I am using the CPS ASEC to analyze likelihood of poverty, setting a poverty threshold using household level income. I need to do a supplemental analysis that includes union status as a control, so I am using the outgoing rotation group question for that measure. The issue I have run into is, it is important for the analysis to maintain all individuals in a household. But, EARNWT assigns unemployed, children, and military a value of 0, even when they are in a household with an individual who did answer the earner survey. Therefore, I can’t really use EARNWT to weight my analyses, because it would exclude these household members with weights of 0. Should I use ASECWT instead? Are there any other suggestions or options?
I am also pooling across years if that is relevant.