nimh – Swing State Project

VA – A foreigner’s (first) attempt at a “good governance” redistricting map for Virginia

Dear all,

This is the first time I've tried using Dave's App and making a redistricting map. I'm hoping you could tell me what things I've done wrong. (For one, I've kept the population of each district very close to the target population, 3,000 off at most, because I didn't know just how close you have to keep it. Could I have been more flexible?)

This is the map, but please do read on:

Virginia

First off, I had an idea with this map. Most of the redistricting maps here are drawn, with great expertise, to either create the most realistically (or unrealistically) Democratic-friendly map; or, in a know-your-enemy / worst-case-scenario kind of way, the most GOP-friendly map; or the map that best fits the current political realities of the state, trying to guess what the real map will end up being.

Since I have nowhere near enough expertise to weigh in, I've just been observing, with some awe. But, as a foreigner, I still have trouble getting over my initial incredulity at the whole American practice of gerrymandering in the first place. The kind of gerrymandering that yields these sometimes mindbogglingly contorted looking districts, tracing unlikely looking paths from, say, the suburbs of one city to those of another halfway across the state, often seemingly without regard to keeping communities together. It's one of the oddest and unhealthy looking aspects of the US political system, for an outsider.

I'm from a country where we don't have any districts at all (it's all PR), and both in my adopted home country and the big European countries I know best (Germany, the UK), nothing like this kind of gerrymandering seems to exist. And often when I see the draft maps here, I can't help fantasizing about what a redistricting map purely based on good governance would look like. (I'm really curious what the new CA and FL maps will look like!)

Continued beneath the fold …

That said, among the many things I have learned about on SSP is the VRA, and I approve. I don't like the 60+% black/hispanic vote sinks that segregate the minority altogether and allow the VRA district incumbent practically guaranteed re-election. But I have totally embraced the need for districts in which minority group candidates are favoured, so minority groups are represented more proportionally in Congress. It's history's bill: it would be great if a black candidate stood as good a chance to be elected anywhere, but the reality is that in too many parts of the country, black/hispanic candidates will only be elected in black/hispanic-majority districts.

I wanted to draw my ideal "good governance" map for a state, the way it would look if there were no bothersome laws, legislative majorities or incumbencies to take account of.

I drew up these criteria:

The number of districts that lean one way or another politically should be roughly proportionate to the parties' general share of the vote. (E.g.: don't stack all the voters of one party into two districts so the other party can easily win in five, when the two parties get about equal amounts of votes altogether).
There should be a number of districts where a minority group candidate would be favoured roughly proportional to the group's share of the overall voting age population. However, districts in which a minority group makes up a small majority (<55%) or a mere plurality, and "coalition" districts in which minorities together outnumber whites should be preferred over segregating individual minorities in 60%+ vote sink districts.
There should be as many competitive districts as possible, both to avoid safe incumbents coasting to victory without having to worry about accountability; and to keep the number of those whose vote doesn't "count" because their party is irrevocably in opposition to a minimum.
Districts should be compact and keep communities of interest together.
While uniting communities of interest is good, it is not beneficial for public policy if the residents of cities and surrounding suburbs are pitted against each other. (A story that struck me was the fight, at a state level I imagine, over public transport in Atlanta, in which the residents of the suburbs managed to block the extension of the city's public transport system because they feared that it would just bring more blacks into their neighbourhoods.) The map should draw cities and surrounding suburbs into common districts where possible.

As you will recognize, but I didn't quite realize beforehand, some of these points make some of the others impossible. Applying point 2 in particular throws a spanner in the works when it comes to points 4 and 5. Creating that many minority-majority districts means contorted shapes, and splitting off black or hispanic city neighbourhoods. Since minority voters tend to vote Democratic, creating more minority districts also means creating more safe Democratic districts, so it's a problem with point 3 too.

Virginia turns out to neatly illustrate all this. My other problem is that I know little about Virginia, so it's hard for me to guess where communities of interest lie exactly in any case. (Any feedback much appreciated.)

