2020 Census Shapefile Thread

[cinyc]

I didn't know until this evening whether, in the 10 "bad" counties, there would be a single voting district (like "voting districts not defined") in each of those counties, or if those 10 counties would be one combined voting district, or if there would be no voting districts in those 10 counties, period.  I now see it's the third case.  I did have some pleasant correspondence with Dave Bradlee last July, when I shared with him my latest failed attempt to fix the voting district situation in Maine (he appreciated my efforts and thought it was interesting).  But I still don't know what he'll use for the basic units for Maine in his app.  It doesn't look like he'll be able to use just voting districts.  He might just do one with block groups as the one option, that he usually does as a second option.  They're better than the old pseudo voting districts, and certainly better than nothing, but not nearly as good as county subdivisions would be, let alone the precincts and single-precinct municipality combos that are the voting districts in 6 of Maine's counties now.  We'll have to wait and see.

The Census Bureau requires wall-to-wall coverage of counties with VTD's. That is why you see psuedo VTD's being created in Lake Erie or Chesapeake Bay. The Census Bureau only recognizes VTD's as entities for PL 94-171 data. They don't exist between censuses.

Maine towns must extend out to the three-mile limit see that circular VTD around an island down by NH. Aren't some of the islands in Portland Harbor part of Portland and some are independent, like the one they accused of botching the count a few years ago.

Incidentally, the Census Bureau has added 10 states to next weeks tranche which will complete 44 states. They also announced the final eight will be done the following (5th) week. They will be a day later because of the holiday.

Census moved up the final 8 (which include DC & PR) to next Friday. So we should have all the 2020PL shps by the end of next week. They've also quietly released the regular Tiger/Line 2020 shp and Geodatabase files this week, which include '20 Block Groups for all states. There are no Cartographic Boundary Series simplified '20 files on the Census FTP site --- yet.

[Kevinstat]

Census moved up the final 8 (which include DC & PR) to next Friday. So we should have all the 2020PL shps by the end of next week.

Time to update my earlier comparison.  I'll note the the three different groups of states which were announced would be released next week (the last group still having a later scheduled release date than the other two).  I guess I won't worry about the weeks anymore for this year's release of census geography, but I'll keep the weekly grouping for 2011 since the states in those groupings were pretty close together, like in one case how the coming Vermont data was announced the day after four other states.  I've rounded all my figures to the nearest 0.1 unless it was exactly #.#5, where I decided not to round either way here.

2011 Release Week 1: LA, MS, NJ, VA
2011 Release Week 2: AR, IA, IN, MD, VT
2011 Release Week 3: IL, OK, SD, TX
2011 Release Week 4: AL, CO, HI, MO, NV, OR, UT, WA
2011 Release Week 5: DE, KS, NC, NE, WY
2011 Release Week 6: AZ, CA, CT, ID, OH, PA, WI
2011 Release Week 7: AK, FL, GA, KY, MN, MT, ND, NM, TN
2011 Release Week 8: MA, ME, MI, NH, NY, RI, SC, WV, DC, PR

2021 Geography Release 1 (Posted on 01/19/2021): AR, CO, MD, NJ, VA
2021 Geography Release 2 (Posted on 01/26/2021): CA, IA, IN, LA, MO, NC, NY, OH, OR
2021 Geography Release 3 (Posted on 02/02/2021): DE, HI, ME, MS, NE, NV, PA, SD, WA, WI
2021 Geography Release "4A" (the original Release 4) (Posting on 02/09/2021 (announced last week)): AK, AL, AZ, CT, IL, MA, MN, MT, OK, TX
2021 Geography Release "4B" (presumably originally Release 5) (Posting on 02/09/2021 (announced this week)): FL, ID, KS, ND, NH, NM, SC, TN, UT, WV
2021 Geography Release 5 (presumably originally Release 5) (Posting on 02/12/2021 (moved up from 02/17/2021)): GA, KY, MI, RI, VT, WY, DC, PR

