Thread by @BryanDMartin_

Hey you, person that works with data! Have you ever implemented or used "regression"? Have you, like me, used regression for years without questioning why it's called regression, and not something that is actually descriptive?

A 🧵 on the linguistic history of "regression":

1/9

Turns out, it's not just because statisticians are terrible at naming everything (we are!). Also turns out, it has origins the eugenic history of statistics.

Before you roll your eyes and dismiss this as another statistician SJW trying to cancel science, let's talk about Galton.

Francis Galton, coiner of the term eugenics in 1883, sought to understand to what degree parents passed on desirable traits in order to promote "improvement of the human breed" [1].

What sort of traits was he focused on? The things he considered desirable, such as "civic worth"

Galton found what was, to him, a troubling conclusion [2]: offspring of parents with extreme values tend to have less extreme traits than their parent.

He called this: "regression towards mediocrity" [3].

Sound familiar? It's now referred to as "regression to the mean"

4/9

"regression towards mediocrity" was Galton's judgement that "quality of parentages" go "far back towards mediocrity".

If you've ever taken a course in statistics, you know that Galton's term, "regression" has stuck.

5/9

I'm not naive enough to think regression will be re-named any time soon. But I do think it's worth questioning why we use the terms that we do, or at the very least understanding their origins, especially when our field has a history that is well-known to be rooted in eugenics.

After all, what exactly is being "regressed" in regression, anyways?? How about something actually descriptive for a change, like "curve fitting"?

(really, what is it about statistics that requires us to use such terrible names for everything?)

7/9

Now, am I suggesting you stop using all regression in light of this? Certainly not. Much of my own research centers around regression. Do I think it's important to understand the history of the field and how that history manifests itself in the present day? Absolutely.

8/9

Sources:
[1] bit.ly/3j2yDlh
[2] bit.ly/31bGUgE
[3] bit.ly/3lO1eMY

HUGE h/t to @paulinetrinh for sharing the fantastic article that taught me about this, and to @rvrohlfs for writing that article! (bit.ly/342miJG)

/end 🧵