Solving the Gerrymander Puzzle

The term gerrymander seems to be thrown around a lot lately, because, well, everyone gerrymanders.  It is to politicians what Instagram is to teenagers.  And despite what everyone protests, both sides do it.  Regularly.  For every Republican who complains about Maryland, there's one who praises what's happened in Pennsylvania.  We all hate the process, but like everything in life, we complain less loudly when it goes in our favor.

I wanted to set out to mathematically figure out the most egregious gerrymandering in the country, and while I could just look at the maps, I decided to be a bit more scientific in my studies.  Below you will find three different metrics to deduce what the most unfairly proportioned states in the country are.  For those of you who don't geek out royally over political statistics, this might be an article you just skim-we'll have another film review later today.

Metric #1: Percentage of Delegation vs. Percentage of the Vote

For this section, I took the 43 states that have more than two members of the House that they send to Congress and compared the composition of their delegation with the percentage of votes cast for a specific party (all of my calculations utilize the Democrats' vote total, though you can do simple math to figure out what the Republicans should be at).  So, for example, if State A has six congressmen, three Democrats and three Republicans, and they gave 52% of their vote to the Democrats in 2012, they would have a split of 2% (a relatively small number).

Using this metric, the five most misrepresented districts are:

1. New Hampshire (50.06% favorable to the Democrats)
2. Rhode Island (45.61% favorable to the Democrats)
3. Maine (37.08% favorable to the Democrats)
4. Nebraska (35.8% favorable to the Republicans)
5. Connecticut (35.13% favorable to the Democrats)

And the closest were:

1. Iowa (-0.26% favorable to the Republicans)
2. Colorado (-1.08% favorable to the Republicans)
3. Nevada (3.45% favorable to the Democrats)
4. Louisiana (-4.06% favorable to the Republicans)
5. West Virginia (-4.5% favorable to the Republicans)

Looking at these numbers, you can kind of see why this isn't a great metric to judge off of.  The metric skews heavily toward rewarding states with even delegations (Iowa and Nevada both have two congressmen of each party) and punishes small states with one-sided delegations.  Whereas a state like Pennsylvania (22.48% favorable to the Republicans) would require 4 seats to get to a proper equilibrium, New Hampshire, Rhode Island, Maine, and Nebraska would all just need one.  Trying to deduce which states would require the biggest swing in the vote is what I tried to solve for in my second metric, but for the curious, this metric showed that the nation has a percentage favorable to the GOP.

Metric #2: Current Delegation vs. Percentage of House Vote Totals

For this section, I took the size of a delegation, multiplied the percentage of votes received by the Democratic Party, and then calculated the difference between what the state sends to Congress to what should be sent based on the size of the Democratic vote totals.  So, for example, if a state has five representatives in Congress, with three being Democrats, but the Republicans received 60% of the vote in the state, the GOP should theoretically have three seats and the Democrats only two.  So the state would be favorable to the Democrats by one seat.

We'll get into the flaws of this plan in a second, but unlike the previous metric, where really only Iowa was at a statistical zero, here we have multiple states that land as a zero: Utah, West Virginia, Louisiana, Colorado, Iowa, and Nevada all show roughly the correct number of representatives based on the numbers.  The six states with three or more seats that favor one particular party are:

1. California (+6 seats favorable to the Democrats)
2. New York (+5 seats favorable to the Democrats)
3. Pennsylvania (+4 seats favorable to the Republicans)
4. Ohio (+4 seats favorable to the Republicans)
5. North Carolina (+3 seats favorable to the Republicans)
6. Massachusetts (+3 seats favorable to the Democrats)

As you can see here, this method is a more accurate portrayal, though it still has flaws.  The model tends to favor states with larger delegations, which seems appropriate since those delegations tend to wield the most collective power in Congress.  The method also calls out two of the five states where one political party actually gained a majority of the House votes cast but not the majority of the delegation, which seems wildly shady regardless of your politics.  Arizona, North Carolina, Pennsylvania, Michigan, and Wisconsin are the five states that this occurred in, though it’s slightly easier to forgive Arizona and Wisconsin since their delegations only favored the other party by one seat (the other three are completely circumventing the public will).

For the record, even if this model were employed, the Democrats would not have the majority, though they would gain eleven seats.  This doesn’t necessarily mean, though, that the Republicans are correctly in charge of the House.  For starters, were this the law of the land, both parties would fight harder to get a candidate on the ballot in every district.   Louisiana, because the state is so heavily gerrymandered, usually has candidates running essentially uncontested.  Were this the law in the state, the Democrats would be able to pick up enough votes throughout the current Republican districts to justify picking up one more seat.  Additionally, these states aren’t all “gerrymandered” in the traditional sense-California, the most "unfair" state listed, actually had its districts drawn by a nonpartisan commission.

To fix the issue of the lack of representation for both parties on all ballots, though, we'd have to look at the only election where we have a guarantee that both parties are on the ballot: the 2012 presidential election.  

Metric #3: Current Delegation vs. Number of Presidential Votes

While this is a generally good indicator of voter preferences considering the relatively partisan nature of the last election, it isn't a perfect one.  Congressmen like Nick Rahall and Mike Coffman, for example, won districts that their presidential nominee didn't.  However, for the sake of argument, let's take a look at the most egregious offenders under this model:

1. California (+6 seats favorable to the Democrats)

2. New York (+4 seats favorable to the Democrats)

3. Massachusetts (+4 seats favorable to the Democrats)

4. Pennsylvania (+4 seats favorable to the Republicans)

5. Ohio (+4 seats favorable to the Republicans)

6. Florida (+4 seats favorable to the Republicans)

It's worth noting that Florida makes a big jump here (it was +2 for the GOP in the House votes), and that California still remains the biggest offender (the irony of course being that the nonpartisan commission was a piece of legislation pushed by the GOP).  In this scenario, the Democrats do gain the House, but by a scant amount: they would gain 221 seats.  This is done principally by offsetting losses in blue states (the only Romney state that they over-index is in Arizona) with a host of purple states and red states.  Indeed, the losses of the state houses in the 2009 and 2010 elections paid a heavy, heavy price to House Democrats.  Had the Democrats held the governorships of Pennsylvania, Ohio, Florida, Virginia, Michigan, and North Carolina and created districts that reflected the president's results (as can be illustrated in Illinois, they could have gone further but let's not get greedy), they make a quick gain of twenty seats and win the House.  This plan would also make for some vastly more interesting swing states; the three closest races for gaining/losing a House seat in 2012 under this model are Kansas, Tennessee, and Florida, which is apparently always going to be a swing state and would have probably resulted in another recount.

This plan illustrates to the Democrats, though, how much more difficult it is for them to win even if they can even the House playing field: 221 seats is hardly a commanding majority, though it is roughly comparable to the Senate (under this metric, the Democrats would have a 52-seat majority, though that would call into question how fair it is that a state like Wyoming gets the same amount of representation as California in the Senate).

What all of these metrics do illustrate, though, is that while the district-drawing map is by no means a perfect or easy system, Republicans have no justification for complaining about a state like Illinois.  The only consistent pattern through all three of these metrics is that the Democrats are underrepresented in the House based on the results of the last election, which is obviously a huge issue for what is supposed to be "the People's House."