Do I need to factor in PERWT when calculating median/average INCWAGE for a sample?


This is a follow-up question to one I asked earlier this week. I don’t have experience weighting or handling data with weight values.

I have created a CSV format data extract for bachelor degree holders (DEGFEILD excluding N/A) that are in the labor force (EMPSTAT excluding N/A and not in labor force). My goal is to use Excel to calculate average and median income for each degree (DEGFEILD).

Do I need to factor in PERWT when calculating these functions (i.e. multipy PERWT values by INCWAGE and then calculate the average) or can I ignore PERWT in this case and use a pivot table and the MEDIAN function?



Yes, I think you should use the weights if you want to values to be representative!



Yes, anytime you are calculating a statistic that is representative of a particular group of people within the population, sampling weights are necessary. Without the application of PERWT the statistics will only represent the respondents in the survey and will not account for differing probabilities of different individuals to be included in the sample. More detail about sampling weights in general can be found in this blog post.