Following this evening’s debate, kindly organised by TCDSU, we want to look at a few claims about whether pro-life laws work or not.

Studies have shown that pro-life laws do not work. Countries that restrict abortion have abortion rates similar to those of countries where it is legal.

This claim comes from a paper written by Dr Gilda Sedgh, among others. Sedgh presented the findings of this paper to the Citizens’ Assembly and the Oireachtas Committee. She is a research scientist at the Guttmacher Institute. The paper is called ‘Abortion incidence between 1990 and 2014: global, regional and subregional levels and trends’ (2016) 388 (10041) The Lancet 258.

The paper estimates abortion numbers and rates for women in different world regions and claims that abortion is as common in countries where it is illegal or allowed only to save a woman’s life as it is in those where it is provided on demand. The only difference is that abortion is safe where it is legal and unsafe where it is illegal. An article in 2012 claimed the same thing: Sedgh et al., ‘Induced abortion: incidence and trends worldwide from 1995 to 2008’ (2012) 379 (9816) The Lancet 625.

Don’t those studies prove that our efforts are pointless?

No. There’s so much reason to be sceptical of those studies. First of all, they’re a collaboration between the World Health Organisation and the Guttmacher Institute. The Guttmacher Institute is an explicitly pro-choice organisation with former ties to Planned Parenthood, which obviously has a vested financial interest in extending abortion access. Imagine if the Pro-Life Campaign had collaborated on a study that found the opposite conclusion. Wouldn’t you be sceptical?

Funding for the study came from the UK government, the Dutch government, the Norwegian government (all countries with on-demand abortion regimes) and the David and Lucile Packard Foundation. This is an NGO which focuses on depopulation through family planning, i.e. abortion in order to save the planet. It would have been incredible if they had reached a result that did not favour legalised abortion.

Even though the researchers and funders may have been biased in favour of abortion, this does not make the findings incorrect.

No, but the study itself makes the findings strongly open to accusations of incorrect data analysis. First off, the Lancet studies do not compare like-with-like. Instead of looking at otherwise-similar countries that have variations in abortion laws, the study compares rich regions (like Western Europe and North America) with poorer regions (like Southeast Asia and sub-Saharan Africa).

This makes it extremely difficult to tell whether the trend toward lower abortion rates in Western democracies really reflects the success of safe and legal abortion, or whether it’s mainly a consequence of the enormous gap in wealth and development that still separates the West from the rest. If you look at Table 4 in the study, you can see that they’ve grouped together all countries that have specific kinds of laws regarding abortion, but regardless of the state of their medical system, levels of poverty and other factors that would be relevant to the incidence of abortion and maternal mortalities. The study fails to control for these realities, and therefore tells us very little.

So when they claim that Western Europe, with liberal abortion regimes, has a lower abortion rate than other regions which prohibit abortion, they’ve reached that conclusion by factoring in countries in Western Europe with very restricted abortion, like Ireland and Malta.

Then we have the fact that even the study itself admits that it can’t rely on its own figures: on page 265, it says, “Our estimates have several limitations. Information about abortion incidence in the developing world is scarce. The quantity and precision of data in developing regions are reflected in the wide uncertainty intervals around estimates for these regions. Empirical evidence to inform and validate the estimated rates for subgroups of married women is also lacking, and research on abortion incidence in these subgroups is needed.” Even they don’t really trust their own findings.

The study also uses a Bayesian hierarchical time series model. The Bayesian approach constitutes a radically different way of doing science: Bayesians categorically reject various tenets of statistics and the scientific method. Bayesians use probabilities for both hypotheses and data. This is philosophically iffy as hypotheses do not constitute outcomes of repeatable experiments in which one can measure long-term frequency.

However, most importantly, the Bayesian approach forces you to push subjectivity to the beginning of the analysis, rendering the end result contingent.

To Bayes, py(y) in the theorem py(y|x) = px(x|y)py(y)/px(x) [where py(y|x) = the conditional probability density function for Y given X = x, and px(x|y) = conditional pdf for X given Y = y] represents the investigator’s subjective prior beliefs about whether a fixed parameter takes the value y. This parameter is an unknown variable, and the prior probability density function denoted by py(y) measures the relative strengths of belief about possible values of the parameter. The Lancet study offers no validity for the chosen prior. They go into the study hoping that they reach a specific conclusion, they structure the model so that they reach that specific conclusion, and then they reach that conclusion. Frankly, it would be amazing if they did anything other than reach that conclusion. Bayesian methods can and do work very well, but only when an article based on Bayesian analysis is transparent with regard to which priors were used and why. To win the intellectual battle, we have to become passionate frequentists (the opposing school of probability) and/or sticklers for strong Bayesian priors.

Scientific American says that Bayes’ theorem ‘can … be used to promote superstition and pseudoscience’.

Click to access Reasons-for-belief.pdf

If anyone is interested in the philosophy of stats, with more on Bayesian disadvantages, this is a good resource.

It still doesn’t make them wrong.

