I’m working with the linked 1850-1880, 1860-1880, and 1870-1880 samples. When I look at a specific state/county/mcd sometimes there are different urban/rural classifications for the households living there in the same year. For example, the Barnstable, Mass. 1850 sample (year_1=1850; stateicp_1=3; county_1=10; mcd_1=200) classifies 3 people as urban and 3 as rural.
Am I missing a more finely defined geography that puts some people in the urban part of Barnstable and others in the rural? I checked incorporation codes and other geographic identifiers and nothing seems different about these 6 observations other than that 3 are classified “urban” and 3 “rural”. This doesn’t seem to be a coding error since I think it’s true of about 10% of the sample.
This observation is due to a fairly specific detail about the coding of the URBAN variable. In the cases where you see different values of URBAN within the same MCD, there are “villages” or other “thickly settled areas” that are considered urban while other areas are considered rural. This is particularly the case for New England cities and towns. A little more detail is noted about this on the comparability tab of the URBAN variable.
Thank you, I very much appreciate your answer and it makes sense to me. It’s too bad that, to my knowledge, these linked datasets do not allow you to distinguish whether an observation comes from a thickly-settled area. So aside from state/county/mcd, I can’t have any information that tells me the name of the area an observation comes from, for New England.
Thanks again,