Application of Epidemiological/Biostatistical Methods to Political/Economic Issues

I was recently asked by friends who worked in the field of public health whether typical epidemiological/biostatistical methods had been much applied to political and economic issues. The answer is a definite YES. The question is not whether socioeconomic issues have been considered in public health studies, since there is a long history of social epidemiology, or social medicine, going back at least to Rudolf Virchow in Germany in the mid-19th century. Geoffrey Rose and Michael Marmot in Great Britain and George Kaplan and John Lynch in the U.S. (the latter also active in Australia) are just a few well known more recent practitioners. Indeed, Virchow famously stated that “[m]edicine is a social science, and politics is nothing else but medicine on a large scale.” Poverty, social status, and educational attainment have been well-studied as associated with prevalence and incidence of disease. A newer socio-economic variable of interest is degree of economic inequality as a predictor of well-being, most famously examined by epidemiologists Richard Wilkinson and Kate Pickett in their book The Spirit Level. This association is by definition an analysis of grouped variables, i.e., an ecological analysis in the statistical sense of that word. The groups are societies, the exposure of interest is income or wealth inequality, and the outcome variable is health status or other measure of well-being or lack thereof, all variously defined. We summarize some of the work from The Spirit Level in our We Can Have a Better Country, in Chapter 7 (“Inequality, Poverty, and Corporate Control in the Public Health Arena”).

Two recent papers by economists further explore inequality and its potential effects on mortality. A 2016 paper from the Brookings Institute, “Later Retirement, Inequality in Old Age, and the Growing Gap in Longevity Between Rich and Poor”, by Barry Bosworth, Gary Burtless, and Kan Zhang explore factors affecting probability of retirement and mortality rates, though using probit analysis and linear regression where epidemiologist/biostatisticians would more likely have used logistic and Poisson regression. They also examine the association of life expectancy at age 50 with income level, for those reaching age 50 in 1970 or in 1990. Those in the top tenth of income in the U.S. have a higher life expectancy than those in the lowest tenth, as would be expected, but the new information is that the gap in life expectancy between top and bottom incomes has grown enormously, as the top income group has grown vastly richer and the bottom income group poorer. For women the gap went from under five years in 1970 to over ten years in 1990. For men, the gap went from two to eight years over the period. Another recent finding is that increasing inequality may be leading to reduced life expectancy in working class whites – see “Rising morbidity and mortality in midlife among white non-Hispanic Americans in the 21st century”, by Anne Case and Angus Deaton, in the Proceedings of the National Academy of Science.

Ernest Drucker explicitly applies methods of epidemiology to look at the issue of mass incarceration in the U.S., in A Plague of Prisons: the Epidemiology of Mass Incarceration in America. He looks at factors associated with the increase of incarceration, the chronic incapacitation that often follows incarceration, and a public health approach to end mass incarceration.

We explored in We Can Have a Better Country, in Chapter 13, “New Directions in Political and Economic Thought,” two interesting analyses that are epidemiological in approach but are on political issues. The authors of these two books are both political scientists. Martin Gilens, professor of politics at Princeton University, in his book Affluence and Influence: Economic Inequality and Political Power in America, brings extensive data to the question of the how responsive our government is to its citizens, and in particular to the question of how this responsiveness varies by level of income of the citizens. His outcome variable of interest is yes/no, according to whether the government enacts policy changes on the various issues considered, and his exposure variable is the level of citizen support for these issues. We would expect that the government would have a higher likelihood of adopting a policy change when that change has a higher level of support among the citizens. The interesting twist is that Gilens does this analysis by income level of the people surveyed on opinions regarding these policy questions. (In epidemiological terms, this is a consideration of “effect modification”.) The analysis method is logistic regression. His conclusion: among the affluent respondents there was a strong association between preference levels and later policy changes. For the respondents with low income there was no association at all.

The second epidemiology-like analysis described in Chapter 13 of We Can Have a Better Country is by Benjamin Radcliff, Professor of Political Science at the University of Notre Dame, from his book The Political Economy of Human Happiness: How Voters’ Choices Determine the Quality of Life. He analyzes with linear regression the association between human happiness and the role of government, variously defined. His conclusion, however “role of government” was defined, was that countries with the more activist governments had a higher mean level of happiness.