Being something of an amateur data nerd, I decided to take a closer look at the popular vote for the House to see if it looked like the Democrats were likely to come out ahead when all the counting is finished and how one might quantify any disparity between vote totals and seat totals. (First of all, a big thank you to whoever posted the earlier diary with the link to the AP's XML file: http://hosted.ap.org/... .)

Using Excel, I was able to generate state-by-state totals as well as overall totals. Two notes on "methodology" here:

1) The AP is still reporting 0 votes in any race where there was only one candidate. I adjusted for this by adding in 175,000 votes for each of these candidates. Totals in the low- to mid-100,000s seem to be fairly common for winning candidates, and these districts are likely disproportionately tilted towards the party of the single candidate.

2) I removed the totals for special elections, i.e. the few cases where people were voting on the same race twice (once to fill the seat immediately, and again when the new Congress convenes). Only the general election totals are counted for these races.

Anyhow...in brief, the Dems are a smidgen ahead if you actually count the zeroes as zeroes, and a smidgen behind with the 175k numbers that I added in. Obviously, not all races are finished counting yet, but based on what was out there when I downloaded the XML file a couple hours ago, combined with the 175k adjustments, here are the overall percentages, along with three ways to look at how the seats would break down if they were actually awarded proportionally:

Interestingly, if you include the Greens and Libertarians and go strictly by percentage, i.e. multiply everyone's percentage by 435 and round to the nearest, we'd end up with what would be called a "hung parliament" in Britain, i.e. nobody has a majority. Discarding the votes for "Other" (which do, in fact, reflect a smorgasbord of parties and independents) still results in a House with no majority.

Of course, systems which use proportional representation generally have a minimum "threshold" below which parties are not awarded any seats, so if you assume that the Greens and Libertarians are out of luck on that basis, you literally get a 1-seat Republican majority (which, if the Dems were to inch ahead by just 1 vote, would flip to a 1-seat Democratic majority, since 435 is an odd number).

However, here's something else to consider: what if the seat totals in each **state** were closer to the proportion of votes cast for House candidates in that same state? I thought this might be more informative in terms of determining where gerrymandering or just underperformance may have helped or hurt the Democrats and the Republicans, so I again ran some numbers.

For example, here's Florida. The GOP are winning the popular vote for the House, 51.1-45.2. I adjust these numbers to reflect the 2-party vote and get 53.1-46.9. Multiply these percentages by the total of 27 seats and you get a 14-13 GOP advantage, compared to the 17-10 that the GOP actually won:

Here's where things get interesting - I ran these numbers for all 50 states, and found the following. (For the few House races not yet called, I just gave it to whoever's ahead in the AP data.)

1) There are 21 states where this sort of disproportionality benefits the Republicans, compared to 12 states where the disparity benefits the Democrats and 17 where it makes no difference.

2) Among the 21 where the GOP benefit are 7 Obama states - Florida (probably), Michigan, New Jersey, Ohio, Pennsylvania, Virginia, and Wisconsin.

3) Also among these 21 are 4 states where Dems may win the total popular vote for the House but the delegation is majority GOP - Michigan, North Carolina, Pennsylvania, Wisconsin - and 1 - New Jersey - where they will almost certainly win the popular vote but the delegation is tied.

4) By contrast, all 12 states where the Democrats benefit are Obama states, and all of them by double digits with the exception of New Hampshire. (2-seat states are exceptionally likely to show this kind of disproportionality - even freakin' * Idaho* would be 1-1 if you take the 2-party percentages and round to the nearest.)

5) If seats were distributed on a proportional state-by-state basis like this, my current numbers would result in 222-213 GOP majority, compared to the 235-200 tally we'll get if all the most recent leads hold. So still a Republican majority, but one that would fall apart with just 5 defections on any given floor vote.

6) Also of note: while 1-seat states by definition cannot show a disproportionality under this system (anything above 50% rounds to 1), the GOP do benefit from carrying 5 of these states (Alaska, the Dakotas, Montana, Wyoming) while the Dems only have 2 (Delaware, Vermont).

The full table of state-by-state estimates is below if anyone is interested. Also, I'm thinking of taking a look at some actual vote tallies (as opposed to just percentages) in certain states for House, Senate, and Presidential candidates to see how much of this may be gerrymandering and how much of it might be attributable to other factors.

## Comment Preferences