I want to analyze public and welfare assistance variables (INCWELFR, INCSSI, INCASIST) as welll as variables for earned income (OINCWAGE, OINCBUS, FTOTVAL). Should I use the ASECWT and AECWTH weights for all of these, or should I use different weights for each one?
The correct weight depends on your analysis. ASECWT is appropriate for person-level analyses. ASECFWT is appropriate for family-level analyses. ASECWTH is appropriate for household-level analyses.
If you want to capture the individual-level variation in person-level income variables (e.g., INCWELFR, INCSSI, INCASIST, OINCEAGE, OINCBUS), you should note that people in the same family will have the same value for FTOTVAL; you may want to account for this in your analysis (this also applies to household-level variables). If you instead choose a family- or household-level analysis, you should systematically restrict your sample to only one person per family or household (e.g., PERNUM = 1 for households, or the minimum PERNUM value for each family in households containing more than one family) to avoid double-counting. Note, however, that this will only report the included person’s value for person-level measures (e.g., INCWELFR); you might consider summing these values across all members of the family or household (for family- or household-level analysis) if relevant and appropriate for your specific research application.
@KariWilliams will we not be using EARNWT for this ? Despite it being as ASEC sample ?
EARNWT is appropriate for analyses of the outgoing rotation group (ORG) or Earner’s Study variables; these are asked of a smaller subpopulation (those in month-in-sample 4 and 8 of the CPS rotation pattern). ASEC variables should be weighted with ASECWT(H).