House of Representatives Elections 2010

Senate races are polled frequently, but House races are not. In fact, for the vast majority of House races, there are no polls at all. Still, using available data we can make some estimates of how each House race will go. The methodology and results are discussed below. But before starting a few basic facts are worth noting. Charlie Cook's Partisan Voting Index (PVI) measures how much more Democratic or Republican a congressional district is than the country as a whole based on the two previous presidential elections. Currently 240 CD's (Congressional Districts) lean Republican and 195 lean Democratic. This effect is due to the fact that Democrats tend to be concentrated in cities and Republicans in rural areas. Of the 240 Republican-leaning CDs, currently 72 are represented in the House by Democrats. Only 10 Republicans represent Democratic-leaning districts. Thus if many districts return to their norm, the Democrats are going to take a big hit in the House, independent of everything else.

Prediction Methodology

To make predictions about how CDs will go without having polling data, we have to assemble some other data. In particular, for each CD we know the PVI, the 2008 election results, whether the incumbent is running again, and if so, how firmly established the representative is (i.e., how many elections has he or she won).

With this data available, here is the algorithm used in the table below. These predictions (given in the last column) will be used as our default value until a poll is published. For every race, each of the following steps is evaluated in order until one of them applies.

1. If a candidate has no major party opponent, the candidate wins.
2. An incumbent who got >= 58% of the vote in a contested 2008 election wins.
3. Any CD where the 2008 race was decided by less than 3% is a tossup.
4. Any incumbent first elected before 2000 wins.
5. Any Republican incumbents in a district that is R+3 or more wins.
6. Any Democratic incumbent in districts that is D+8 or more wins.
7. An open seat in an R+5 or more district goes to the Republican.
8. An open seat in a D+5 or more district goes to the Democrat.
9. All other races are Tossups.

Here are some notes on why these rules were chosen. First off, for the past decade, the reelection rate for House incumbents not under indictment has been 94-98%. Even in the Republican wave year of 1994, 90% of all incumbents were reelected. Reasons for this effect include incumbents being better known and having more money than their challengers and having a track record of bringing home the bacon.

For half a century, the party in the White House has done very poorly in the midterms, almost without exception, so the Republicans have the wind at their backs this year. For this reason, we predict the Democrats will not be able to knock off any Republican incumbents in Republican (>= R+3) districts. In contrast, Democrats not in strong Democratic districts (D+5 or more) will be potentially vulnerable. However, any representative who got 55% of the vote in 2008 clearly has substantial personal popularity and will be tough to unseat. Any congressman elected before 2000 has survived both the Bush years and the Democratic waves of 2006 and 2008, and can probably survive anything other than being YouTubed in bed with a dead girl or a live boy (reverse for congresswomen). Open seats are harder to predict since none of the advantages of incumbency are in play, so swing districts (D+4 through R+4) represented by relative newbies who barely won last time are rated tossups and depend on the actual cast of characters and local details. HI-01 and LA-02 are heavily Democratic districts where Republicans were elected due to unusual circumstances that won't be repeated, so we rate them as Democratic takeovers. Once again, these predictions are only used when there has been no unbiased poll published for a race.

The bottom line here is that to keep control of the House, the Democrats are going to have to hang onto a very substantial number of the 72 red CDs they now occupy. In the absence of polling data, it is only a guess how well they will do. But also keep in mind the fact that the voters are often illogical. Generic House polls currently favor the Republicans despite the fact that virtually every poll shows that even though the voters dislike the congressional Democrats by large margins, they dislike the Republicans even more. Let's hope more actual polling data shows up soon.

Results for Each Congressional District

The data displayed below is available in .csv format for you to download and analyze yourself. Also available is the polling data. Finally, descriptions of the most hotly contested House races, with photos of the candidates and many relevant hyperlinks can be found here.

Below is a list of the Democratic and Republican candidates for the House of Representatives elections in 2010. The colors indicate the predicted party, not the incumbent one. When a poll is available, it is used, otherwise the model is used. The columns are as follows.

1. CD = Code for the Congressional District (click on it for race description or map)
2. PVI = Charlie Cook's Partisan Voting Index
3. Open? = Is this an open seat?
4. Democrat = Democratic candidate (incumbents are marked with *)
5. Democat's polling score if available
6. Republican = Republican candidate (incumbents are marked with *
7. Republican's polling score if available
8. Winner = What the poll or model predicts for 2010
9. Po/Mo = Whether a poll or the model was used here
10. Pollster/Why? = If this is a poll, which pollster, otherwise which rule in the model