Statalist


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: Heteroskedastic probit


From   Richard Williams <Richard.A.Williams.5@ND.edu>
To   "statalist@hsphsun2.harvard.edu" <statalist@hsphsun2.harvard.edu>, "statalist@hsphsun2.harvard.edu" <statalist@hsphsun2.harvard.edu>
Subject   Re: st: Heteroskedastic probit
Date   Thu, 30 Apr 2009 11:07:37 -0500

At 09:48 AM 4/30/2009, Maarten buis wrote:

The variables n5a and Le8c look suspicious, are these measured on a very
small scale, i.e. are the values on these variables much larger than the
other variables? If that is the case, then you might be able to make it
easier for Stata by using a larger scale, for example by dividing these
variables by a milion.

-- Maarten

Good point. Even if convergence wasn't a problem you'd probably want to rescale to make the coefficients more intelligible.

Also, given how insignificant most of the variables are, it doesn't surprise me that there may be convergence problems, especially if normscapab is a dichotomy. i.e. collinearity of the estimates can be a big problem.

Alejandra can try estimating the same model with oglm (available from SSC) but I suspect it won't help. My experience is that oglm tends to give gigantic standard errors when other programs just say that it can't be done. But, if normscapab is totally insignificant in the variance equation, that probably just means that you don't need to have it there in the first place.


-------------------------------------------
Richard Williams, Notre Dame Dept of Sociology
OFFICE: (574)631-6668, (574)631-6463
HOME:   (574)289-5227
EMAIL:  Richard.A.Williams.5@ND.Edu
WWW:    http://www.nd.edu/~rwilliam

*
*   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/



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