2021 Geo Release 1 (5 States): Weeks 1, 1, 2, 2, 4 in 2011 (Mean Week 2.0; Median Week 2)
2021 Geo Release 2 (9 States): Weeks 1, 2, 2, 4, 4, 5, 6, 6, 8 in 2011 (Mean Week 4.2; Median Week 4)
2021 Geo Release 3 (10 States): Weeks 1, 3, 4, 4, 4, 5, 5, 6, 6, 8 in 2011 (Mean Week 4.6; Median Week {4, 5})
2021 Geo Release 4A (10 States): Weeks 3, 3, 3, 4, 6, 6, 7, 7, 7, 8 in 2011 (Mean Week 5.4; Median Week 6)
2021 Geo Release 5 4B (10 States): Weeks 4, 5, 6, 7, 7, 7, 7, 8, 8, 8 in 2011 (Mean Week 6.7; Median Week 6)
(2021 Geo Release 4 overall: Mean Week 6.05; Median Week 7)
2021 Geo Release 6 5 (8 States or State Equivalents): Weeks 2, 5, 7, 7, 8, 8, 8, 8 in 2011 (Mean Week 6.6; Median Week {7, 8})

2011 Release Week 1 (4 States): Releases 1, 1, 2, 3 for 2021 Geo (Mean Release 1.75; Median Release {1, 2})
2011 Release Week 2 (5 States): Releases 1, 1, 2, 2, 5 for 2021 Geo (Mean Release 2.2; Median Release 2)
2011 Release Week 3 (4 States): Releases 3, 4, 4, 4 4A, 4A, 4A for 2021 Geo (Mean Release 3.75; Median Release 4A)
2011 Release Week 4 (8 States): Releases 1, 2, 2, 3, 3, 3, 4, 5, 4A, 4B for 2021 Geo (Mean Release 2.9 2.75; Median Release 3)
2011 Release Week 5 (5 States): Releases 2, 3, 3, 5, 6 4B, 5 for 2021 Geo (Mean Release 3.8 3.4; Median Release 3)
2011 Release Week 6 (7 States): Releases 2, 2, 3, 3, 4, 4, 5 4A, 4A, 4B for 2021 Geo (Mean Release 3.3 3.1; Median Release 3)
2011 Release Week 7 (9 States): Releases 4, 4, 4, 5, 5, 5, 5, 6, 6 4A, 4A, 4A, 4B, 4B, 4B, 4B, 5, 5 for 2021 Geo (Mean Release 4.9 4.2; Median Release 5 4B)
2011 Release Week 8 (10 States or State Equivalents): Releases 2, 3, 4, 5, 5, 5, 6, 6, 6, 6 4A, 4B, 4B, 4B, 5, 5, 5, 5 for 2021 Geo (Mean Release 4.8 4.1; Median Release 5 4B)

[cinyc]

Census released the 2020 PL shapefiles for a bunch of states today: AL, AK, AZ, CT, FL, ID, IL, KS, MA, MN, MT, NH, NM, ND, OK, SC, TN, UT & WV.

Here's what the VTD files look like:

Alabama's VTD shapefile contains about 80% actual VTDs (1,566/1,837):


Alaska's VTD shapefile is all actual VTDs:


Arizona's VTDs are all actual outside of 9 Greenlee County VTDs:


All of Connecticut's VTDs are actual:


In contrast, only about 2/3rds of Florida's VTDs are actual:


All of Idaho's VTDs are actual:


All but 92 of Illinois's 10,000+ VTDs are actual:


By contrast, only 158 of Kansas' 4,240 VTDs are confirmed actual:


Almost all of Massachusetts' VTDs are actual. What's pseudo appears to be largely water:


All but the 3 water VTDs in Minnesota are marked as actual:


All but 9 of Montana's VTDs are actual. The 9 pseudo VTDs appear all to be in Sanders County:


It appears that New Hampshire's pseudo VTDs are in almost all of the multiple-precinct towns, unfortunately:


All but 1 of New Mexico's VTDs are actual. Why that 1? I don't know:


All of North Dakota's VTDs are actual:


All but 52 of Oklahoma's 1,947 VTDs are marked as actual:


