Poverty data in 1950 census sample


I’m trying to calculate the percentage of people in each state that were below the poverty line in 1950. It seems like the 1950 1% sample available through iPums-USA is the best source of data on this. However, I’m getting confusing results. When I look at the micro-data, it appears that only 10% of the people in 1950 were at or below the poverty level. However, this article suggests that 40% of the population in 1950 was below the poverty rate (http://www.cepr.net/documents/publica…). Is there something obvious I’m doing wrong? Perhaps I should subset the data to heads of households?



This is likely related to the fact that in the 1950 Decennial Census income information was only requested of “Sample-Line” individuals. When examining variables from the smaller, sample-line population you should use the SLWT weight variable for representative estimates of the national population. Variables that require the SLWT weight will note that the universe is composed of Sample-Line individuals in the Universe tab.

I hope that helps.