Don’t miss Joseph Wright’s response, “Abortion and Public Policy.”
Pro-lifers who visibly and vocally support a pro-choice presidential nominee can be found in every election cycle. Over the years their arguments have adopted a boring consistency. They criticize Republican presidents for their inability to nominate judges who will overturn Roe v. Wade, credit President Clinton for the 1990s abortion decline, blame President Reagan for the fact the abortion rate did not decline much during his administration, and talk vaguely about the importance of other issues not related to the sanctity of human life.
This year pro-life supporters of Barack Obama have at least tried to add a little methodological rigor to their arguments. Many have been eagerly citing a study that was released in August by Catholics in Alliance for the Common Good. This study analyzes abortion data from almost every state from every year from 1982 to 2000. The spin given to the study is that it shows that state level pro-life laws only have a marginal impact on the incidence of abortion. Furthermore, increased expenditures on TANF and other welfare programs reduce abortion rates.
But this isn’t quite right. A closer look at the results provides evidence of the effectiveness of pro-life legislation. Furthermore, there are methodological problems with the analysis performed by Catholics in Alliance for the Common Good. A correct analysis demonstrates that pro-life laws are effective—and casts serious doubts about whether more generous welfare benefits are correlated with reductions in the incidence of abortion.
What the Catholics in Alliance Study Really Finds
Suppose one takes the results of the Catholics in Alliance study at face value. Contrary to the spin, the results actually provide some evidence that pro-life legislation is correlated with abortion declines. The authors analyze four different types of state level abortion policies: 1) public funding of abortion through Medicaid, 2) parental involvement laws 3) partial birth abortion bans and 4) informed consent laws. I will discuss their analysis of each of these laws below.
1. Medicaid Funding of Abortions
In the Catholics in Alliance study, the authors run two regressions that analyze abortion data from nearly all 50 states from every year from 1982 to 2000. In the first regression, the authors find that public funding of abortion by Medicaid increases state abortion rates by approximately 13 percent. This finding is statistically significant and is consistent with other research on the issue. In the second regression the authors find that Medicaid funding of abortion increases state abortion rates by approximately 10 percent. This finding does not reach conventional levels of statistical significance, but comes extremely close. However, the authors give little attention to the significant finding in the first regression and emphasize the (narrowly) insignificant finding in the second.
2. Parental Involvement Laws
The results of the Catholics in Alliance study indicate that state level pro-life parental involvement laws only have a marginal impact on state abortion rates. However, the authors overstate the importance of this finding. Their study analyzes overall abortion rates, not minor abortion rates. Since parental involvement laws would only have a direct effect on minors, it is unsurprising they only have a marginal effect on the overall abortion rate. In addition to this, there is significant social science evidence that pro-life parental involvement laws are able to reduce abortion rates among minors. Furthermore, my recently released study shows that more protective parental involvement laws—those that require parental consent and those that require the involvement of two parents—result in even larger decreases in minor abortions.
3. Partial Birth Abortion Bans
The authors find that state level Partial Birth Abortion Bans have little effect on the incidence of abortion at the state level. This finding is consistent with my studies and other research on the subject.
4. Informed Consent Laws
The one area where the results of the Catholics in Alliance study contradict previous research involves their analysis of informed consent laws. They use two separate variables to analyze informed-consent laws at the state level. The first variable indicates whether a state passed an informed-consent law and the second indicates whether or not a state informed-consent law was enacted. The difference between these two variables is that two states, Michigan and Indiana, both passed informed-consent laws in the 1990s, but their enactment was delayed due to judicial challenges.
The results of the Catholics in Alliance study indicate that when a state passes an informed consent law, it results in an 18 percent increase in state abortion rates. When a state puts this law into effect, the abortion rate declines by approximately 18 percent. This means the total impact of an informed consent law that is both passed and enacted is close to zero.
However, the fact that enforced laws are much more effective than unenforced laws provides evidence that informed consent laws work. More importantly, the finding that unenforced informed consent laws increase abortion rates by 18 percent causes one to seriously question the other findings of this study. Quite honestly, there is no plausible explanation why an unenforced law should result in such a sharp and statistically significant increase in the incidence of abortion. Aberrant results like this are often evidence of some underlying problems with the statistical model. These methodological problems will be discussed in the next section.
Where Catholic Alliance for the Common Good Went Wrong
There are two main problems with the methodology used by the authors of the Catholics in Alliance study.
1. Inclusion of Potentially Biased Data
The authors are not able to analyze abortion data from every state from every year from 1982 to 2000. This is for a couple reasons. First, during the late 1990s California, New Hampshire, Oklahoma, and West Virginia did not report abortion data to the Centers for Disease Control. Second the authors exclude Alaska and Hawaii in their analysis and exclude certain years from Florida, Iowa, Louisiana, West Virginia, and Wyoming.
However, there exist other data points that should have been excluded as well. In certain years, Alabama, Iowa, and Illinois only reported data for abortions performed in hospitals. Since a relatively high percentage of abortions are performed in clinics, this underestimates the actual number of abortions performed in these states. As such, this data should have been excluded from their analysis.
Furthermore, the authors should have excluded data from Kansas. Kansas is a statistical outlier for a couple reasons. First, for every year between 1992 and 1999, 40 percent of the abortions in Kansas were performed on out-of-state residents. This is by far the highest percentage in the country. Second, between 1991 and 1999 the abortion rate in Kansas increased by 69 percent. This increase cannot be attributed to changes in demographics, economics, or legislation and is instead likely due to notorious late term abortionist George Tiller expanding his abortion practice. As such excluding data from Kansas seems reasonable.