All of South Carolina's VTDs are marked as actual:


All of Tennessee's VTDs are marked as actual:


8,555 of 9,007 Texas' VTDs are actual. The pseudo VTDs appear to largely be in rural counties:


In contrast, only 19 of Utah's 2,745 VTDs are actual:


Finally, like ME, WV didn't submit a complete county file to Census. Of the counties submitted, only about 1/4ths are actual VTDs:

[cinyc]

Here's the CONUS map of the VTD files released so far. CA, OR and HI did not participate in the VTD program; GA, KY, MI, RI, VT, WY, DC & PR are scheduled to be relased on Friday.

124,770 of 144,468 VTDs (including AK)  released so far are flagged as actual (86.4%):

[jimrtex]

The Texas precincts in Harris County appear to be actually actual.



Earlier the Texas Legislative Council (Lege Council) which is responsible for redistricting support in Texas had released shapefiles for election precincts based on 2010 Census Geography. In Harris County, there were a number of precincts that were in multiple parts. In some cases it appears that the 2020 streets did not exist in 2010, so neither did the census blocks.

I suspect that it was the Lege Council that acted as the liaison for the BBSP. They were able to map VTD's on the proto-2020 geography, as well as projecting it backwards to 2010.

Prior to 2010, the Census Bureau did not permit use of artificial lines, except if they coincided with a legal boundary, such as that of a city, county, or township. The Census Bureau did use those artificial boundaries for census block boundaries, which were in turn used to aggregate populations for VTD's. From a Census Bureau perspective, VTD's only exist for purposes of delivering PL 94-171 redistricting data. They don't exist in between censuses.

The Census Bureau added the Pseudo/Actual flag to permit states to indicate when a VTD did not correspond to election precincts. If a legislative district was built using election precincts, a "P" flag would indicate that the population might not be exact. If a legislative district was built with VTD's, the "P" flag could indicate that you had just split an election precinct.

Some states require election precincts to be built from census blocks. This allows legislative districts to be defined in terms of census blocks (or block groups, census tracts, and counties). In census geography a county is a collection of census blocks, and the state and county ID form a common part of the block ID for all blocks in the county.

Since the 2010 Census, the Census Bureau has permitted VTD's to use artificial boundaries. In their guidance for the BBSP, they note that these artificial boundaries will be held as census block boundaries, and thus they would comply with state law requiring election precincts conform to census block (that is, the census bureau would split a physical block bounded by streets or other visible features to match VTD boundaries, just as they would city limits).

The Census Bureau is indifferent to the value of the Pseudo/Actual flag. Some states might be indifferent as well. There is a single liaison in each state for defining census blocks and VTD's. They may delegate this to individual counties. That is likely what happened in West Virginia.

In Texas, it may be a case of smaller counties not supplying data to the Lege Council. In rural counties, election precincts may be defined by the highway or road that leads to the polling place. The road is what defines the community, it used as a bus route, and mail route. It does not divide the community but unites it.

*One elections precinct in Harris County does not have a corresponding VTD. It corresponds to a traffic circle. It is 162 square feet, and a perimeter of 62 feet. It could fit in a master bedroom if you could find a way to get it through the door. The Census Bureau deprecated small unpopulated census blocks. That must be the case here. The Census Block no longer exists.

[cinyc]

The final 8 2020PL VTD files were released today, one day early:

DC's was a 129-14 mix of actual and pseudo VTDs :



All but 6 of Georgia's VTDs are flagged actual:


All of Kentucky's VTDs are flagged actual:


Almost all of Michigan's VTDs are flagged actual. Those that are not generally are near water:


All of Rhode Island's VTDS are flagged actual except one water VTD:


All of Vermont's VTDs are flagged actual:


Wyoming's VTDs are a mix of actual and pseudo:


Finally, all of Puerto Rico's VTDs are flagged pseudo:

[cinyc]

Here's the final national VTD map. 86.5% of the 158,444 VTDs were flagged actual in 2020:



CA, OR and HI didn't participate in the '20 VTD program; WV and ME only provided partial county coverage.