I think you have an interesting dilemma. Multiple imputation would
"fill in" for the missing parent information, but in some respects it
really isn't missing as in "we failed to observe it" but missing as in
"not existent in this family". It's like trying to get the count of
male patients of an OB-gyn. In short these are structural missing
cases.
MI might not do too much violence to these data but I really have my
doubts. I wonder if perhaps you need to do separate analyses, one for
two parent and the other for single parent families (possibly mom only
and dad only, so three groups) and then perhaps plan to make
comparisons using -suest-. An intermediate position might be to
average the parent scores, or combine using the maximum value for the
two parents.
Jay
*
* For searches and help try:
* http://www.stata.com/help.cgi?search
* http://www.stata.com/support/faqs/resources/statalist-faq/
* http://www.ats.ucla.edu/stat/stata/
