Cases of WTSUPP=0 in ASEC prior to 1988


Looking at the ASEC samples between 1976 and 2016 I found that quite a few cases have a zero value in the weight variable WTSUPP. Most of these cases pertain to years prior to 1988 (if I’m not mistaken, the year in which vacant households started to be included in the data) although year ranges from 1976 and 1999, they do provied information in other variables such as age, sex, occupation or wage income and one or two of them even have a positive value in the household weight variable (HWTSUPP). From these I assume these repondents generally agreed to participate in the survey and that at least most of them aren’t vacant households.

Is there a known explaintion for these cases that I missed in the documentation?

Thanks so much for your help!

Weight values of zero can be treated similarly to negative weight values. These have been historically included to account for oversampling. In general, we suggest that since there aren’t many of these values that researchers simply drop these observations from their analysis. In fact, this is exactly what a zero weight will do in weighted data analysis.