What weight should I use in IPUMSCPS
The IPUMSCPS samples are weighted, with some records representing more cases than others. This means that persons and households with some characteristics are overrepresented in the samples, while others are underrepresented.
User Note: Weight variable names were recently changed to incorporate basic monthly data into IPUMSCPS. Weight variables specific to the March supplement have been renamed: PERWT is now WTSUPP; HHWT is now HWTSUPP. BLSWT has been renamed WTFINL; it can be used to replicate published Bureau of Labor Statistics figures based on basic monthly survey questions and should be used in analyses of nonMarch data. Weights appropriate for householdlevel analyses of nonMarch data are also available (HWTFINL).
To obtain representative statistics from the samples, users must apply sample weights. Follow one of the following procedures:

For most personlevel analyses of March supplement IPUMSCPS data, apply the WTSUPP variable. WTSUPP gives the population represented by each individual in the sample.

For personlevel analyses of nonMarch IPUMSCPS data, apply the WTFINL variable. WTFINL gives the population represented by each individual in the sample. Analyses to replicate published Bureau of Labor Statistics figures should also use WTFINL rather than WTSUPP.

For personlevel analyses analyzing the summary health insurance variables (HCOVANY, HCOVPRIV, HINSEMP, HINSPUR, HCOVPUB, HINSCAID, HINSCARE, andHINSMIL), apply the HINSWT variable rather than WTSUPP.

For personlevel analyses using the IPUMSCPS variables EARNWEEK, HOURWAGE, PAIDHOUR, and UNION, apply theEARNWT variable.

For householdlevel analyses of March supplement data, weight the households using the HWTSUPP variable. HWTSUPP gives the number of households in the general population represented by each household in the sample.

For householdlevel analyses of nonMarch data, weight the households using the HWTFINL variable. HWTFINL gives the number of households in the general population represented by each household in the sample.