The probabilities as such are comparable in the sense that they may
refer to the chances of similar events occurring in different samples.
However, what you seem to be interested in measuring is the INCREASE in
probability due to some independent variables, relative to the corresponding
"training set" where those IV were presumable absent. In that case, of
course, they are not comparable. What is comparable in that case is either
(1) the increase in probability (predicted probability minus baseline) or
(2) the size of the effect (i.e. the size of the logit coefficients in
either case relative to the other).
Hector
-----Mensaje original-----
De: SPSSX(r) Discussion [mailto:
[hidden email]] En nombre de Paul
-
Enviado el: 23 January 2007 08:00
Para:
[hidden email]
Asunto: comparing predicted probs from logistic regressions of different
DVs?
I have two logistic regressions, each with a different binary DV
(DV1 and
DV2). The baseline probabilities (proportion of 1s in the training
set) are
different: .15 for DV1 and .25 for DV2. My intuition is that the
predicted
probabilities of the two regressions are not comparable. Is there
something
I can do to make them comparable?
_________________________________________________________________
The MSN Entertainment Guide to Golden Globes is here. Get all the
scoop.
http://tv.msn.com/tv/globes2007/?icid=nctagline2
Locked
