I did two experiments and found different results!  I wonder why the results of the two experiemnts are different?
- Experiement #1:  I sorted first the data by "caseID", where CaseID is the unique ID of the respondents.  Then, using caseID as subject, I successfully run the syntax and saved the utilities.  I think it is called individual conjoint analysis.  Subsequently, using such utilities data , I computed the utilities for the gender, marital status, and religion groups.  
- Experiment#2:  I sorted the data by gender and run the syntax using gender as subject.  I did the same for the other grouping variables.
 
Thanks.