Dear Tushar,
With only one observation per subject and dichotomous data, there is not enough information to estimate intersubject variability. Logistic regression without interindividual variability is a simple estimation problem doesn't need any approximation (Laplace or so). In NONMEM you can just specify > $ESTIMATION MET=0 LIKE MAX=500 You should omit or rename the "ID" column, as NONMEM would expect ETAs when ID as in the data set. Best regards, Mats Mats Karlsson, PhD Professor of Pharmacometrics Dept of Pharmaceutical Biosciences Faculty of Pharmacy Uppsala University Box 591 75124 Uppsala Phone: +46 18 4714105 Fax + 46 18 4714003 From: owner-nmus...@globomaxnm.com [mailto:owner-nmus...@globomaxnm.com] On Behalf Of Phyllis Hui-Min Chan Sent: 26 January 2013 13:39 To: nmusers@globomaxnm.com Subject: [NMusers] FW: Intersubject Variability in Logistic Regression > From: Garimella, Tushar > Sent: Friday, January 25, 2013 3:43 PM > Subject: Intersubject Variability in Logistic Regression > > Dear NONMEM-users, > > We are working on a logistic regression problem using a dichotomous response (1 observation per subject, either 0 or 1) and came across the following control stream in the NM Users Guide: > > $PROB DICHOTOMOUS RESPONSE > $DATA data > $INPUT ID DOSE DV > $PRED > LOGIT=THETA(1)+THETA(2)*DOSE**THETA(3)+ETA(1) > A=EXP(LOGIT) > P=A/(1+A) > IF (DV.EQ.1) THEN > Y=P > ELSE > Y=1-P > ENDIF > $THETA .1 (25,100) (1,3,5) > $OMEGA 1 > $ESTIMATION MET=COND LAPLACE LIKE MAX=500 > $SCAT ETA(1) VS DOSE > > We are discussing the role and interpretation of the ETA as shown in the control stream in the context of logistic regression (for example: is it appropriate to estimate intersubject variability in logistic regression, and how can this be interpreted with dichotomous data) and would like to seek input from the group. > > Thank you in advance for your help > > Best regards, > > Tushar > > >
