Each household and person has a weight that should be used in producing population-level statistics. The weight reflects the probability sampling process and estimation procedures designed to account for nonresponse and undercoverage. Unweighted counts can be very misleading and should not be used in demographic or labor force analysis.

To produce an estimate of a population level, simply sum the final CPS person weights for all sample persons having the desired characteristic. To make an estimate using a continuous variable (for example, hours worked or earnings), sum the variable multiplied by the weight for the appropriate set of persons. Rates, averages, and other types of estimates can be made by manipulating these two types of weighted totals. For example, to compute the average number of hours worked at all jobs by full-time employees who worked last week, first sum the weights (PWSSWGT) for each person with the variable PRFTLF=1 and PEMLR=1 (full-time workers who were at work last week). Next, for the same set of people, sum the weight*hours worked at all jobs (PWSSWGT*PEHRACTT). Finally, divide the weighted total number of hours worked by the weighted number of full-time workers.