Virginia redistricting - data table

(All these data from within Dave's app. I noticed that if you download the data on race by congressional district (18+ population, hispanic and non-hispanic by race) from the census site, there's slight variations, though never more than 1% up or down.

Here's maps with some more detail:

Redistricting map NoVa

Redistricting map Richmond

Redistricting map Hampton Roads

How does this stack up with my criteria?

Fulfilled: My redistricting map creates six Republican districts, four Democratic districts, and one Democratic-leaning district.
Fulfilled: The map creates two districts in which a minority group has a plurality (blacks in VA-3 and VA-4) and a third district in which the minority groups together outnumber non-hispanic whites (VA-11). (I tried to group together disproportionally hispanic towns and neighbourhoods in VA-11, so there’s at least one district where they make up as much as 23% of the VAP – as close to having a district of their own as possible.) That's three minority-favoured districts compared to one now. Pitfall: while non-hispanic whites make up no more than 45% of each of these districts' VAP, they are 43%-45% in each, meaning that disparate turnout rates could also end you up with no minority Congressmen at all. Unlikely in an Obama year, but a concern otherwise.
Failed: My redistricting map actually makes most districts less competitive. This is due to applying point 2. In order to create two more minority-favoured districts, I had to take black votes out of largely white districts, shoring up Republican majorities there. I also took some from the existing minority-majority VA-03, a Democratic vote sink – but that means that instead of having one D+38 district in the south, I ended up with two, still safe D+17-23 districts. Same in the north – by taking black and hispanic precincts from VA-8, I reduced that Democratic vote sink from D+32 to a still safe D+19, while creating an additional safe Democratic seat in VA-11 (D+18). All in all, I went from four arguably toss-up seats (VA-2, VA-4, VA-5 and VA-10) to one (VA-10).
Partial: In the Northeast and the Southeast, districts are pretty contorted looking because of heeding point 2 (though I did manage to cut VA-03 short of stretching all the way up into Richmond neighbourhoods). Elsewhere, some districts are reasonable compact (VA-05 and VA-09, as well as VA-10 up north), but VA-1 stretches a long way across the state… How could this be done better, and do these districts unwittingly split any communities of interest?
Partial: Creating three minority-favoured districts meant splitting a number of cities, in particular Richmond, but also Norfolk, Hopewell, Danville and Franklin – plus the agglomeration in NoVa.

Here's maps of each individual district:

VA-01

Redistricting map: VA-01

VA-02

Redistricting map VA-02

VA-03

VA-03 redistricted

VA-04 (or as I like to call it, the dragonboat

VA-04 redistricted

VA-05

VA-06

VA-07 (the crab, or is it a lobster?)

VA-08

VA-09

VA-10

VA-11

VA-11 redistricted

Mapping the exit polls II – Republicans, conservatives and Tea Partiers, oh my!

Continuing to gratify my curiosity about what the exit poll data looks like when mapped, I was looking at what they said about the number of voters who approve of the Republican Party and the Tea Party movement, respectively.

Nationally, as I noted in the previous diary, 42% of the voters expressed a favorable opinion of the Republican Party. Unfortunately, the same question wasn't exactly asked about the Tea Party. Instead, the exit poll asked voters: Do you support or oppose the Tea Party movement? A total of 40% responded that they somewhat or strongly supported it. Very close to the number of people who thought favorably of the GOP. But are they necessarily the same people?

Just to throw in a couple more layers:

According to the exit poll, no less than 41% of the voters identified themselves as conservative. This is, as DCCyclone commented on the previous diary, an upset: in the past, self-described moderates always outnumbered conservatives, but this time conservatives formed the largest group.
Only 36%, on the other hand, identified themselves as Republican, when asked "No Matter How You Voted Today, Do You Usually Think Of Yourself As A Democrat / Republican / Independent?".

I compiled these data, state by state, in this Google Docs spreadsheet.

