An Improved Redistricting Strategy

As redistricting proceeds at the national and state levels, it is in the interest of conservatives to consider helping the GOP make the most of its election wins in 2010. In states where it is very unlikely to win every national district (and for every state legislature map), most redistricting strategies will focus on the time honored method of constructing a limited and minimal number of “packed” districts for Democrats and creating as many districts as possible for the GOP to win by a small but healthy margin, say 55-45. A packed district will contain as many Democrats as possible and is “thrown away” to the opposition in order to make GOP candidates viable in more places elsewhere; e.g. a state with 12 districts may make 2 or 3 such that Democrats win them 80-20 or more, and the rest are won by GOP candidates by no more than 60-40. The proportion or # of packed districts ultimately relies on the risk averseness of the map drawer in the face of what damage he or she projects wave elections can wreck and other constraints such as the VRA and ancillary rules.

While modern redistricting software combined with lots of voting returns data and statistical analysis make this exercise more doable than ever, I have come to wonder: are redistricting strategies limited by considering only general election returns and not additionally the effect of primary returns? That is to say, if redistricting made allowances to consider that not all Ds and Rs are the same, might it be used to alter the kind and quality of the opposition’s candidates in addition to altering the proportion of the opposition’s voters?

Redistricting as currently conceived seems to focus exclusively on general election outcomes, and appears to treat all Ds and Rs as interchangeable units. But this is not true in practice – different voters choose different people to represent them in the primary, and the winner emerges to compete in the general election. Let us assume that moderate voters are more likely to choose more centrist politicians, and that more liberal voters are more likely to choose more liberal politicians. This allows the citizens of a district to choose a candidate that best represents their voters for that district, which may be relatively “centrist” or “extreme,” to enhance his or her general election winning probability. This is very important, especially to the democrats who have a large contingent of voters who appear to prefer ‘blue dog’ democrats (fewer these days, but nevertheless); if every democrat had to run as a Nancy Pelosi liberal, they would have drastically lower chances of winning districts they currently occupy outside of the coasts; consider examples such as Mike Ross (AR 4), Dan Boren (OK 2), Jim Matheson (UT 2) and many more.

But what if you could coax the democrats into selecting more extreme, liberal candidates using the same redistricting tools that are already in place, and applying them to primary elections in addition to general ones? What if, by distinguishing between voters in the same party, and by drawing district lines accordingly, moderate and very liberal democrats could be grouped in such a way that blue dog democratic voters are unable to select blue dog democrat politicians? Would this be worth pursuing, and if so for what gains?

If you are of the opinion that the GOP prospers when democrats are at their most liberal public face, then the answer to the last question is “yes,” and the gains are every district that democrats currently win by advertising themselves as centrist and moderate while governing as anything but. Strip away the legislative electoral viability of “centrist” democrats, and the remaining liberal majority has a far harder time accomplishing its goals in normal America, which doesn’t normally swallow very liberal well. If successful, such a strategy would serve to a) ensure that your GOP politicians are always facing the most liberal opponent possible, b) make very liberal (and unpopular) politicians (e.g. Pelosi, Rangel, Grijalva, Waxman) the public persona of the Democratic Party and c) enhance the election chances of GOP candidates. A trivial but effective demonstration of how a-c are possible is presented numerically.

Let’s say that there are 100 voters, all of whom vote in both the primary and general election, and they are to be put into 2 districts of 50 voters apiece. Let’s further say that in this area of 100 people, there are 30 republicans, 10 of whom are moderate, and 70 democrats, 20 of whom are moderate. Say further that I have decided to “pack” District A 100% full of democrats (democrat wins 50-0) to make District B winnable by 30R-20D. Using traditional redistricting, the problem is done and I can stop. But can we do better by considering the effect of which district we place the moderate democrats in? I say we can. Let’s pretend that we put 20 moderate democrats in with the republicans, so that district A is 50 liberal democrats, 0 moderate democrats and district B is 20 moderate democrats, 10 moderate republicans and 20 conservative republicans. Every year, the voters in district A select a liberal democrat in the primary (by 50 liberal to 0 moderate vote). Every year, the voters in district B select a conservative republican in the primary (by 20 conservative to 10 moderate vote) and a moderate democrat (by 20 moderate to 0 liberal vote). In an average year, the democrat will win in A and lose in B. But say we get a democratic wave year, or a particularly “centrist” democratic candidate like Heath Shuler(NC) or Bradley Ellsworth(IN). In that year, say that 6 of the 10 moderate republicans vote democratic because they feel that they are electing a ‘pragmatic, centrist candidate.’ The centrist democrat wins 26-24. Now you have a democratic politician in districts A and B. That is bad.

