I’m trying to break down the metro area destinations for people of a certain race who migrated from a specific state (migplace) in the 2013-2017 ACS extract. I’m getting a large chunk of people in “Not identifiable” areas, however.
Does this refer to people moving to places not defined as metro areas or to people who may be in a metro area but we don’t know about or both? Trying to decide if the metros that are reported have too small of a percentage of people compared to the NA category.
Here’s a screenshot of my results. I’m using the SDA interface:
I also ran the query for the metro row variable, and the indeterminable and not-in-a-metro estimates don’t quite add up to the total reported in the “not in identifiable area” (23.3 vs 26.6)
Does this [“Not in identifiable area"] refer to people moving to places not defined as metro areas or to people who may be in a metro area but we don’t know about or both?
Both.
The source ACS microdata contains no information directly about metro areas. The only geographic information given about a person’s residence is the state and Public Use Microdata Area (PUMA). PUMAs are special-purpose statistical units that must have at least 100,000 people in them. IPUMS USA assigns MET2013 metro area codes based on PUMA information, but a PUMA may straddle metro area boundaries. See the MET2013 variable description for details on how IPUMS handles those cases. In general, smaller metro areas–especially those with less than 100,000 residents–cannot be identified in the microdata, and even some larger metro areas do not match well with PUMA-based coding.
What does it mean to have a “mixed” metropolitan status?
The METRO variable is also derived from PUMA information. A household with a “mixed” metropolitan status resides in a PUMA that includes some area both within and outside of metro areas. See the METRO variable description for more info.
Not all “mixed” PUMAs have a 0 code in MET2013. In cases where a PUMA straddles metro area boundaries, MET2013 assigns a metro area code based on where the majority of the PUMA’s population lives (though only if there’s not “too much” mismatch between the metro area and its assigned PUMAs).
Am I better off not using the metro areas identified?
