Re: Re: st: MIXLOGIT: marginal effects

[Clive Nicholas <clivelists@googlemail.com>]

Christopher 'Kit' Baum replied:

> I don't understand it either, and I agree wholeheartedly with the sentiment. The undergrad textbook from which I teach Econometrics,
> Jeff Wooldridge's excellent book, has a section on the LPM; I skip it and tell students to stay away from it. Unfortunately, much of the
> buzz about the usefulness of the LPM has arisen from the otherwise-excellent book by Angrist and Pischke, Mostly Harmless
> Econometrics, in which they make strong arguments for the use of the LPM as an alternative to logistic regression.

Having recently finished reading "Mostly Harmless Econometrics", I was
left nursing many beefs about the book, some of them possibly quite
controversial. For one thing, they write throughout of schools,
classrooms and pupils, and yet not once do they even give a nod to
multilevel models, the regression modelling technique surely most
suited to modelling outcomes in such situations. It's only been around
25-odd years: perhaps it flashed past them. I _would_ say that I
thought that was a quite astonishing omission to make, but it isn't
really: after all, it wasn't developed by econometricians, was it? I
have other beefs about their book (why is there no discussion of the
merits of OLS models with panel-corrected standard errors; why did
Angrist and Pischke (2009) pointedly omit to mention that lagged
variables often play role as suitable instruments in IV models, etc.),
but I'll leave it there on the beefs.

On the positive side, I should also say that, having started to read
the quite superb paper by John Antonakis and colleagues (2010; as well
as Antonakis' full-version podcast - available on YouTube - discussing
the subject: well worth a watch), their case for why IV-2SLS is
necessary if you wish to deal with the problem of endogenity in your
model specification fully vindicates Angrist and Pischke's choice to
put the method front and centre in the book, even if I found most of
the material contained in that fourth chapter largely inaccessible.

>
> One of my econometrician colleagues has come up with a nifty example of how, in a very simple context involving a LPM with
> a binary treatment indicator, the LPM gets the sign wrong! A logistic regression, even though it fails to deal with any further issues
> regarding the treatment variable, gets the right sign.

I have to concede that I didn't know that obtaining incorrect signs on
the coefficients were yet another bad feature of such models, but it's
well worth making a mental note of. Interestingly, Pampel doesn't
mention it in his primer.

-- 
Clive Nicholas

[Please DO NOT mail me personally here, but at
<clivenicholas@hotmail.com>. Please respond to contributions I make in
a list thread here. Thanks!]

Angrist JD and Pischke J-S (2009) "Mostly Harmless Ecomonetrics: An
Empiricist's Companion", Princeton, NJ: Princeton University Press.

Antonakis J, Bendahan S, Jacquart P and Lalive R (2010) "On Making
Causal Claims: A Review and Recommendations", The Leadership Quarterly
21(6): 1086-1120

Pampel FC (2000) Logistic Regression: A Primer (Sage University Papers
Series on QASS, 07-132), Thousand Oaks, CA: Sage.
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/