Importing Econometric Models into Psychiatric Epidemiology, A Start

Leaving aside earlier history of probability and statistics that anticipated least squares regression, a potted history in this area can start with William Farr and his 'hero of concept,' the Marquis de Condorcet (Nicholas de Condorcet).

Sidebar 1: Let me ask one of your to make a comment about Condorcet for me to edit and add as a later sidebar, with attribution. Why was Condorcet important to Farr? How might Farr have learned about him? You can answer these questions by reading my copy of Humphreys' version of Farr's selected works, but it might be easier to search for Condorcet (and France) in the online version of Farr's work. In the process, you will learn that Farr's studies in France were in part financed by an elderly benefactor, and that Farr was accompanied by a wealthier physician-friend to study with Pierre Charles-Alexandre Louis. Many characterize Louis as the physician who formalized clinical trial elements such as randomized assignments.

Sidebar 2: While his friend was hiking the Alps, what was Farr doing, which is pertinent to his abiding interests in what we now call neuropsychiatric epidemiology of developmental disabilities?

Sidebar 3: Find and describe Farr's post-Snow description of factors x,y,z, etc., mentally held constant while the source of water supply was allowed to vary, an early conceptualization of what was to become the familiar multiple regression model for a univariate response. (Hint: You can find a rendition in one of my papers.)

Fast-forward, and skip over a lot of progress until you encounter:

(1) Sewell Wright, son of P. Wright, and discover their contributions to this topic, with particular focus on path analysis, and its contribution to a field called "Population Genetics," which now is seen no more than rarely as an acedimic department in universities. Can you figure out why and how it came and went?

(2) Contributions made to model-building in econometrics, which should be understood by epidemiologists. Why not read January Tinbergen's Nobel Prize lecture, which is about as good an explanation of models and model-building as I have seen lately:

January Tinbergen Nobel Prize lecture, 1969

P.s. Next up, error terms, Keynes, and Haavelmo, as we work our way toward comparison of Directed Acyclic Graphs and their sequences of univariate response models with simultaneous equations models for a multivariate response world.