Statalist


[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

st: Standardized tobit coefficients


From   "SEVIGNY, ERIC" <SEVIGNY@mailbox.sc.edu>
To   <statalist@hsphsun2.harvard.edu>
Subject   st: Standardized tobit coefficients
Date   Wed, 1 Oct 2008 21:32:52 -0400

Dear Stata Users:

I am trying to obtain standardized tobit coefficients per Long (1997).
Initially, I calculated the standardized tobit coefficients after
estimation per Roncek (1992) as follows:

foreach var of varlist x1 x2 x3 etc. {
      quietly sum `var' 
      display as result "beta* `var' =" (_b[`var']*r(sd))/.5697083)
}

where .5697083 is sigma reported by the tobit model.

Long (1997) criticized this approach because sigma is conditional on x.
He suggests instead to use the unconditional variance of y* computed
with the quadratic form, where r_(y*)^2=B'Var(x)B+s_e^2 (which I read as
the unconditional variance of Y* equals the variance/covariance matrix
of x plus an error term). Unfortunately, I am having difficulty with
transferring this formula into Stata. I am unfamiliar with using and
accessing matrix information, and hope someone on the list could offer
some help.

Thanks in advance.


Eric L. Sevigny

References:
J. Scott Long (1997) Regression Models for Categorical and Limited
Dependent Variables. Sage Publications.

Roncek, D.W. (1992). Learning More From Tobit Coefficients: Extending a
Comparative Analysis of Political Protest. American Sociological Review
57(4): 503-507.

*
*   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