The new cover article in @NautilusMag is about the sordid history of eugenics in the field of statistics.

I had a conversation with @aubreyclayton a few months ago, when he was beginning his work on this article.

1/ https://twitter.com/aubreyclayton/status/1321631931038081026

I think this article has a lot of value, as it brings to sharp focus the very racist views of some of the founders of our field: Galton, Pearson, Fisher, in particular. I applaud Aubrey for his very careful research about this topic.

Some screenshots from the article:

2/

As I told Aubrey, in the year of our lord 2020, we don't need awards or monuments to honor people with racist views. We can make use of their contributions to statistics without holding them up as exemplary humans (which they’re not!!)!

3/

Statisticians (and other quantitative thinkers) often feel that because they work with numbers, they are inherently objective, and couldn’t possibly be biased or racist in any aspect of their lives. But, this logic is flawed.

4/

You can understand how to objectively test a null hypothesis while still holding subjective (and, dare I say, even incorrect) views in other areas of your life. Aubrey provides a nice example of the fallacy of believing oneself to be impartial here:

5/

Aubrey also makes some important points about the importance of recognizing our biases: as he puts it, “statistics is a human enterprise”.

I agree!! We need statistics to be an inclusive field. Let's keep working to make it happen.

6/

However, much of the second half of Aubrey’s article amounts to an argument that the mathematical/statistical framework underlying the past 100 years of statistics is somehow invalid. I do not agree!!!

7/

Statisticians have recently issued statements about hypothesis testing: that we should consider a totality of evidence, including effect sizes, confidence intervals, and so on, instead of using a “bright line cut-off” like p<0.05. All true! https://www.tandfonline.com/doi/full/10.1080/00031305.2019.1583913

8/

While these statements DO offer good common-sense advice for how scientists should draw statistical conclusions from their data, they DO NOT say to throw away the baby with the bathwater, or that the field of statistics is inherently flawed.

9/

Many non-statisticians seem to misunderstand this point (and make claims that e.g. "the ASA is against p-values”), despite statisticians’ best efforts to make their views clear.

10/

So: a nice article by @aubreyclayton that brings to light the racist history of the founders of statistics, but that unfortunately implies that the underlying framework of statistics is at fault (as opposed to the statisticians). Statistics is objective; people are not!

11/

Fisher information is a valuable statistical concept, even if Fisher himself was a piece of *&@#.

And there is nothing wrong with the statistical or logical framework underlying Fisher's exact test --- the problem is if you apply it to promote your personal agenda.

12/

P.S. I strongly disagree with the claim that “Fisher . . . is the only other person with a legitimate claim as the most influential statistician of the 20th century.” ....

13/

.... I can think of plenty of statisticians — including some still living, and some who I have the pleasure to know personally — who have a claim to this. Aubrey’s claim makes more sense if it is limited to the first half of the 20th century.

14/

P.P.S. I’ve been an associate editor for Biometrika — one of the top journals in our field — for almost 10 yrs. It's a very prestigious theoretical & methodological stats journal (less theoretical than Annals of Stats; maybe slightly more so than JASA Theory & Methods).

15/

As a community, we should know this journal’s origin story . . . but I personally don’t think we need to cancel it in 2020. However, if it were named after Pearson/Galton, or had the word “eugenics” in its name, I'd want it to be renamed.

16/fin