Fisher [actually Cox] and logistic regression

(it might be useful to have a ‘history’ category?)

I think I heard @f2harrell say Fisher described logistic regression, ie it can be attributed to him. Do I remember this right? Is there a reference?

I added a history tag.

I never said that about Fisher and I’ve never seen that in print. DR Cox developed multivariable logistic regression: The Regression Analysis of Binary Sequences on JSTOR (1958). Single-predictor logistic models for bioassay had been developed decades earlier. I don’t have the primary reference for that.

3 Likes

Indeed, we provide various historical references in our obituary to David Cox, including the development of the logistic regression approach as an alternative to probit by Berkson in 1944 for use in bioassays.

6 Likes

I had no idea @f2harrell was responsible for the first logistic regression and proportional hazards routines in SAS! And here we here, 40+ years later and Frank is still producing excellent work in Hmisc, rms, rmsb, reproducible research, and more!

5 Likes

Agree. It is hard to believe that anyone could show such consistent long-term commitment to making data analysis methodology accessible to the community. That’s why we really felt compelled to highlight it.

1 Like

I really appreciate that Pavlos. In truth I needed the SAS procedures and much of what else I’ve written for my own work in cardiovascular clinical epidemiology at Duke University. The pure gift to the community is the documentation :slight_smile:

6 Likes