Hi,
So based on the boundary files for PWPUMA00, using ARCGIS, I was able to get the file that consists of the polygons for each PUMA. Using some operations, I was able to calculate the longitude and latitude for each of them. Now I’m matching the longitude and latitude with each PUMA in the IPUMS dataset that I am working with, but I realized that some of the polygons seem to be missing from the file that I got from ARCGIS.
For instance, as you can see from the photo that I attached, there is no polygon of State(Minnesota) PWPUMA00(2703) based on ARCGIS so I am not able to match it with the accurate longitude and latitude. For now, I’m matching it with the longitude and latitude of the area closest to the data point, but I was wondering if you have any thoughts as to where this error is coming from. Please let me know if you have any thoughts!
Sorry, the photo failed to be attached so I’ve attached it through this reply:
As noted on the PWPUMA00 variable description, PUMAs defined for place of work (PWPUMA00) differ slightly from PUMAs defined for place of current residence (PUMA). In most cases, the two are identical. For a few cases, however, multiple PUMAs of residence are combined to form a larger PUMA of work. The tables found on this page provide specific details regarding how PWPUMA00 relate to PUMAs.
As far as I can tell, the reason the boundary file contains no polygon for MN place-of-work PUMA 2703 is that it did not officially exist. I checked the 2016, 2015, 2012, 2011, and 2007 ACS samples and the 2000 5% sample, and I found no PWPUMA00 code 2703 for MN in any of those samples. Which sample are you using? You may have found a sample that contains some erroneous codes in it.
Hi,
Thank you for your reply.
I’m using 2005 ACS sample and there seemed to be several data points that seem to be “non-existent” based on the PWPUMA00 (which I believe is for years 2005 to 2011). For instance, for State(Minnesota) PWPUMA00(2703), there are 9 data points (not considering personal weight). Should I just use the longitudes and latitudes of whatever polygons are closest in such cases?
Additional question: So I’m trying to calculate the distance between the place of residency one year ago and the current place of work using MIGPUMA1 codes and PWPUMA00 codes. In some cases, I realized that even if the datapoint indicates that the person has “moved within state, within same PUMA”, the distance is not necessarily 0. Is this because the codes/polygons for MIGPUMA1 and those for PWPUMA00 are defined differently sometimes?
Thank you very much for your help.
>> I’m using 2005 ACS sample and there seemed to be several data points that seem to be “non-existent” based on the PWPUMA00 (which I believe is for years 2005 to 2011). For instance, for State(Minnesota) PWPUMA00(2703), there are 9 data points (not considering personal weight). Should I just use the longitudes and latitudes of whatever polygons are closest in such cases?
These definitely appear to be errors in the 2005 sample. I suppose all that we know about their locations is that they’re in the state of Minnesota, so the best approximation of their location would be the population center of Minnesota, available via this Census web page.
>> Additional question: So I’m trying to calculate the distance between the place of residency one year ago and the current place of work using MIGPUMA1 codes and PWPUMA00 codes. In some cases, I realized that even if the datapoint indicates that the person has “moved within state, within same PUMA”, the distance is not necessarily 0. Is this because the codes/polygons for MIGPUMA1 and those for PWPUMA00 are defined differently sometimes?
The status of “moved within state, within same PUMA” only tells you about a person’s place of residence, not their place of work, so if a person with such a status has differing MIGPUMA1 and PWPUMA00, the simplest explanation is that they work in a different Migration/Place-of-Work PUMA than the one they live in.
