In addition, note the -modeldiag- package
written up in Stata Journal 4(4). Type
. search modeldiag
which indicates how to install. You can do things like
. indexplot, show(rs)
after a -logit- call. Thus Maarten does things
from first principles, whereas -modeldiag- will
often do the spadework for you.
Nick
[email protected]
Maarten Buis
> > I am interested in doing outlier analysis on logit
> regressions. Dfits and
> > some other commands do not seem to work following logit.
> What techniques
> > are available to use with logit or probit regressions?
>
> If you type -help logit- you also get information about what you can
> -predict- after -logit- which includes various influence statistics.
> After you have -predict-ed them you have variables containing those
> statistics which you can than use to create your favourite graphs,
> see example below.
>
> *--------------begin example--------------------
> sysuse auto, clear
> logit foreign price mpg
> predict rs, rs
> predict dbeta, dbeta
> predict hat, hat
> sort price
> gen long obsnr = _n
> twoway scatter rs obsnr, mlabel(obsnr) msymbol(none) mlabposition(0)
> twoway scatter dbeta obsnr, mlabel(obsnr) msymbol(none)
> mlabposition(0)
> twoway scatter hat obsnr, mlabel(obsnr) msymbol(none) mlabposition(0)
> list foreign price mpg if obsnr==74
> sum foreign price mpg
> *---------------end example----------------
*
* 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/
