While I think that Catholics in Alliance for the Common Good should have made some kind of public announcement about the revised study, I in no way fault Professor Wright for this.

On to Wright’s specific points:

- First, I agree with Wright’s interpretation of some of the economic variables. My research and other research suggest that abortions decline when the economy is doing well.

- Since Wright does not make public his new findings on how spending on the Women Infants Children (WIC) program affects abortion rates, we cannot be sure if the changing results in this particular case were due to either his use of new post-1997 abortion data, or the elimination of potentially biased data.  However, since Wright says the new findings were due to the use of new post-1997 data, I will take him at his word for it.

- I am not arguing that the statistical model that Wright uses is a bad model as such. My concern is that when this model is used to analyze the metric he chooses to measure welfare benefits, it may be producing some spurious results.  In his regression model, Wright analyzes per capita AFDC/TANF spending. This variable was at a high level during the early 1990s and fell during the rest of the decade. This is largely because fewer people were receiving welfare as the economy improved. As such, I am concerned that this variable is, in effect, serving as a proxy for the strength of the economy and does not actually measure the relative generosity of state welfare benefits.

Now, the regression results indicate that increases in this welfare variable are correlated with long term reductions in the abortion rate. My concerns are 1) the reductions in abortion rates during the 1990s might be due to the short term effects of the improving economy and not the high levels of per capita welfare payments in the early 1990s and 2) the states with the very worst economies in the early 1990s might have had the sharpest increases in both per capita welfare benefits and abortion rates and hence might have seen the most substantial declines in both variables (not because of the high level of early 1990s welfare benefits, but due to more dramatic improvements in their economic conditions). That is why I am concerned that these findings might be spurious.

In his working paper, Wright analyzes some other metrics to measure the amount spent on welfare benefits and finds similar results. This is certainly a welcome addition to this study. I have not had time to examine these other metrics carefully. However, Wright’s analysis would be more persuasive if he included in his regression a variable measuring state AFDC/TANF benefit levels for a single mother on welfare. The magnitude of such a variable would not fluctuate with the strength of the economy. Furthermore, I also think it would be a good idea, if Wright would run some regressions starting with data from the mid 1990s or another time when per capita welfare spending was not unusually high.

- I am a little puzzled at Wright’s apparent consternation at my decision to list a series of peer reviewed studies which indicate that pro-life laws reduce the incidence of abortion.  If one is interested in a particular social science question, consulting a range of peer reviewed literature certainly seems like a reasonable way to proceed.  Even Phillip Levine, whose work Wright praises, suggests in some of his research that pro-life laws reduce the incidence of abortion. Furthermore, in Sex and Consequences Levine conducts a thorough review of the research on pro-life legislation and concludes  ”the evidence generally supports the proposition that abortion restrictions reduce the likelihood of abortion” (Levine 1996; 132). [1]

In the literature review included in his working paper, Wright himself cites about 5 studies which show that pro-life laws reduce abortion rates.  Even though he states that the literature has “yielded mixed results” the only studies he cites with contrary findings are one that analyzes parental involvement laws on overall (not minor) abortion rates (Blank 1996)[2] and another study that only analyzes one year of abortion data (Medoff 2002).[3] Between Wright’s working paper and his dialogue with me, he has only identified one methodologically sound, peer reviewed study which finds that abortion laws have a marginal effect (Haas-Wilson 1997).[4] Conversely, there are numerous studies showing that public funding restrictions and parental involvement laws reduce the incidence of abortion. These studies utilize a variety of statistical techniques to analyze data from a range of sources and years. The findings do not seem to “depend on the data used and the time period covered” as Wright suggests.

Furthermore, if Wright would simply acknowledge the differences between passed and enacted informed consent laws or admit that his findings on parental involvement laws might be different if he were analyzing minor abortion rates, he would seem more fair minded to those of us who feel that various types of pro-life laws are effective  at reducing abortion rates.

- Finally, I appreciate the fact that Professor Wright thanks me both in his Catholics in Alliance for the Common Good paper and the working paper available on his website. I am also pleased that he was willing to run some regressions where he excluded data from states where data collection may have been inconsistent over time. That having been said, he has not pursued my other suggestions.  For instance, he has yet to 1) publicly present results of regressions on the full 1982 to 2000 dataset with weighted data, 2) analyze minor abortion data (which is relevant to his research since he analyzes pro-life parental involvement laws), 3) include in his regression state TANF/AFDC benefit levels for a single mother on welfare, 4) provide an explanation as to why he excluded certain datapoints from Florida (1997-2000), Louisiana (1997-2000), West Virginia (1992-1993), Wisconsin (1992-1993) and Wyoming (1992-1993). In fact, these excluded datapoints are mentioned neither in his Catholics in Alliance for the Common Good study nor his working paper.

- I will conclude by saying that sometimes social scientists simply disagree. This can be healthy. Often times, the peer review process can help resolve disagreements about empirical questions.  However, it is unfortunate that Professor Wright seems to interpret reasonable concerns about his analysis as personal attacks on his character.

Michael J. New is an Assistant Professor of Political Science at the University of Alabama and is a visiting fellow at the Witherspoon Institute.

[2] Blank, Rebecca, Christine George, and Rebecca London. 1996. “State Abortion Rates: The Impact of Policies Providers, Politics, Demographics, and Economic Environment.” Journal of Health Economics 15: 513-553.

[3] Medoff, Marshall.  2002. “The Determinants and Impact of State Abortion Restrictions.”American Journal of Economics and Sociology 61: 481-493.

[4] Haas Wilson, Deborah. 1997. “Women’s Reproductive Choices: The Impact of Medicaid Funding Restrictions.” Family Planning Perspectives 29(5) 228-233.