I was wondering if you could confirm that that the a_indinus_xw weight variable should contain 3,933 observations (or about 7.71% of the total indresp file )where a_indinus_xw=0? I was analyzing a subpopulation of the data, and noticed that the subpopulation # of obs was lower than expected, due to this weight being zero.
Many thanks for your response.
#2 Updated by Olena Kaminska over 7 years ago
inresp dataset consists of proxy and full respondents. If you are using information available for proxy respondents, you can use everyone in the dataset and the correct weight will be a_indpxus_xw. If you are looking at full interviews only, the correct weight is a_indinus_xw. Data documentation has further information on this.
#3 Updated by Sung Park over 7 years ago
Thank you for your response! I believe when I compare the a_indinus_xw and the a_indpxus_xw weights, proxy respondents account for most of the 3,933 observations. However, there are 699 cases where the values for both weights are zero (a_indpxus_xw=0 and a_indus_xw=0), and it seems like these 699 cases have values for questions that were asked of full respondents. Should these cases be considered proxy or full respondents?
#4 Updated by Olena Kaminska over 7 years ago
The proxy weight of 0 in indresp is given to temporary sample members whom we will not follow in the future. These are non-ethnic minority respondents living with ethnic minorities. For details, please look at the weight design section of documentation and at Berthoud et al. working paper on ISER website.