[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: ICE interpretation

From   "Michael I. Lichter" <>
Subject   Re: st: ICE interpretation
Date   Wed, 29 Oct 2008 11:22:53 -0400

It sounds like you may be on the right track. How many cases were missing data on i.gene1 that were non-missing on response1? Another way of putting it: what was the n for the logit on your original data, and what was the n for the data after ice?


Sham Lal wrote:
Dear all, I'm currently having some issues with the interpretation of an imputed dataset. The study aims to identify gene associated with response to drug treatment. I imputed a series of clinical response variables in the imputation equation, to gain what I would expect to be more robust estimates. I did not impute the gene's. Upon using mim and performing various logistic regressions, using for example, xi: mim: logit response1 i.gene1

using this imputed dataset gives me a nearly identical estimates (ORs, SE, 95%ci) as my unimputed dataset (changes occurred 4th decimal place). So in fact i've not gained anything from using an imputed dataset. Could this be because not imputed by predictor variable, the gene?

Any thoughts appreciated
Regards, Sham

Research Assistant
ARC Epidemiology Unit
Epidemiology Research Group
School of Translational Medicine
Room 2.514, Stopford Building
University of Manchester
Oxford Road
M13 9PT
Tel: +44 (0)161 275 1611
Fax: +44 (0)161 275 1657

*   For searches and help try:

Michael I. Lichter, Ph.D.
Research Assistant Professor & NRSA Fellow
UB Department of Family Medicine / Primary Care Research Institute
UB Clinical Center, 462 Grider Street, Buffalo, NY 14215
Office: CC 125 / Phone: 716-898-4751 / E-Mail:
*   For searches and help try:

© Copyright 1996–2023 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index