These famous words, which Mark Twain attributed (probably erroneously) to British Prime Minister Benjamin Disraeli, have served as a warning that statistics, if used carelessly, can be as misleading as outright lies.
In the mid-1990s I was asked by the United Nations organization UNIFEM to examine cross-cultural statistics on women’s participation in science, technology and medicine (STM). The goal was the compilation of tables for the UN fact book The World’s Women: 1995 .
This being well before the days of easy internet data searches, UNIFEM sent me cartons and cartons of policy papers, data sheets, and women’s organization pamphlets. I poured over the materials for months, trying to contrive tables that would fairly put the status of women in STM in each country in comparative perspective.
I was forced to conclude that the task was impossible. Some data sets only included the percentage of women in STM in tertiary institutions, while others only included women in government research institutes; some collections lumped together women of different ranks or in different scientific specialties, while others meticulously differentiated ranks and subfields; some included even the social sciences and humanities in their definitions of “science,” while others included the physical sciences but not mathematics or engineering. Ultimately, I wrote a piece (the only signed one in the volume) explaining the absurdity of trying to derive meaningful comparisons from such sources.
We find a different sort of methodological morass if we examine the UN’s comparative Gender Inequality Index (GII) (Human Development Report,
United Nations Development Program, 2016). The GII is calculated from three components, one of which is purportedly a measure of reproductive health; it is that component that I’ll discuss here.
The reproductive health component of the GII is computed by combining two indicators: maternal mortality per 10,000 live births and teen pregnancy rates per 1000 women. Once could ask: What about other indicators that in some countries are much more significant than maternal mortality and teen pregnancy rates in assessing the situation? At the very least, the data should include availability of contraception and legality/accessibility of abortion. In the United States one should also track the number of abortion doctors harassed or murdered, the number of clinics and schools intimidated out of offering birth control and abortion information, the number of states with outrageously restrictive TRAP laws, the number of raucous demonstrations at abortion clinics, the number of “Crisis Pregnancy Centers” staffed by anti-abortion zealots masquerading as neutral reproductive health advisers, the decreasing availability of abortion and certain forms of contraception, and so on.
The likely explanation for the UNDP’s use of maternal mortality and teen pregnancy and nothing else is that those are the data that are readily available from most countries of the world with some degree of accuracy (although the ratio of reported teen pregnancy to actual teen pregnancy undoubtedly varies widely from country to country). This is the typical reason why “proxy data” are used: you use the data that’s easiest to get, not the data that would really be most revealing, and you hope that the proxy data at least correlate with what you’re trying to measure.
In the case of the UNDP’s proxy data for reproductive health, what they do correlate with is the overall wealth and level of development of a country. More than anything else, low maternal mortality means a relatively affluent population and a highly developed health care system.
Similarly, the use of teen pregnancy rates skews the results in favor of the wealthier countries, where in fact it’s reasonable to expect that a low teen pregnancy rate signifies widespread availability of birth control as well as adequate sex education in the schools. But note that `teen’ is defined to include 18- and 19-year-olds. In many low-income countries where most men and women do not go to college and a high proportion of the population is rural, early marriage and child-bearing are culturally accepted and not correlated with failure in life. (This was the case in the U.S. until well into the 20th century.) Why then should the GII penalize low-income countries for a high teen pregnancy rate?
Because of what is counted, and equally importantly, what is not counted, there are some curious oddities in the GII rankings. Take South Korea, for example, which ranks tenth in the world in the GII. Because South Korea has a reported teen pregnancy rate of 1.6/1000 (the lowest in the world) and a maternal mortality rate of 1.1/10,000, the reproductive health component of its GII appears close to ideal, and that leads to its high GII ranking. But abortion is almost completely banned in South Korea. Doctors have been prosecuted for performing the procedure, the illegal abortion industry is booming, and there is a large black market in abortion pills obtained illegally over the internet. Women activists have been trying to get the draconian abortion laws revised for years, and have only recently succeeded in convincing the legislature to revisit the issue.
To cite another example, Cuba has a GII ranking of 62 out of 159 countries, while the U.S. ranking is 43. Why does Cuba come out poorly compared to the U.S.? By far the main reason, ironically, is Cuba’s reproductive health component, which is greatly affected by its relatively high teen pregnancy rate of 45.6 per 1000 vs. 22.6 for the U.S. Meanwhile, Cuba has a health care system that is a model for developing countries, and it is the only country of Latin America that has had unconditional reproductive freedom for women for over half a century. (In addition, women constitute 48.9% of the Cuban legislature and 27% of the Cuban Academy of Sciences; the latter figure is the highest percentage in the world.)
My point is not to castigate the UNDP for using statistical methodology that favors the wealthy countries and underrates low-income countries such as Cuba that have progressive traditions in women’s health and women’s rights. Rather, the lesson here is that issues of reproductive health and gender equity are far too complicated to be captured by a single number, especially one that’s computed from proxy data.