Wading into Akaike again…

In Uncategorized on May 5, 2010 by Randy

Apologies to all about my stridency during the seminar. I regret that much of our normal discourse was stifled by the red herring that I threw into the room.

Having said that…

Let me tell you two short stories in which Akaike figures. A few years ago, I sat in on a highly rigorous doctoral seminar on psychometrics taught by one of the giants in the field, Roderick P. McDonald. Causal modeling (does this sound scary to a philosopher?) for analysis of complex social science data was the overt subject matter. McDonald described his objective as to get us to the point where we “were not a danger to ourselves nor an an annoyance to others”. He took us through BIC and AIC, as well as 16 other criteria for assessing model fit given parsimony, included and omitted variables, number of cases (“n”), and the characteristics of the raw data. Long story short, AIC and BIC are unstable with respect to parsimony (number of parameters between competing models) and “n”. He called these two criteria “useless nonsense”, as were all the others for various faults with respect to the core criterion for model fit: the noncentrality parameter.

He also told a story of a meeting in 1972 where Akaike shared the dais with Joreskog (developer of LISREL – the standard statistical package for doing the kinds of analyses I alluded to above) and a doctoral student in education from the University of Chicago, J.W. Keesling. Keesling presented his thesis paper on causal flow modeling. His design looks  like Joreskog’s LISREL model a few years later. Sometime in the 1990’s practitioners of causal modeling began calling the design the Keesling-Wiley-Joreskog model (including Keesling’s mentor, David Wiley). I suspect Rod McDonald was more than just a casual observer of this more complete attribution of the analytical design. And Akaike’s information criterion, discussed at the same symposium, became a default goodness-of-fit measure for LISREL and its knock-offs. Big-time meeting — ex post.

Grad students: guard your intellectual property! And be skeptical of received wisdom of from philosophers who discuss parsimony in the ether without looking in the dirt where the concept is a core feature of practice.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: