Question about 1950, 1960, 1970 tract-level housing and population data


Dear Colleagues:

I am sending this message directly, as I ran into some problems getting messages through on the user forum.

I am finding that the 1960 Census tract housing unit data are indicating a large upward spike in population and housing in some poor urban Census tracts between 1950 and 1960. This upward spike drops back to previous levels in affected tracts by 1970. I’m not sure how to address the issue and if the observed pattern is real or reflects data error or measurement issues. The problem appears to occur in a few areas that were subjected to urban renewal and new housing development, though this is not a wholly consistent pattern. Also, this does not seem to explain why the housing unit and population data spike upward, then drop, over the 1950, 1960, and 1970 Censuses. It’s incredibly unusual behavior, even considering urban renewal and related interventions in those urban neighborhoods.

Have you observed this issue before? I have enough concerns about the accuracy of the 1960 data (or its non-compatibility with Censuses before and after) that I’m tempted to drop analyses that use it as an outcome variable.

Is this a known problem associated with the use of found ICPSR data for the 1960 Census?

Clayton Nall


Hi Clayton,

Would you be willing to post a few of the census tracts that seem to have this issue? I can then take a look at them and compare them to our internal NHGIS data. I want to make sure we don’t have a data problem in the 1960 data.


Dave Van Riper