US House Redistricting: Ohio

[Nichlemn]

13 and 2 also could well go Democratic with such large portions of Cincinatti and Cuyahoga.

A Generic R could easily hold 2, given that Eastern Hamilton County is more Conservative than the rest.  The Current one is R + 13, and the one drawn here is not much different.

Actually both CD 1 and 2 as I drew them would have voted McCain in 2008. CD 1 is closest at 51.5-48.5, and CD 2 would have been 54-46. In an even statewide year like 2004 they both would be about 56 or 57% R.

What's your approximation for CD-6 & CD-7.  They both seem to have the largest chunks of Franklin county.

They are fairly close and probably would have been for Obama by 51-49 or so, but would be R in most years. With precise precinct data I could adjust that by shifting Champaign (and maybe Logan) to 7 and bringing 4 deeper into Columbus. CD 8 could move into Springfield if needed and give up some R areas to CD 6.

Youve drawn a map where Democrats could win OH-01, OH-06, OH-07, OH-09, OH-10, OH-14, OH-13, OH-16, and probably OH-02 if Schmidt is still around in a good Dem year.  Republicans are going to have to concede a seat to Democrats in Columbus to keep the surrounding districts at least 54%-55% Republican.  

CD 9, 10 and 16 are strongly D and with CD 11 make up the 4 certain D seats in this map. As I noted, CD 1 and 2 would have voted for McCain, so it would take an even larger Dem wave than 2008 to topple the seats. I agree that as drawn CD 6 and 7 are vulnerable, but both are close with 2008 data. I'll post an update that can keep them R in that year.

2008 wasnt really a wave and 2006 wasnt that big.  We still havent seen a Dem +50 wave since 1974.  

Technically, the last D+50 election was 1948 (although 1958 and 1974 were both D+49).

[Nichlemn]

It's not at all obvious how aggressive a gerrymander should be. It's certainly not as simple as "draw a map that will be safe for the whole decade".

In good years for the GOP, Torie's map would likely be 13-3. In particularly bad years, it could easily be 3-13. But that doesn't tell us anything in and of itself, because we haven't put a value on an additional GOP Congressman as a function of how many seats the GOP holds. If you think it's important to have a strong "buffer" against Democratic waves in order to reduce what a disunified Democratic Congress could do (e.g. pass ObamaCare),  then you'd weight this highly. However, perhaps you would prefer to weaken this in exchange for an additional GOP House member in GOP waves, which could potentially be enough to pass a controversial bill or override a veto.

Neutral-ish years (where control of the House is at stake) are also tricky. It depends on how much variance in the Congressional vote is determined by PVI. Currently, this appears to be on the rise, but there is nonetheless still a good chance that Democrats could win in Republican-leaning district otherwise neutral years. Conceding a seat to shore up all other districts would reduce these chances, but it should do so by more than one expected seat.      

This paper on optimal gerrymandering may be helpful for mappers (warning: math).