Redistricting and electoral fairness: the view from Eno precinct

As if the election wasn’t annoying enough, I got redistricted this year here in North Carolina. I haven’t moved, but I’m in a completely different congressional district — or, rather, I will be when the 113th Congress convenes in six weeks or so. I wasn’t nuts about my future-former representative, and I like the new guy considerably less, but in the big picture, it doesn’t make much difference, because they’re both in Congress now, and they’ll both be in Congress come January.

But in the bigger picture, redistricting seems to have made a heck of a difference. Republicans won 9 of North Carolina’s 13 seats in the U.S. House of Representatives this month. Combined with the state’s vote for Romney, a newly elected Republican governor, and a re-elected majority in both houses of the General Assembly, North Carolina looks like a very red state, yes? In fact, Democrats won a majority of votes cast in North Carolina Congressional elections this year, even while winning less than a third of the available seats. Welcome to the wonderful world of partisan redistricting.

Details, research, and some history after the jump.

The power of redistricting

In 2010, Republicans won control of both houses of the North Carolina General Assembly for the first time since 1896. That meant that, for the first time in more than a century, Republicans were in charge of drawing the state’s congressional districts. The party that controls this process, obviously, can have a tremendous influence on congressional elections, as Robert Draper recently explained in The Atlantic. This is not a new issue — during the Jim Crow era, southern states typically drew district maps to limit the influence of black voters, but the power of redistricting commissions has grown in recent years with the power of mapping software and detailed demographic studies. (Draper’s article is worth a read; the details of the process are fascinating.)

It was expected, then, that the Republicans would make gains in North Carolina simply as a result of redistricting. But I was curious as to what those gains were, and whether they simply reversed a Democratic bias of the previous set of maps. So I did a bit of research.

The table below shows the total statewide vote in North Carolina congressional elections from 1996 to 2012. For each party, it shows the total votes received by candidates of that party in all districts, the percentage that represents of the statewide vote, and the number of congressional seats won. The last two columns are the takeaway. First, the number of seats that the Republican candidates would be expected to win based on statewide vote (that is, the percentage of total statewide votes received by Republican candidates for Congress times the number of seats available, which is how seats would be allocated in a parliamentary system), and then the number of seats gained by the Republican party over that expectation. The last column, in other words, is the advantage to the Republican party provided by the districting map over a parliamentary system.

Source: 2012 election results in Raleigh News & Observer and Wikipedia.
year Republican Democratic R-predicted R-benefit
votes % vote seats votes % vote seats
2012 2,141,037 48.8 9 2,217,735 50.6 4 6.3 +2.7
2010 1,440,913 54.1 6 1,204,635 45.2 7 7.0 -1.0
2008 1,901,517 45.1 5 2,293,971 54.4 8 5.9 -0.9
2006 913,893 47.1 6 1,026,915 52.9 7 6.1 +0.1
2004 1,743,131 51.1 7 1,669,864 48.9 6 6.6 +0.4
2002 1,209,033 53.9 7 970,716 43.3 6 7.0 0.0
2000 1,514,806 54.5 7 1,193,600 43.0 5 6.5 -0.5
1998 1,014,010 53.3 7 827078 43.4 5 6.4 +0.6
1996 1,339,515 53.3 6 1135731 45.2 6 6.4 -0.4

This year, in all thirteen Congressional elections combined, Democratic candidates received 2,217,735 votes to the Republicans’ 2,141,037 votes. That’s a majority of votes to Democrats (50.6% to 48.8%), yet Republicans won majorities in nine of the thirteen districts and therefore won nine of thirteen seats. That’s a gain of 2.7 seats over a parliamentary allocation. How is that possible?

How does that compare to the results with Democrat-drawn districts? In five elections with the previous maps, from 2002 to 2010, Democrats gained an average of 0.48 seats over parliamentary allocation. That’s a slight advantage, but it’s within a rounding error — you can’t give a party half a seat. Over the previous eight elections, going back to 1996, Democrats gained an average of only 0.2 seats per election. I would say, then, that those maps were fairly drawn, within reasonable limits. As a general principle, fairness would seem to demand that the advantage conveyed on the majority party over the life cycle of a set of Congressional district maps is less than 0.5 seat.

Two and seven tenths does not round down to zero.

To be fair, there’s a danger in looking at just one election. In 2010, the Democratic-drawn map seems to have resisted the Republican tide to an extent; the Democrats won a full seat over parliamentary allocation. But that advantage is dwarfed by the Republican advantage in 2012. Looked at one way, the recent redistricting resulted in a gain of four seats out of thirteen for Republican candidates!

