I’m pooling 10 years of data from IPUMS. I’m using SAMPWEIGHT for weighted estimates. In the user notes on weights, the only method mentioned to account for the pooling is to divide by 10, but it also states that other “more sophisticated methods of adjustment” are available. What are these methods?
Thank you!!
A more sophisticated method would be to multiply the weight in sample x by (number of observations in sample x)/(number of observations in pooled sample). In many cases this fraction will be very close to 1/(number of pooled samples).
Thank you!
Quick question–wouldn’t it make more sense to multiply the fraction (# of observations by year/total observations) by the sample weight, rather than divide? Thanks!
Yes, that is correct. I must have mis-typed my previous answer. I’ve edited the previous response so to avoid confusion.