We generally assume that the donor and recipient samples are drawn from the same population (basically, occupied U.S. households). In principle, the donor and recipient (ACS) should look similar in terms of the distribution of shared variables (X). In practice, we know that they do not -- though it is usually pretty similar. Could allow for the donor weights to be (re-)calibrated so the weighted X marginals in the donor look similar to ACS.
We generally assume that the donor and recipient samples are drawn from the same population (basically, occupied U.S. households). In principle, the donor and recipient (ACS) should look similar in terms of the distribution of shared variables (X). In practice, we know that they do not -- though it is usually pretty similar. Could allow for the donor weights to be (re-)calibrated so the weighted X marginals in the donor look similar to ACS.
See Section 5: https://cran.microsoft.com/snapshot/2015-07-15/web/packages/StatMatch/vignettes/Statistical_Matching_with_StatMatch.pdf