History, redistricting, and race

As I said earlier, this practice isn’t new. We call it gerrymandering after Massachusetts governor Elbridge Gerry, who in 1812 (!) approved congressional redistricting that aided his own Democratic-Republican Party at the expense of the Federalists. One bizarrely drawn district around Boston looked like a salamander, as a Federalist newspaper noted — hence “Gerry-mander”:

the original gerry-mander

Cartoon from the Columbian Centinel of Boston, 1812, via Wikipedia.

Southern states, in particular, have routinely drawn electoral maps to limit the power of minority voters. North Carolina was a pioneer in that effort, in fact: after the Civil War, resurgent Democrats drew the “black second” congressional district to concentrate eastern N.C.’s African-American voters into a single district:

North Carolina Congressional Districts, 1872-1883

Map by the author, from LEARN NC.

(A generation later, Democrats quit fooling around and simply disfranchised African-Americans completely via constitutional amendment, with literacy requirements and the original “grandfather clause.”)

The Voting Rights Act of 1965 established federal oversight of state elections and voting districts to prevent racial bias in districting, but that hasn’t entirely ended the practice; it’s only made electoral cartographers more clever. Texas gained four seats in the House of Representatives in 2010, due to population gains from Latino immigration, yet although a majority of the state’s Latino residents vote Democratic, the new districts were carved in such a way each had a Republican majority.

Moreover, Federal oversight only prevents overtly racial bias in redistricting. Drawn in 1992, North Carolina’s 12th district snaked along I-85 to capture black and Democratic votes in urban areas, resulting in the election of North Carolina’s first African American congressman in a century; after protracted legal battles, the Supreme Court ruled that the district was legal because “race in this case correlates closely with political behavior.” The new Texas maps may yet be overturned by federal courts, and North Carolina’s new ones are being challenged, but my guess is that they will almost certainly survive, because there isn’t anything obviously racist about them. Partisan, yes, but not racist.

Geography and democracy

Take my district, for example. (Please.) For twenty years I lived in the fourth district, represented (except for two years after the 1994 elections) by Democrat David Price. But the fourth district was redrawn to concentrate Democratic voters from urban precincts, and now I’m in the sixth, with northern Orange County and several counties north and west of me that I almost never visit. If I drive a mile to the east I’m in Durham, in the fourth district; five miles west I’m in Hillsborough, also in the fourth; five miles south I’m in Chapel Hill, also in the fourth. Practically anywhere I go when I leave my house is in the fourth district. But my house is not.

Here’s the map from 2002 to 2010:

Several districts were already a bit nutty, but compare with the current map:

Note the long and winding fourth district. The scheme worked; Price won three-quarters of the vote, while the Democratic vote in neighboring districts was spread just thin enough not to matter. Instead of the Black Second, we have the Blue Fourth, which concentrates all of the Democratic Party’s demographic groups. It makes that 1872 map look charmingly quaint.

Am I fairly represented? It’s hard to make a case one way or the other; I’d have no standing to sue. Am I meaningfully represented, though? Probably not. I would argue that if districts are going to bear no relationship to lived geography — if voters are as a matter of course not to share representation with others with whom they live and work — then the notion of congressional districts is no longer logically or democratically valid. We ought to scrap the idea entirely and go to a parliamentary system, with a statewide vote for a congressional delegation.

If, however, we are going to have electoral districts, they should conform to lived geography and to existing municipal borders. Unless absolutely necessary to divide population equally among districts, we ought to try to follow a set of rules like these, in order of priority:

  1. Cities and towns should not be divided.
  2. Cities and towns should be districted with suburban precincts with which they have an economic relationship.
  3. Counties should not be divided.
  4. Borders of districts should be convex unless following existing municipal lines.

But there’s no guarantee that a simple map like that would meet my test for fairness above — that given a fairly large number of districts, the apportionment of seats roughly matches statewide voting patterns. No system is perfect. Local representation and broad-based democracy will necessarily conflict at times, and we could reasonably embrace one principle or the other, or try to strike a balance as we have for most of our history. But the current system isn’t serving either purpose; it’s merely partisan.

Whether it’s specifically Republican, only time will tell. I’d like to see data from states whose maps flipped the other way this year, but I don’t know that there are any states in which Democrats took control of the legislature in 2010 after a decade of Republican control. I haven’t had time to dig into that, nor to compile any sort of national data, and probably won’t. But although Republicans are using these tactics most aggressively now, I see no reason to think that both parties won’t be exploiting the power of Big Data in 2020. This could be only the beginning.

Unless we manage to reform the redistricting process, of course. Groups like this one are pushing for reform, and they have an interesting mix of conservative and liberal groups supporting their cause. But any reform would have to be approved by the legislature — which means, most likely, by the majority party, which either benefits from the current arrangement or expects to benefit from the next round of redistricting. I wish them luck, but I wish I could be more optimistic.