Are statistics racist?

Question

Evidence-Based Medicine (EBM) is a proposition and an area of study of medicine for which I am very fond. However, a few days ago, talking to some friends, I was confronted with a very critical approach to MBE and to Biostatistics itself.

"In 1992, the journal of the US Academy of Medicine announced the formulation of “a new approach to teaching the practice of medicine”. Known as Evidence-Based Medicine (EBM), it “emphasizes the examination of evidence provided by clinical research”, for which it “requires the application of formal rules to assess the evidence” (Guyatt et al., 1992).

A few years later, the scope of EBM had expanded considerably: “it is the meticulous, explicit, and sensible use of the best available evidence in making decisions regarding the care of the individual patient” (Sackett et al., 1996). For this purpose, the level of evidence is classified into categories, according to multiple criteria, among which, the statistical-mathematical measures to establish the chance that a certain effect is due or not to chance have a prominent place. Thus, concepts such as “frequency distributions”, “standard deviation”, “statistical significance” (usually measured as “p-value”), “correlation”, “regression” and “randomization” and tests such as “chi square”, “analysis of variance” and “Fisher's exact test” have become everyday tools, not only in clinical research, but in all areas of the biological and biomedical sciences.

The examples cited above are just a few of the many statistical concepts and methods used in EBM. However, they were not selected arbitrarily, but all of them were developed, among many others, between the end of the 19th century and the beginning of the 20th century by Karl Pearson (1857-1936) and Ronald A. Fisher (1890-1962). Both are considered the formulators of modern statistics, who developed according to a well-defined purpose: to improve the hereditary endowment of the Human race."

All fragments were taken from this book.

Available for free at https://www.fm.usp.br/museu/portal/livros-para-download (portuguese)

The tools that we use on a day to day basis to interrogate data and understand the world, were developed by white supremacists for the express purpose of demonstrating that white men are better than other people. In Statistics for Social Justice, we use statistical tools to expose and analyze racism in many aspects of society. Is this a contradiction?

Answer 1

In general it is important to check scientific results and methods for racism and other biases, so it's a fair question.

One needs to distinguish, however, between the historical origins and the present-day usage.

The statistical methods in EBM are fundamentally just tools. While they were developed by people with racist views and with the purpose of supporting racist views (as the quote describes), the tools themselves are not (necessarily) racist.

To give an analogy, many other fields of science also have historical roots in things that are (or should be) unacceptable now. For example, some mathematics and mechanics was developed for war, to be able to calculate the path of cannon balls or such things. Chemistry was used to develop bombs. Nuclear physics was, for a time, closely tied to the development of the nuclear bomb. Medical surgery was also developed mainly by military surgeons. That doesn't mean that our understanding of mechanics, of chemistry, of the properties of atoms or or medical surgery is now militaristic.

With statistical methods, it's similar. They are fundamentally neutral tool, despite of what they have been used for or are used for today.

However, this doesn't mean that the applications are always harmless or unbiassed. With statistical methods and EBM, it is possible that other biasses and also racism come into the actual results. For example, if a new drug is only tested on a white European or US population (because the manufacturer is based there), it may not work on an African or Asian population or have very different side effects (this can be due to genetic or lifestyle differences). Ideally, drug trials should be performed with samples from all populations, but of course the manufacturers may focus on some markets where there is most money, and poorer countries may not have the option of requiring manufacturers to do test in their own population.

So you can have a situation where all the EBM literature says some intervention works, but practitioners say "anecdotally in our experience, it doesn't" - and the practitioners are actually right if they work with in a different population than the one on which the EBM studies were done.

Therefore, there can indeed by implicit racism in EBM as well and one should always be aware of these issues, but it's a separate issue and has nothing to do with racism in the historical development of the mathematical methods.

Answer 2

Are statistics racist?

No!

Statistics are a mathematical/analytical tool that can be applied to any field, such as: zebra populations, industrial processes and products, interstellar research, shoe sizes as well as medical "research".

Some people may have and may want to use statistics for racist purposes. That has nothing to do with the fundamentals of statistics. It would be an aberrant application of statistics.

Lies, damned lies, and statistics.