2. Failing to Weight The Data
More importantly, the authors of the Catholics in Alliance study do not weigh their data by a measure of state population. This is a mistake for a couple reasons. First, small fluctuations in low population states can cause large percentage increases, potentially biasing the results.
Secondly, unweighted data distorts the analysis and makes it appear as if pro-life laws are less effective than they really are. Between 1992 and 1999 the overall national abortion rate declined by 16.7 percent (for the 46 states that reported data in both years). However, if one averages the abortion rates in each state in 1992 and 1999 (unweighted) the abortion decline comes out to 23 percent. This is because a number of low population states experienced large abortion declines during the 1990s.
The Catholics in Alliance study analyzes the effects of state level pro-life laws by comparing abortion rates in states that have adopted these laws to the overall national trend. Unweighted data exaggerates the national decline and makes these laws seem less effective than they really are.
Correcting the Analysis
The authors of the Catholics in Alliance study undertake a complicated statistical analysis where they attempt to differentiate between those factors that affect state abortion rates in the short term and those that affect state abortion rates in the long term. For my analysis, I collected the same data from the same set of states and years. The means and standard deviations of the variables I uses are very similar to those that they report in Table 4 of their study. However, despite my best efforts I was unable to replicate their results. As such, I proceeded to run three regressions using conventional statistical techniques.
The first regression is on the same set of independent variables over the same set of states and years that were analyzed in the Catholics in Alliance study.
The second regression keeps the same set of independent variables. However, I eliminate the biased data from Alabama, Iowa, Illinois, and Kansas, and weight the data by the number of women in each state who are of childbearing age (15 to 44).
The third regression is identical to the second except for the fact that I add some additional explanatory variables. I include variables that measure the racial breakdown and age demographic of women of childbearing age. Additionally, to better measure the effect of the economy, I also add variables that measure the change in the state unemployment rate and the real change in per capital personal income. These regression results can all be found in Table 1.
The first regression provides little evidence that more generous welfare benefits reduce abortion rates. The regression results indicate that more state spending on TANF or AFDC actually increases the incidence of abortion. Furthermore, the regression results indicate that Family Caps—policies that do not grant additional welfare benefits to women to have additional children out of wedlock—actually reduce state abortion rates. This regression also provides little evidence that pro-life laws are effective at reducing abortion rates either. None of the variables measuring the impact of various types of pro-life laws achieve statistical significance.
However, when potentially biased data is removed, the data is weighted by the population of childbearing women, and additional demographic and economic variables are included, the results change. Once again, the variables measuring welfare spending are insignificant. But more importantly, the regression results indicate pro-life laws are effective. Specifically, Medicaid funding of abortions increases abortion rates and informed consent laws reduce abortion rates. Both of these findings are statistically significant. Interestingly, in the final regression model only 4 of the 24 independent variables achieve conventional standards of statistical significance. Two of these four variables measure the effect of pro-life laws. Overall these findings contribute to the body of academic and policy literature which argues that pro-life laws are effective at lowering abortion rates.
Some Final Thoughts on Informed Consent Laws
The primary area of disagreement between me and the authors of the study published by Catholics in Alliance involves the effectiveness of informed-consent laws. As such, I wanted to further analyze these informed-consent laws. National Right to Life’s 1996 Convention Yearbook highlights 8 states that were the first to adopt Casey style informed consent laws. These are informed consent laws that are modeled after Pennsylvania’s law that was found constitutional in the Supreme Court’s 1992 Casey vs. Planned Parenthood decision. All of these laws came into effect in either 1993 or 1994. Table 2 compares the abortion decline in these 8 states to the national average.
Examining these data, one finds that the overall national abortion rate declined fifteen percent from 1992 to 1999. Interestingly, six of the eight states that first adopted Casey style informed-consent laws experienced above average abortion declines during the 1990s. Mississippi led the way with an impressive decline of 54 percent in its state abortion rate. Now, it is true that Ohio’s abortion-rate decline was somewhat below average; however, there are reports of enforcement problems with its informed-consent law. Similarly, North Dakota’s abortion rate declined little, but since North Dakota has one of the lowest abortion rates in the country, the small decline is unsurprising. Overall, the average abortion decline in these 8 states is approximately 25 percent, 10 percentage points greater than the national average. It appears that these informed consent laws played a role in these abortion declines.
This summer Catholics in Alliance for the Common Good released a study which questioned the effectiveness of some types of state level pro-life legislation. It argued that greater spending on welfare programs was a better strategy for lower abortion rates than enacting pro-life laws. In some respects they actually deserve credit. Catholics in Alliance attempted to make a meaningful contribution to the ongoing debate about pro-life voting by commissioning a methodologically rigorous study to analyze a complicated issue—the incidence of abortion at the state level.
Unfortunately, their study did not acknowledge any of the previous academic or policy research on pro-life legislation. As such, they did not engage or find fault with previous research indicating that pro-life laws were effective. Overall, it seems that Catholics in Alliance was primarily interested in making the case that welfare spending was the best way to reduce abortion. They even refused to properly acknowledge and publicize their own findings which indicated that certain types of pro-life laws were effective.
My analysis of their data indicates that welfare spending only has a marginal impact on the incidence of abortion. Additionally I find that both public funding restrictions and informed-consent laws are effective at reducing state abortion rates. This adds to the body of research which finds that pro-life laws are effective. Hopefully this sizeable body of work will inform ongoing discussions among Catholics, pro-lifers, and others who are deciding how best to cast their vote on Election Day.