Dave's Redistricting App

[Kevinstat]

Whichever court redrew the lines in 1994 seem to have cleaned up the lines a lot better in Houston than they did in Dallas-Fort Worth, from a worse starting point.

[Kevinstat]

Reversed situation for Maine. Green district takes in as much Republican territory as possible (and may have been won by Bush 2000, but I wouldn't know):

Dave's app for Maine is rather useless because Maine's "voting districts" according to the Census Bureau (which Dave uses in his app) are chunks of territory in the same State House, State Senate and (County) Commissioner districts from before the 2003 redistricting in Maine.  Congressional districts didn't matter: Albion and Benton in Kennebec County were in different CDs back then (and still are, but the other way), but were both (along with Unity Township which was and still is in all the same election districts as Albion) in "Voting District 141063, Kennebec County" as they were all in the Senate District 14, House District 106 and Commissioner District 3.  Wayne and Fayette in Kennebec County were also in different CDs back then but shared all state and county elected officials (they're now in the same CD but in different House districts) and are thus shown as being in the same voting district by both the Census Bureau and Dave's app.  Some of the voting districts are incorrect even by that standard (Cape Elizabeth and Limington were split between House districts but are shown as being entirely in one of them (in Cape Elizabeth's case the district entirely in that municipality that included most of that town; Limington didn't have a whole district but a majority of its population was probably in the same House district the entire town is shown as being in).  Some islands and island towns and one of the two chunks of Lewiston that were in the southern Lewiston House district but in the all-Lewiston Senate district (the rest of that House district was in a Senate district based in Auburn) are shown on Dave's app as having no people even though they did as of the 2000 census and probably the estimates used as the entire popluation of that "voting district" is attributed to another island or a portion of the mainland in all the same districts in the first case or the other chunk of Lewiston with those same districts in the second case.

Perhaps Dave could be persuaded to use Block Groups for Maine as they would be smaller and be confined to one municipality in much of the state.  Maybe the Census Bureau also has polygons for county subdivisions.

[Kevinstat]

Dave should upload smaller and more sensibly-based polygons for Maine and New Hampshire before he adds any more data to states which already have useable building blocks.  Also, apps for the 1-CD states would be nice to facilitate drawing legislative districts or hypothetical congressional districts (like the kind drawn in the "The Trond can't help it..." thread) in those states.

[Kevinstat]

Will he also do one-representative States one day ? I long for Vermont...

Me too!  I mean that was the one-representative (or 1-CD as I put it) state I am most interested in playing with a Dave's App and that I was thinking most about when I wrote the second sentence in my last post.  Do you live closer to Vermont than to any other state with only one U.S. Representative, like I do.

[Kevinstat]

I gerrymandered Maryland to a 5-4 GOP advantage, and a 6-3 GOP in a good year.

Maryland has only 8 U.S. House seats now, and I haven't heard that it might gain a seat.  Is it within the realm of the reasonably possible?

[Kevinstat]

Not really as even those would have to cross each other. The current setup is an obvious attempt to keep a chunk of Republican exurbia out of Brad Miller's district. And NC-12's long stringy shape makes things a bit more difficult. NC Democrats kind of drew themselves into a corner really though I'm sure they'll find a way to preserve all the Democratic-held seats when redistricting comes around. But it's kind of tricky.

Not really.  The map on the upper left is equivalent to the case in North Carolina.  The map on the upper right is the same, but the two districts no longer have point contiguity.  The connecting sections can be made extremely narrow.

The bottom map is a bit more complex with 4 discontiguous sections in the green section.  But they can be connected.  The map in the lower right is topologically the same as the 3 concentric circles.  I think it is similar to making balloon animals.

Does point contiguity count as contiguity for federal court standards (a) always (including if two or more districts "cross each other" at that point and rely on that point for contiguity) (b) only if no two districts both depend on that point for contiguity and "cross each other" at that point, or (c) never?  I remember reading somewhere that point contiguity didn't count, but I've also read people (perhaps on this forum) write that it does.

[Kevinstat]

For those who have wanted to draw state leg districts or whatever for At-Large states, there is now data for every state except Alaska up.

Hooray!

Dave also now has block groups available for New England states, which, for Maine and New Hampshire at least, are better than the voting districts which for Maine were the tracts of territory in the same county, state senate district, state house district and county commissioner district (but not necessarily the same congressional district) before the 2003 redistricting (so under the lines drawn based on the 1990 census) and in New Hampshire were the wards of cities (or perhaps some non-cities) with all other municipalities in each county being lumps into "Voting Districts not defined, [Whatever] County" (except for Carroll County which apparently has no cities and had the entire county in the "CARROLL Voting District").

In both of those states (at least), however, some block groups include multiple towns, in Maine at least one pair of towns (Rome and Vienna) which were not only in different "Voting Disrricts" (as they were (and still are) in different state house districts) but were each coterminous with one "Voting District."  Overall, however, the block groups are definitely more useful, although it would be nice if the name of the census tract were included in the block group name rather than just "Block Group 1" or "Block Group 2" and if I new how the '(part)'s of block groups were divided (at least adjacent "Block Group 1"s of different census tracts in the same county are in different districts ('disticts' here being the indivisible building blocks of 'CD's)).