fbpx

Damned Lies versus Statistics

Statisticians are familiar with the claims that one can lie with statistics and that there are lies, damned lies, and statistics. There is also the Stalinist aphorism (actually Kurt Tucholsky’s) that the death of one man is a catastrophe, while the death of hundreds of thousands is a statistic. These are all bad enough, but recently in Law and Liberty, Richard Gunderman argued that the application of statistical models to human life is dehumanizing, and that the discipline of statistics is tarnished by its founders’ association with racism and eugenics. Gunderman attempts to show this with a biographical account of three key figures in the history of statistics, Francis Galton (1822 – 1911), Karl Pearson (1857 – 1936), and Ronald Fisher (1890 – 1962).

The Cum Ergo Propter Hoc Fallacy

 Gunderman falls into the questionable-cause or cum ergo propter hoc fallacy, which is when a person sees a correlation between two things, and wrongly assumes that one has caused the other. In this case, Gunderman has assumed a causal relationship between “being a statistician” and “being a eugenist and racist,” and the logical error lies in believing that the former causes the latter. Ironically, Pearson himself is the one that first explicated this fallacy within the context of Galton’s work on correlation.

Gunderman does give nice intellectual biographical sketches of this trio of statisticians, although he incorrectly gives Fisher credit for the t-distribution; this was actually Guinness Brewery’s chemist William Gosset’s contribution. Gosset was on good terms with both Pearson and Fisher, who were bitter enemies, and his work on the t-distribution was validated by Pearson, not Fisher though the latter understood its importance while the former did not. As far as I have been able to determine, Gosset himself held no abhorrent views on race and population control.     

Regrettably, such views of Anglo-Saxon supremacy were common among not only English statisticians, but the British intelligentsia of the time. John Maynard Keynes, Rudyard Kipling, George Bernard Shaw, and Virginia Woolf held similar views. The fact that three English statisticians were eugenic racists in the late 1800s and early to mid-1900s should surprise us as little as the discovery that three representative German professors expressed antisemitic opinions in the same decades. Cum ergo propter hoc, indeed. 

The Falsum Dilemma Fallacy

Gunderman does admit that while unfortunately Galton and Pearson were both racists and eugenicists, Fisher’s “views on race were somewhat nuanced.” I am speculating, but perhaps this is because Fisher, unlike the atheists Galton and Pearson, was an Anglican who on some level understood that his beliefs related to race and eugenics did not align with his Christian faith. For all of the criticism the Catholic Church receives regarding its prohibition of birth control, it is certainly an anti-racist view: all human life has equal worth. In the end, social Darwinism is necessarily eugenic racism, as non-statistician Margaret Sanger epitomizes. Gunderman also deserves substantial credit in that he does not seek to cancel these thinkers’ contribution to statistics, mathematics, and science, despite their abhorrent beliefs.

Gunderman concludes his essay by stating that Sophocles, Shakespeare, and Tolstoy offer unsurpassed insights into humanity without using statistical analysis. This is another logical fallacy: falsum dilemma or false dilemma. This is when only two options are offered, although there are additional possibilities. The same fallacy appears in the statement, “Either you are in favor of gun control or you support gun violence.” One can reasonably argue that one of the purposes of the Second Amendment to the United States Constitution is to ensure that gun control does not lead to gun violence and governmental tyranny. Likewise, it is a falsum dilemma to set literature and statistics in opposition. Human beings have explored the truth along many avenues, not just these two. 

Gunderman sets up a straw man when he suggests that statisticians view their field as the only or best way to understand humanity.

The false dilemma paves the way to further confusion. Gunderman’s illustration is anachronistic since two of these writers wrote before statistics existed as its own subject of inquiry and the third was contemporaneous with the advent of statistics. But there is a deeper issue here: The purpose of literature is not the same as that of statistics. If one wants insight into the depth of the human soul, then yes, please read literature. But if one wants to draw conclusions from scientific data, then the only recourse is statistics. There are myriad applications of statistics where it has proved its worth as a discipline, to mention but one: showing the inhumanity of COVID-19 lockdowns.

The Ignoratio Elenchi Fallacy     

In his closing sentence, Gunderman writes, “Properly applied, statistics can enlighten us, but to regard statistics as the best or only window on human reality is to engage in an essentially dehumanizing project with moral and political consequences that can prove nothing short of disastrous.” After 1,500 or so words, I was surprised that Gunderman admitted that statistics can have a useful purpose, but this admission was couched in another informal fallacy. The straw man or ignoratio elenchi fallacy is when one characterizes his opponent’s position in an over-simplified way, making it easy to knock down. Gunderman sets up a straw man when he suggests that statisticians view their field as the only or best way to understand humanity. He makes no convincing argument that this was true even for Galton, Pearson, and Fisher, let alone for statisticians writ large.

With Gunderman, I concur that we should critique the misuse of statistics. There are plenty of legitimate targets within the science, such as using bad data, or misapplications of statistical inference. However, the unconscionable beliefs of Galton, Pearson, and Fisher are not among them.