I am currently working with the United States Census data provided by IPUMS International. In the Sample Characteristics Documentation for the United States, it states that 1970 and 1980 are self-weighting with an expansion factor of 100. However, “perwt" (person weight) is comprised of both 100 and 101 for 1970, and 20 and 21 for 1980. I am unsure as to which of the two is correct and if I should be using the weights provided for these years.

After looking through both the IPUMS USA and the IPUMS International versions of the samples, my best guess is that this is a rounding error that somehow ended up in the IPUMS International version of the data causing a limited number of cases to have values of 100 instead of 101 (in 1970) or 21 instead of 20 (in 1980). However, I am still looking for explicit confirmation of this (in the IPUMS USA versions of the same sample, there are only values of 100 and 20, respectively). In the meantime, I suggest running your analysis using both the provided weights and applying the same weight of 100 (in 1970) or 20 (in 1980) to all households and ensuring that your results are robust across the change. I will update you if I am able to uncover an explicit explanation for the variation in the flat weights.