Daniel,
I played a bit with a dataset to reproduce your outcome. I got an output
similar to the one you get from -prvalue- when I ran a model with a small
dataset and a variable that predicts the outcome perfectly. Stata then
drops that variable from the model, as well as all observations with the
value of that variable that predicts the outcome perfectly. Then when I
run -prvalue- I get your same outcome.
If, instead, I run the same model excluding the problematic variable *and*
the problematic observations myself (so it's the same model with the same
sample, only that I drop the variable and the observations, not Stata),
then I get the same regression results, but now -prvalue- runs well.
Did you get a message such as this?
note: [varname] != [value] predicts failure perfectly
[varname] dropped and [n] obs not used
I have used the Spost commands many times and they always worked well. It
must be some sample issue, I guess.
Mariano
Dear list,
I am trying to use some of the commands from the spost package, but
something odd is going on.
I have estimated a logit regression and I tried to use any of -prvalue- or
-prtab-. Unfortunately the commands never report any result. For example:
. prvalue
logit: Predictions for ew_undecided
Pr(y=1|x): . 95% ci: ( ., .)
Pr(y=0|x): . 95% ci: ( ., .)
[...]
Or
. prtab cw1_ew_exp~d
logit: Predicted probabilities of positive outcome for
ew_undecided
cw1_ew_ex
p_spd Prediction
-2
-1
0
1
2
[...]
So I can run the commands but they never show a prediction result. I
believe it is something very obvious and easy, but I just can't figure out
what it might be.. Any hints?
Daniel
*
* 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/