I am writing because I am working on an analysis for various characteristics at both a household level and person by household size. For household level data, I am using HHWT to get the total number of people in NYC, and for person level data I am using PERWT to get the number of people in those household sizes in NYC. When looking at 1-person households (NUMPREC == 1), I figure that the total number of households (HHWT) and total number of people (PERWT) should be the same since we are looking at 1-person households.
However, my analysis shows the sums of these numbers are slightly off with results that show 1,023,468 1-person households in NYC and 1,023,412 persons in 1-person households in NYC.
I was just wondering if there is an explanation in regards to sampling or classification that I should be aware of for this small discrepancy between these data points or if they should be exactly the same.