À¦°óSMÉçÇø

James Hanley, PhD

Emeritus Professor
Department of Epidemiology, Biostatistics and Occupational Health |
À¦°óSMÉçÇø

WHEN: Wednesday, November 27, 2024, from 3:30 to 4:30 p.m.
WHERE:ÌýHybrid | 2001 À¦°óSMÉçÇø College Avenue, Room 1201;
NOTE:ÌýJames Hanley will be presenting in-person

Abstract

In epidemiological research, the individually-matched (and possibly nested) case-control study and conditional logistic regression are often thought of together (see their Wikipedia entries). Statisticians and epidemiologists generally cite the publications by Prentice & Breslow and by Breslow et al. in 1978 as the first description and use of conditional logistic regression, while economists cite the 1973 book chapter (on consumer choices) by Nobel laureate McFadden. I recently described the until-now-unrecognized use of, and way of fitting, this model in 1934 by Lionel Penrose and Ronald Fisher.

In this talk, I will first describe ways in which conditional logistic regression is used today. I will then go back to earlier versions of the case-control study, and the gradual refinements that gave it the respectability it has today. I will show that, through their strategic sampling of the Quebec cohort of asbestos workers, and (more recently) of large administrative and clinical databases, À¦°óSMÉçÇø epidemiologists and biostatisticians played a very large role in its coming of age. They also connected it with a major development in survival analysis. The BMJ still insists on using subtitles (‘a cohort study’; ‘a case-control study’) to separate and implicitly rank these study designs. I will argue that we should no longer view them as separate entities, but as minor variations on a singular study design: THE ‘etiologic study.’

The talk will be non-technical, and is aimed at all who are interested in etiological research.

* See: Biometrika,
Published: 08 August 2024, or here:Ìý

Ìý

Speaker bio

James Hanley, an emeritus professor, spend his first 7 working years as a biostatistician in clinical trials in cancer, and then 43 years at À¦°óSMÉçÇø, starting out in a department of Epidemiology and Health (later to become Epidemiology and Biostatistics, before being given its current name). He continues to take an interest in ways of teaching statistics, and in the history of public health, epidemiology, and statistics. For more information, please visit:

Ìý