Hi folks, This question is not specifically related to multiple imputation, but I thought folks might have suggested readings from the missing data literature on treatment of "not applicable" responses. For example, some measures have respondents rate characteristics of relationships with various family members, and items might be left blank if the respondent did not grow up with--for example--a father. On the one hand, it seems odd to treat such responses as missing (and apply modern missing data techniques to them). On the other hand, if such items are believed to be indicators of a larger construct (e.g., conflictual family relationships), then it seems as though it might be okay to treat N/A responses as missing.
I haven't seen this issue addressed much in published papers, but I thought I'd see if I may have missed something. I'd be interested in any leads people can share. Thanks, Jon -- ***Please note change of email to [email protected]*** Jonathan Mohr Assistant Professor Department of Psychology Biology-Psychology Building University of Maryland College Park, MD 20742-4411 Office phone: 301-405-5907 Fax: 301-314-5966 Email: [email protected]
