09 September 2011

Colorado State Legislative Redistricting Almost Done

Two final plans for the Colorado General Assembly's House and two final plans for its Senate will be considered by the blue ribbon redistricting commission that draws state legislative boundaries on Monday, with a final vote to come a week from Monday. Realistically, any plan adopted is almost sure to survive a court challenge.

One of the main impacts of either plan in metro Denver is to have what is now my House District 3, current held by Daniel Kagan, shift to the south, changing it from a fairly safe Democratic district to one of the most competitive districts in the state. The maps are a bit hard to read but Wash Park and West Wash Park looks like they will probably be ending up in House District 6 (as it was under the 1980 redistricting map), although there aren't enough landmarks on the map that I looked at to discern if some of the neighborhood may end up in House District 2, along with SW Denver.

Metro Denver Democrats had strongly objected to the redistricting plan for the area when it was first proposed and testified against it, but it is hard for someone who hasn't been following the details very closely, like myself, to see if there have been any subsequent modifications to the provisions of the plan that Democrats had found to be objectionable.

Colorado's Congressional redistricting was not accomplished in the 2011 legislative session, so that decision is being made in the District Court in and for the City and County of Denver in litigation that is currently pending. This state trial court will take a bit longer to make a ruling, but is also almost sure to adopt a plan that will survive a court challenge, which would be directly to the Colorado Supreme Court, since it involved election law.

I'm not paying attention, in part, because I lack the time and resources to really give sufficient attention to the details to really make sense of it. Analyzing a single state redistricting plan takes an hour or two and a fair amount of practical political knowledge off the details that won't show up on any map, as well as a few hours in advance trying to draw your own districts in an effort to make sense of what is even possible given shifts in Colorado's population from the 2000 Census to the 2010 Census. Analyzing the Congressional District proposals being floated in the redistricting court case isn't quite as taxing in terms of time and political knowledge (and I've already done the work of tinkering with the new census data to see what is possible), but there is even less room for public participation in that process than there is in the state legislative redistricting process.

For what it's worth, while the redistricting process is inherently political, the process used to draw state boundaries which dampens the impact of partisanship, involves group decision making by a group chosen for political balance, is not reliant entirely on the expertise of experts chosen expressly for their partisanship, and is basically deadlock proof is a superior one to the process envisioned by the U.S. Constitution that requires Colorado's bicameral state legislators to adopt a map, which is almost always followed by litigation, whether or not a map is adopted, either claiming gerrymandering or to have a map drawn because the legislature failed to do so. In times of divided government, this leads to an insular process with district lines drawn by a judge in the course of litigation without sufficient public input and in house expertise. In times of undivided government, this process almost insures a gerrymander in favor of the party in power that despite almost inevitable litigation is highly unlikely to be reversed by the courts, reducing the effectiveness of democratic checks on the party in power.

The only arguments to recommend the current system at the federal level is that the decentralization of the redistricting process allows states where Democrats gerrymander in their own favor to counterbalance states where Republicans gerrymander in their own favor, while leaving states with divided government with either a negotiated result or a result drawn by a judge which on average is likely to be less partisan than a map drawn by either party acting alone.

An important unresolved question in political theory, which I've explored but never had sufficient time to address with any real mathematical rigor, is how much of a difference in political result can be produced by redistricting under what circumstances. This depends upon (1) how many districts are drawn out of a state; a single district for the entire state, or a very large number of districts that must be equal in population, contiguous and compact to some extent greatly reduce the impact of redistricting, while a fairly small number of districts increase the impact of redistricting, (2) the extent to which contiguous areas are political homogeneous, (3) the proportion of the electorate that is effectively up for grabs in partisan elections and their geographic distribution, with formally unaffiliated voters in one place often having political leanings similar to the strongest political party in their neighborhood, (4) the extent to which someone drawing districts favors a "conservative" strategy of drawing many safe seats and few competitive ones, or a "high risk, high return" strategy of maximizing the number of competitive seats, (5) the extent to which redistricting reduces the incumbency advantages of the individuals who currently hold office, (6) the extent to which shifting district composition changes the optimal political strategy for members elected from that office to take to be elected and once in office, and (7) the extent to which redistricting shifts the balance of power of factions within political parties even to the extent that the partisan balance is unchanged.

Even quantifying the impact of redistricting is hard. One of the better yardsticks is to compare the overall porportion of legislative votes at that level cast for Democrats and Republicans respectively in some race where there are top of the ticket races the pull people to the polls even when the legislative race itself is uncompetitive, and to compare that to the outcome expected under a map by some method that assigns each party's candidate a probably of success based upon the political make up of the distict with an empirically validated formula (the result is non-linear - a 60-40 district is almost surely safe, a 50-50 district is in principle one with even odds, and a 55-45 district, while not entirely unwinnable,is a major challenge for even a moderate candidate from the minority party to win) and then looks at the range of outcomes that could result in a Monte Carlo analysis relative to the benchmark of proportional representation. You could then assign gerrymandering ranks to different possible maps.

You'd also probably want to use a large number of examples of real, rather than manufactured data, to evaluate the role of redistricting options, because the practical ability of a map to influence results is heavily dependent upon the underlying geographical spread of political leanings in a state there are correlations in relationships of space and political leanings that exist in the real world but aren't well understood analytically with sufficient accuracy. For example, an important factor in almost all redistricting battles is the fact that the higher a place's population density the more liberal it is likely to be, but subtle differences in how that is estimated in "fake" data could have a big impact on the results.

One of the reasons that it is politically acceptable to let a commission draw state senate and state house lines, but not Congressional districts, is that any map of 35 state senate districts or 65 state house districts that must all be roughly equal in population, contiguous and reasonably compact, with some guidelines regarding how compactness rules are implemented, is that it is much harder to be really effective at having a political impact through gerrymandering subject to those constraints, than it is when drawing a map with just 7 Congressional districts, without making a really concerted effort to draw a really tortured map that a commission with a healthy dose of partisan balance and moderation is unlikely to adopt.

In a state with three or four members of Congress and a highly unhomogeneous political landscape, redistricting can easily de facto decide the mix of the Congressional delegation come election time, in a way that is grossly at odds with the overall partisan vote for legislators at that level.

The more seats there are in a map, the more marginal an impact the redistricting process has on the ultimate electoral results, all other things being equal. The larger number of seats also makes it harder to demonstrate partisan biases in the outcome and to have enough knowledge to evaluate the impact of the map on the likely results. Shifts in partisan balance as a result of redistricting is much more obvious in a map with seven maps.

Of course, once the lines are drawn, it will be time to roll up my sleeves, as everyone involved in electoral politics does, to figure out what strategies make sense in the 2012 election campaigns for legislative offices. Districts that look competitive on paper, like Congressional District 7 in Colorado after the 2000 Census redistricting, can become far less competitive once someone has succeeded in winning the seat and getting relected once - incumbency advantages start to mount and recruiting highly qualified challengers becomes harder.

Redistricting is particularly a big deal in Congressional races. Term limits for state offices mean that open seats crop up with regularity in state legislative races, but in Congressional races it isn't uncommon for the advantages of incumbency to allow a particular member of the House of Representatives to hold the same seat for decades and the shift in their constituency that comes from redistricting may be the next best thing to the level playing field of an open seat that is available to someone challenging an incumbent.

No comments: