Stata The Stata listserver
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

Re: st: RE: Econometrics Theory Questions on Dummies and Correlation Analysis


From   Chris Ruebeck <ruebeckc@lafayette.edu>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: RE: Econometrics Theory Questions on Dummies and Correlation Analysis
Date   Mon, 18 Apr 2005 22:43:25 -0400

To take Nick's example further, I think that Sam's point in a binary outcome was that finding a mean equal to one of the two outcomes would tell us the same thing as finding the variance equal to zero---and the mean, in this case, would actually tell us more than the variance did! On a more general basis, there is only one degree of freedom (once we fix the number of observations) in a binary outcome variable, so reporting both the mean and the variance is redundant. We choose the mean because it is easily interpreted (especially when the binary outcomes are 0 and 1) as the fraction of one outcome.

I recognize, though that this is a small eddy in the greater river of discussion to which Nick refers.

Chris

On Apr 18, 2005, at 5:39 PM, Nick Cox wrote:


I agree that means and variances of non-binary nominal codes don't usually make sense. The only exception I can think of is that zero variance has a clearcut interpretation, namely that all observations are in the same category. But you clearly don't need the variance to tell you that, and I don't think this is a point at issue in this thread.
*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/



© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index