42%, 40%, 41%, 36% – these numbers are all quite similar. The biggest similarity, however, is between Tea Party supporters and conservative self-identifiers. Taken state by state, the difference between these two values by state was an average of just 2.6%. By comparison, the difference between any other pair of these four values ranged between 4% and 9%.

So how do the maps compare?

Somewhat or strongly support the Tea Party movement

Considers oneself, on most political matters, a conservative

Again, I used the Google Charts Wizard to create the maps, and they're really plain, and annoyingly, I didn't figure out how to include legends. Or titles. If someone knows how to tweak the URL to include them, or has a easy mapping tool to suggest, thank you.

In lieu of a legend, though: all these maps are set to the same scale, and basically, bright green stands for 25%; yellow for 37.5%; and bright red for 50%.

Top and bottom of the hit lists are roughly the same. Texas is the reddest state of all (ex aequo with Arkansas when it comes to conservatives). Indiana and Arizona are up next when it comes to Tea Party states, with South Carolina and Arkansas coming next, while it's roughly the other way round for the number of conservatives. Vermont and Hawaii dangle at the greenest end (or bluest, really), with far fewer Tea Party supporters than anywhere else.

Tea Party – the mild appearance of the conservative movement?

However much some Tea Party cheerleaders like to describe it as a fiercely independent movement with people from all kinds of backgrounds, it is striking how closely the number of people who support the Tea Party hews to the share that identifies itself as conservative. In 20 out of 26 exit poll states, the difference between the two numbers is no more than 3%. The only exceptions are NY, CT, DE and PA, where Tea Party support is 4-6 points higher than conservative self-identification, and SC, AR and LA, where conservative identification is 5-8 points higher than Tea Party support.

That suggests an interesting regional pattern. Unlike you'd maybe expect, the 'Tea Party' label apparently elicits less resistance in some of the liberal states, and less passion in some of the conservative states, than the label 'conservative' itself. I associate the Tea Party movement with irrational rage, myself, but this suggests that, if anything, it functions as a somewhat milder rebranding of conservatism – like "progressive" being substituted for "liberal"?

Tea Party support – only marginally dependent on individual candidates?

You'd expect the presence of local Tea Party champions running heated Senate campaigns to have an impact on how many people support the Tea Party movement, but it doesn't necessarily seem so. Rand Paul, Ayotte, Rubio, Johnson, Buck, Angle and McDonnell ran in very different states, but the level of Tea Party support in their states in all of them was a fairly consistent 35-43%. Neither does there seem to be much pattern in how Tea Party support in those states compares with sympathy for the GOP or conservative self-identification.

In WI, NH and KY, the homestates of Johnson, Ayotte and Paul, Tea Party support badly lagged GOP favorability (map below). But in Ken Buck's Colorado, it was GOP favorability that lagged, and in the homestates of McDonnell and Angle, the posterboys of Tea Party failure, the numbers pretty much aligned.

Compare Tea Party support with conservative self-identification and the dice roll differently. In Nevada, conservatives outnumbered Tea Partiers. But before you count that as a sign of Angle's weakness, the number of Tea Party supporters in Delaware, O'Donnel's home state, clearly outnumbered the number of conservatives.

In short, it's possible that these candidates collectively dragged down the numbers of Tea Party supporters, conservatives as well as GOP sympathizers, but they don't seem to have particularly affected the Tea Party's standing in particular. Or at least any such effect is drowned out by more structural factors determining conservative and GOP affinities.

More maps:

Favorable opinion of the Republican Party

Usually thinks of oneself as a Republican

There's a striking difference between this pair of maps, but it's a pretty obvious one: there's a big gulf between expressing a favorable opinion of the GOP and actually thinking of oneself as a Republican.

The difference is by far the biggest in Arkansas and New Hampshire, presumably for very different reasons. (New Hampshire has a big tradition of voters identifying as Independents, while Arkansas still presumably has many Dixie Democrats. Or at least former Dixie Dems who've started considering themselves Independents but aren't quite ready to embrace the GOP, even if they've come to have a favorable opinion of it.) In Arkansas, 51% of the voters thought favorably of the GOP, but only 29% identified themselves as Republican. In New Hampshire, that was 48% and 30%. And in no less than 11 of the 21 states where both questions were asked, 9-12% more people expressed a favorable opinion of the GOP than identified themselves as a Republican. Theoretically, I suppose, could you see those states as possible "growth areas" for the GOP?

