Can original data be different from IPUMS's data?

I am using Mexican Census data from INEGI (original data website) for child mortality. Recently I also extracted the same Census data from IPUMS, but the numbers seem to be a little bit different for specific geographic levels. Do you have any comments or suggestions for this? Can the original data be different from IPUMS’ data?

Yes, IPUMS microdata samples may differ. IPUMS samples are harmonized to facilitate comparative anaylsis. Some samples are drawn especially to IPUMS specifications, such as Mexico 1990. Especially large samples are reduced to ~10%, such as Mexico 2015. These details are explained in the descriptions for each sample: https://international.ipums.org/inter…