On 2004-11-02, at 19.09, Constantine Daskalakis wrote:
The trouble is that you are computing a quantity (RRR) that is
"non-standard" (ie, non-linear) from the logistic regression model.
Wald-type tests/CIs via the delta method often perform poorly.
I have a suggestion. If data is non-linear, why not log-transform it
before the wald test? Using the auto.dta I posted earlier it would look
like this:
sysuse auto
logit foreign length
// RRR for length=1 vs length=10
nlcom ln((exp(1 * _b[length] + _cons) / (1+ exp(1 * _b[length] +
_cons))) / ///
(exp(10 * _b[length] + _cons) / (1+ exp(10 * _b[length] +
_cons)))) //
// RRR for length=1 vs length=100
nlcom ln((exp(1 * _b[length] + _cons) / (1+ exp(1 * _b[length] +
_cons))) / ///
(exp(100 * _b[length] + _cons) / (1+ exp(100 * _b[length] +
_cons)))) //
The estimates would then have to be exponentiated to describe RRRs with
CIs. On my data this gives very reasonable estimates.
Any comments?
Michael
------------------------------------------------
Michael Ingre , PhD student & Research Associate
Department of Psychology, Stockholm University &
National Institute for Psychosocial Medicine IPM
Box 230, 171 77 Stockholm, Sweden
*
* 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/
