There appears to be no “city” variable in any 1970 sample… is there a crosswalk anywhere which would allow me to construct the city variable within one of these samples, perhaps by using the variables countyicp, metarea, and metro in the met 1% samples? If not, would I plausibly be able to make an accurate crosswalk using those variables?
The finest resolution geographic variable in the 1970 metro samples is the CNTYGP70 (county group, 1970 - details about the 1970 country group variable, including a crosswalk between CNTYGP70 codes and county codes are available here). This variable consists of counties or groups of counties that contain at least 250,000 persons. If a county’s population exceeds 250k people, then it receives a CNTYGP70. If a county’s population is less than 250k, it will be merged with one or more counties until group’s population exceeds 250k.
Thus, the smallest geographic entity you can identify in 1970 is a single county. The units identified in our countyicp variable are single counties whose population exceeds 250k (e.g., single counties in the CNTYGP70 variable).
METAREA identifies metropolitan statistical areas, which are single counties or groups of counties. METRO is only available in the 1970 state samples, so you can’t combine METRO, METAREA, and COUNTYICP (or CNTYGP70) to isolate principal cities within metropolitan areas.
Thus, at this time, there is really no way to create a CITY-like variable for 1970. The best you can do is identifying single counties with a population of 250k or more.
Dave Van Riper
IPUMS research scientist