Instead, let’s consider what happens if we make sure to put very liberal primary voters in district B. Every year, the voters in district A select a liberal democrat in the primary (by 30 liberal to 20 moderate vote). Every year, the voters in district B select a conservative republican in the primary (by 20 conservative to 10 moderate vote) and a liberal democrat (by 20 liberal – 0 moderate vote). In this scenario, even if there’s a wave year, the republican still wins district B – let’s say by 27-23 (20 conservative republicans, 7 moderate republicans to 3 moderate republicans and 20 liberal democrats). This happened because the democrat base demanded a candidate too liberal to win in a conservative leaning district. How did this happen conceptually? Conceptually, it happened because the redistrictor made sure to spread the liberal democratic voters around to make sure that only liberal democrats get selected in primaries, democrats so liberal that even in a bad year, moderate republicans/RINOs won’t defect. This achieves a-c: always make sure the opponent is as liberal as possible, always make sure that liberals are the face of the party (particularly in heartland states) and use it to protect GOP candidates in ‘swing’ districts (those where the victory margins are less than 10 or so).

This sort of thinking, as far as I know, is counterintuitive. Most of us are only too happy to see Alan Grayson and Carol Shea-Porter gone, because we think their policies are hurtful and that they themselves are nasty. This impulse is in line with how we live our personal lives: evict the jerks and if we have to live with people we don’t like, at least make them palatable. But in the world of redistricting, this is exactly backwards. We WANT our opponents looking extreme. We WANT our opponents looking out of touch, nasty, etc. When the opposition hands you an Alan Grayson and a Heath Shuler, you need to keep Grayson and retire Shuler (ideally retire both, but you’re not going to get a Texas or Florida with 100% of either party). Politicians are generally loathe to think this way because they a) accumulate grudges over time, and b) want to face the most moderate possible opposition voters, because in theory there’s a higher chance you can get their crossover vote. What GOP politicians forget is that a) people they dislike are often also disliked by the American people and b) voters who are moderate select candidates that are moderate, and who try to steal RINO crossover voters.

This plan is straightforward in the sense that it doesn’t require one to change fundamentally the basic overarching redistricting strategy. What it does is go one step further and squeeze more money’s worth out of any given map layout in order to eliminate the political viability of blue dog/centrist democrats who pose the greatest electoral threat to conservative republicans. “How do you tell a centrist from a liberal voter,” you may be asking yourself. I’m willing to let the statisticians decide, but I can think of several metrics that can be mined from data available at the county and precinct level: first, look at ticket splitting; e.g. West Virginians who vote 20 points against Obama then vote democrats into every other office. Every state has at least some voters like this (note, this strategy will be more effective at the state level). Second, consider districts that have large swings between elections in terms of vote share. Some districts appear to go R or D rain or shine, while others seem to swing as many as 15 points or more every other election. These are the places you find your moderates. Let’s help our opposition pick the worst candidates possible.

We wouldn’t be very conservative if we didn’t consider the potential weaknesses of a plan which at its heart involves some central planning and an attempt to influence the outcomes of the decisions of free people. Therefore, a word of caution is advised – we should consider that a plan like this might be hard to implement based on the demands it makes regarding the availability of certain data, and the assumptions it makes about voting behavior (moderates select moderates &extremists select extremists, for example). I welcome dissenting views here, especially by people who have worked these streets before. Human nature doesn’t always seem to act like our models say it ought to, as 80 years of social engineering has demonstrated. The PA GOP found that out the hard way this last decade, as their redistricting gains blew up in their face midway through the decade.

The basic idea – to use the ingrained tendencies of our opposition to make them show their worst face to the voting public – should at least be considered. If this idea is not pleasing, then let us at least consider that the current method of considering only the general election and skipping the primary election is at best incomplete.