California, Oregon and Pennsylvania form an interesting contrast. While more Pennsylvanians identified themselves as Republican (37%) than did Oregonians (27%) and Californians (30%), Republican appeal in all three states is basically limited purely to those actively identifying themselves as such. In each state, there's just 3-4% voters who doesn't identify as Republican but could, judging on their favorable opinion of the party, soon start to.

Republican self-identification vs the other maps

When you look at the map of the GOP's favorability rating, while the absolute values are clearly lower than in the maps further above on conservatives and Tea Party supporters, there's at least a striking similarity in the relative order of things. No matter whether you look at the most conservative, Republican or Tea Party-friendly states, among those on which exit polls were published, Arkansas, Texas and South-Carolina are in the top 5. Among the states where all three questions were asked, California is always in the bottom 3, and Oregon and Delaware are in the bottom 5.*

The map showing Republican self-identification, on the other hand, doesn't quite match up. Most strikingly, Arkansas is only the shared 19th most Republican state. Considering that there are almost double as many conservatives as there are Republicans in the state, about a third of the Democrats and Independents in Arkansas must describe themselves as conservative. The same must have held true in other Southern states not long ago, but judging on the numbers for South-Carolina, Louisiana and even Missouri, many more conservatives there have already just started identifying themselves as Republican too.

Tea Party support vs Republican favorability

While the national difference between Republican favorability and Tea Party support was small, there are only a few exit poll states where Tea Party support was higher than GOP favorability. They were Oregon (36% vs 30%), Colorado (41% vs 38%), Washington (37% vs 35%), and two states where the difference was just one point (California and Ohio). The West Coast seems to like the Tea Party better than the GOP – though it doesn't seem to particularly like either. Next up in line are Arizona, where GOP favorability was just 1 point higher than Tea Party support, and Nevada, Pennsylvania and Delaware, where it was 2 points higher. Still a lot of Western states in this list.

There was a larger number of exit poll states where Republican favorability was the higher value, and here the difference was sometimes large. In South Carolina, GOP favorability was ten points higher than Tea Party support (53% vs 43%), though both numbers were above-average. In Iowa (45% vs 36%) and Wisconsin (46% vs 37%) it was nine points higher; in Arkansas (51% vs 43%) it was eight.

I was quickly tempted to say something about the Upper Midwest and the South being less specifically Tea Party-friendly than the West, or something, but that's wrong of course. The Tea Party movement is more popular in Arkansas and South-Carolina than in CO, NV, WA, OR or CA. It's just that the difference between the values for the GOP and the Tea Party are most in the GOP's favour in SC, AR, IA and WI, and in the TP's favour in WA, OR and CO.

Basically, it seems that much like was the case with the "conservative" vs the "Tea Party" labels, the Republican Party brand, perhaps surprisingly, is somewhat more polarizing than that of the Tea Party movement. The standard deviation for the state-by-state GOP favorability numbers is 6.3 and that for conservative self-identification is 7.1; whereas for Tea Party support it's a somewhat smaller 5.3.

(For the Democratic Party's favorability ratings, meanwhile, it's just 4.4, suggesting that the party's support is more evenly spread across state than the GOP's. Part of that is surely due to Dem-voting minorities in otherwise red states, but either way it can't be strategically advantageous, can it, when it comes to winning Senate seats and Electoral Votes for the Presidency? It means that more votes get 'lost' in hopeless states, though I suppose it also means that fewer excess votes are wasted in safe states.)

————

* OK, cheating just the smallest bit: on the conservative share of the electorate, Oregon is shared 5th/6th/7th lowest.

Mapping the exit polls: Obama job approval and Democratic favorability

So I was browsing through the exit poll results, as one does after an election, especially one gone badly wrong. And I thought I'd map out some of the state-by-state results. (I really hope someone else here hasn't already been doing this; if so – my apologies.)

Arguably, part of the usefulness of exit polls is that they are a kind of substitute for regular opinion polls, just with a massively larger sample, and conducted strictly among actual voters, rather than a likely or registered voter screen. So what did they say about Obama's job approval? And what did they say about how the voters looked upon the Democratic and the Republican Party?

Moreover, what do those things look like on a map? Because you need maps. Maps are cool.

According to the national exit poll, with some 17,5k respondents, a combined 45% of voters somewhat or strongly approve of how Obama is handling his job as President.

I thought this was surprisingly high. After all, the population that turned out to vote was a relatily hostile selection: only 45% of the voters said that they had voted for Obama in 2008, while another 45% said they had voted for McCain. So you would expect that these voters approved less of Obama than a wider registered voter or voting-age population sample would. Yet the 45% approval rate the exit poll found is right in line with what opinion polls have been finding since July.

The national exit poll found that 43% of the voters had a favorable opinion of the Democratic Party. Almost the same share, 42%, had a favorable opinion of the Republican Party. Those numbers are roughly in line with what the last two polls on the question that are listed at Polling Report, by CNN and CBS, found. Those found a 46% favorability rating for the Dems, and 44% and 41% ratings for the Republicans.

While job approval and favorability aren't the same thing and Obama's favorability rating would probably have been a couple of percentage points higher, I still thought it was striking that these three numbers are so close to one another.

How do the numbers vary from state to state (here's Arizona)? I tabulated the values for these questions for all the states that were covered in the exit poll in a Google Docs spreadsheet.

I used the Google Charts Wizard to create the below maps. They're really, really plain. Annoyingly, I didn't figure out how to include legends. Or titles for that matter. If someone knows how to tweak the URL to include them, or has a better easy mapping tool to suggest, thank you very much! Some of you guys make amazing maps. (I tried Manyeyes, but its color range doesn't quite work for these values, other than in the bottommost map; plus, ManyEyes and SSP don't work well together, since SSP doesn't accept the "style" tag).

Obama job approval:

Obama job approval by state, according to 2010 exit polls

In lieu of a legend: basically, I set bright green to reflect a 33% approval; yellow to relfect a 50% approval; and bright red to reflect a 67% approval.

There is only one state that falls just outside this range: West-Virginia, where Obama's job approval among mid-term voters was in the tank at just 30%. I set bright blue to reflect a 17% approval, so that's why WV appears as a blueish green.

Democratic Party favorability:

Democratic party favorability by state, according to 2010 exit polls

I used the same colour range as above, so they are directly comparable. This does mean that this map doesn't have a lot of color range. First off, in some of the states with the highest Obama job approval (VT, NY, HI), this question wasn't asked. Secondly, interestingly, opinions about the Democratic Party are less polarized.

In some of the states with a high Obama job approval (CA, OR, DE), opinions about the party were less favorable. Whereas in some of the states with a low Obama job approval (IN, KY, AR, and above all, WV), the party was viewed more favorably. So it all levels out a bit more.

The Manyeyes mapping tool did do a good job in neatly illustrating this contrast. Where does Obama's job approval outdo the party's favorability rating, and where is it the other way round?

In short, the West likes Obama better than it does the Democratic Party; but the Appalachian Midwest (for lack of a better label) probably doesn't, even taking into account the apples/oranges aspect of comparing job approval and favorability. West-Virginia really doesn't. While Obama's job approval, as noted, is down at 30%, the Democratic Party's favorability rating is up at a decent 45%. That 15-point difference compares to difference of at most 7 points in all the other exit poll states.

I want to do more maps focusing on exit polls results on opinions about the Tea Party; the shares of the electorates identifying themselves as liberal and conservative; opinions about the health care law; the Democratic share of the white electorate with under and over $50,000 incomes; and opinions on whether the government should do more or is already doing too much. Please warn me if this has all already been done. 🙂