your intuition is right that this should not happpen, but since this kind of thing has been done probably a few hundred thousand times in clarify without this problem, I'm guessing there is something in your code doing it. Hard to tell without seeing the code tho. so I'd check again and if you don't find it send us a copy and we'll see if we can spot the cause.
Gary
-----Original Message-----
From: "Patrick Leblond" <patrick.leblond(a)hec.ca>
Date: Wednesday, Oct 19, 2005 5:55 pm
Subject: [clarify] Predicted Probabilities with Logit
Hi, I have a large binary time-series cross-sectional dataset with multiple imputation (through Amelia) for which I obtain the mean predicted probability and its C.I. through Clarify for a set of hypothetical values for the independent variables. However, if I calculate the (point) predicted probability using the estimated coefficients (obtained through the "mi" function in Clarify) and the specified hypothetical values, I get the opposite result. That is, in the Clarify case the mean Pr(Y=1) is 0.0064 while in my calculation it is 0.999. And I have double-checked my calculation many times. How can I explain this difference? Shouldn't the two results be somewhat similar? Thank you. Patrick Leblond
Assistant Professor
HEC Montreal
Department of International Business
3000 Cote-Ste-Catherine Road
Montreal, QC H3T 2A7
Canada
Tel: +1 514 340 5670
Fax: +1 514 340 6987
E-mail: patrick.leblond(a)hec.ca
--
Clarify mailing list served by Harvard-MIT Data Center
[Un]Subscribe/View Archive: http://lists.hmdc.harvard.edu/?info=clarify
Hi,
I have a large binary time-series cross-sectional dataset with multiple imputation (through Amelia) for which I obtain the mean predicted probability and its C.I. through Clarify for a set of hypothetical values for the independent variables. However, if I calculate the (point) predicted probability using the estimated coefficients (obtained through the "mi" function in Clarify) and the specified hypothetical values, I get the opposite result. That is, in the Clarify case the mean Pr(Y=1) is 0.0064 while in my calculation it is 0.999. And I have double-checked my calculation many times. How can I explain this difference? Shouldn't the two results be somewhat similar?
Thank you.
Patrick Leblond
Assistant Professor
HEC Montreal
Department of International Business
3000 Cote-Ste-Catherine Road
Montreal, QC H3T 2A7
Canada
Tel: +1 514 340 5670
Fax: +1 514 340 6987
E-mail: patrick.leblond(a)hec.ca