No, but here are some studies that show that laws prohibiting abortion do affect abortion rates. By showing links between abortion access and birth/fertility rates, we show that laws restricting abortion are successful, insofar as children being born, not aborted, can be considered a success (it can):

  1. Edith Gutiérrez Vázquez and Emilio Parrado, ‘Abortion Legislation and Childbearing in Mexico’ (2016) 47(2) Studies in Family Planning 113: We documented a systematic association between legalisation and fertility change in Mexico. In comparing change in childbearing propensities between 2000 and 2010, we estimated that abortion legalisation reduced the number of births in Mexico City by an additional 4 percent relative to the changes that would have occurred without the law. Our estimate implies that abortion legalisation prevented approximately 4,000 births in Mexico City in 2010.
  1. Phillip Levine et al., ‘Roe v Wade and American Fertility’ (1999) 89(2) American Journal of Public Health 199: States legalising abortion experienced a 4% decline in fertility relative to states where the legal status of abortion was unchanged. The relative reductions in births to teens, women more than 35 years of age, non-White women, and unmarried women were considerably larger. If women did not travel between states to obtain an abortion, the estimated impact of abortion legalisation on birth rates would be about 11%. A complete recriminalisation of abortion nationwide could result in 440,000 additional births per year.
  1. Marianne Bitler and Madeline Zavodny, ‘Did abortion legalisation reduce the number of unwanted children? Evidence from adoptions’ (2002) 34(1) Perspectives on Sexual and Reproductive Health 25: (This journal is published by the Guttmacher Institute, so if anyone accuses you of ignoring pro-choice evidence, throw this at them. Also the fact that they use the phrase ‘unwanted children’ in a serious manner.) The estimated effect of abortion legalisation on adoption rates is sizable and can account for much of the decline in adoptions, particularly of children born to white women, during the early 1970s. These findings support previous studies’ conclusions that abortion legalisation led to a reduction in the number of ‘unwanted’ children; such a reduction may have improved average infant health and children’s living conditions.
  1. Michael New, ‘Analysing the Effect of Anti-Abortion US State Legislation in the Post-Casey Era’ (2011) (11) (1) State Politics and Policy Quarterly 28: In this study, a series of regressions on a comprehensive time series cross-sectional data set provides evidence that several types of state-level anti-abortion legislation result in statistically significant declines in both the abortion rate and the abortion ratio. Furthermore, a series of natural experiments provide further evidence that abortion restrictions are correlated with reductions in the incidence of abortion.
  1. John Donohue, Jeffrey Grogger and Steven Levitt, ‘The Impact of Legalised Abortion on Teen Childbearing’ (2009) 11(1) American Law and Economics Review 24: The legalization of abortion in the 1970s changed the composition of women at risk of bearing children out of wedlock some 15–24 years later. This composition effect reduced out-of-wedlock teen birth rates by 6%, which accounts for roughly 25% of the observed decline in unmarried teen birth rates over this period.
  1. Christine Valente, ‘Access to abortion, investments in neonatal health, and sex-selection: Evidence from Nepal’ (2014) 107 Journal of Development Economics 225: In Nepal, how close a woman lived to an abortion centre determined how likely she was to give birth.
  1. Caitlin Knowles Myers, ‘Power of the Pill or Power of Abortion? Reexamining the effects of young women’s access to reproductive control’ IZA DP No 6661, June 2012, 2017 version here: Policy environments in which abortion was legal and readily accessible by young women are estimated to have caused a 34 percent reduction in first births. This is not explainable by increased access to contraception.
  1. Stefanie Fischer, Heather Royer, Corey White, ‘The Impacts of Reduced Access to Abortion and Family Planning Services: Evidence from Texas’ National Bureau of Economic Research Working Paper No 23634, July 2017: Between 2011 and 2014, Texas enacted three pieces of legislation that significantly reduced funding for family planning services and increased restrictions on abortion clinic operations. In-state abortions fell 20% and births rose 3% in counties that no longer had an abortion provider within 50 miles. Births increased 1% and contraceptive purchases rose 8% in counties without a publicly-funded family planning clinic within 25 miles. This amounts to more than 3,000 additional births, including 2,562 caused by abortion clinic restrictions and 668 linked to lack of funding for non-abortion reproductive resources like centers that distribute birth control pills and condoms.
  1. Phillip Levine and Douglas Staiger, ‘Abortion Policy and Fertility Outcomes: The Eastern European Experience’ (2004) 47(4) Journal of Law and Economics 223: We distinguish among countries with severe, moderate, and few restrictions on abortion access and examine the impact of changes across all three categories. As we hypothesize, the results indicate that countries that changed from very restrictive to liberal abortion laws experienced a large reduction in births. Changes from modest restrictions to abortion available on request, however, led to no such change in births despite large increases in abortions, which indicates that pregnancies rose in response to more liberal abortion availability. This evidence is generally consistent with the situation that was brought about by changes in abortion access in the United States.
  1. Bussarawan Teerawichitchainan and Sajeda Amin, ‘The role of abortion in the last stage of fertility decline in Vietnam’ (2010) 36(2) International Perspectives on Sexual and Reproductive Health 80: (Guttmacher again. One of the more horrible studies I have seen. Really hate the tone of pushing for higher abortion rates for minority groups in Vietnam’s central highlands). Ethnic groups in Vietnam such as the Kinh, Chinese, Tay, Thai, Muong and Nung have low fertility rates and higher abortion rates. [Study cannot confirm link between contraceptive use and abortion. However then confirms that lower use of contraception = higher abortion rate = lower fertility? Check]
  1. Libertad González, ‘The Effect of Abortion Legalization on Fertility, Marriage and Long-term Outcomes for Women’ (June 2016, WP, Brunel): Using birth records and survey data, we find robust evidence that the legalisation of abortion in Spain in 1985 led to an immediate decrease in the number of births to women aged 21 or younger.
  1. Extract from Anna Glasier, ‘Contraception’ in J Jameson and Leslie de Groot (eds) Endocrinology (Saunders 2010) 2417: “Despite high prevalence of use of contraception, unintended pregnancy is common and so is induced abortion. While much of the decline in population growth has been achieved by contraceptive use, no country has achieved low fertility rates without access to abortion.”

As always, pro-choice responses welcome. We will publish responses on this blog to encourage discussion on the issue.

The Statistics Fairy