ACS Data Users Group

I am working with the county assessor's office to pull all the property owners in a specific fire district. This county is very helpful and willing to send me what they have. There is an online GIS system where I can look at the boundary lines, below are the options available. The county is currently pulling a list of all the property owners in this fire district

The county is currently pulling a report for me of all property owners in this fire district. They are including the parcel #'s. But, I don't know if that is a data point that can be used for identifying the demographic data I'm seeking, i.e. household demographics (age, # per household, # of children under 18 at a minimum. I know much more is available and will likely grab that as well, but the primary focus is trying to get an accurate pull of data with these squirrely boundary lines. TIA for any help/ suggestions on what to ask from the county that I might be able to plug in for pulling the demographics.

One simple way to do this is to upload an address list: https://geocoding.geo.census.gov/geocoder/geographies/addressbatch?form

which will return the list (csv of xls xlsx file) with the geocodes attached. With the correct selections it will return all the geographies (FIPS codes) in the Census hierarchy. All you need is one address in each tract or block group. You can then download the relevant ACS tables eg B01001 ( A-I) for example to get Age x Sex x Race. It is best if you use a software package such as R.

Dave

Thanks. I did find that, but it's not practical for me to look up every single household that make up the border of the fire district one at a time. I do see a batch look up there but it doesn't tell you the file layout format required for the upload options. I wish I could see map even at the county level that would allow me to zoom in and then I could go in and scoop up the block group #'s manually that way. It would still be a time consuming task but doable.

All you need is one address in each tract (or block group) If you have an address list take a random sample of 9000 addresses. 10,000 is the limit for addressed in one batch on the geocoder. Use the batch option on the geocoder as I indicated above. There are several options on the web page link. I think that I gave you the link where you upload a spreadsheet. The fire or police department must have an address list. You can also get an address list for a county using the Census tigerline database. As an FYI on the geocoder as I think I said it will take csv, xls or xlsx files.

A better way to do this is to take the geo (shapefile etc) for each of the fire districts and look for non zero intersections with the block groups from the relevant counties. With this process for the minimum area of an intersection you shouldn't use zero but a number like 30,000 as the geos intersect somewhat because of streets along the boundaries. R will do all of this for you very nicely ( sf package) I got all the blockgroups for each school district in Pennsylvania this way. There are 500 school districts. It took about 4-5 minutes on my computer.

If I'm understanding the "fire district" problem correctly, we're talking about a geographic area that is NOT based on census blocks. If that is the case, I agree with David Dorer about using shapefile intersections. I have done this often when trying to link demographic characteristics from the ACS to election results at the "precinct" level (precincts are defined independently of census geography). Following the "intersection" method there are ways to get some assurance that the estimates for the characteristics of interest are reasonable. For example, if a county is partitioned into separate fire districts and shapefiles are available for these fire districts, it is possible to get the estimate of some characteristic (e.g., children under 18 years) for each fire district, sum up across all fire districts, and compare the result with the ACS estimate for the county. This will at least tell you if your estimates are close to the "true" result for each fire district. It is important to use an appropriate coordinate reference system (CRS) when doing this. The shapefile for a fire district may be using degrees (longitude, latitude), but the census shapefiles contain columns with total area in square meters. I program in Python, but I'm sure the approach in either Python or R is similar.

Dear Doug,

If a fire district splits a blockgroup (or tract) take the blockgroup with the larger intersection area for example.

Dave