Note: I've worked on this entry with Devin McCarthy, a colleague at FairVote who has done excellent work on our new Monopoly Politics 2014 report about U.S. House elections. We'll both chime in with more thoughts in the comments.
Last week, FairVote unveiled its Monopoly Politics 2014 congressional projections, calling 211 seats for Republicans and 163 for Democrats. Take note of them, download the spreadsheet, and let us know how we do. The only changes we expect to make between now and November 2014 will be if an incumbent decides not to run or if districts are redrawn - and in either instance, the formula governing our projection model will remain the same.
We were pleased that the projections have received a good deal of major media attention, including a New York Times blog post and my segment on Chuck Todd's The Daily Rundown.
The projections also provoked some discussion here at Daily Kos, and we'd like to respond to some of the comments and critiques we received.
First, here are some general clarifications on the methodology of the projections.
* We're being cautious: We are more interested in being right about every single projected winner than in projecting as many races as possible. That helps explain our perfect record in 2012.
* We're keeping our model simple: Every race in the country is projected using the exact same formula based on just a few points of data from recent elections. Those surprised by a particular projection may not be factoring in this one-size-fits-all approach.
* We're using simple categories: FairVote uses the terms "Safe," "Likely," "Lean," and "Toss Up" to describe the competitiveness level of races because they have become common parlance for discussing congressional projections. As we're using them, though, the terms really just refer to how close we're projecting the election to be. "Toss Up" races, for instance, refer to races in which we project the two party margin of victory to be less than 6%. That is, admittedly, a fairly cautious definition of "Toss Up" that is likely to include more races than other predictors' definitions.
* Incumbents of both parties get a significant edge, and Republicans right now have far more incumbents: Incumbents on average win by a margin of about 10% to 15% more than a nominee of their party would in an open seat race in their district. Incumbent advantages include the ability to build on their last campaigns, the media attention they earn during their term and the ability to provide non-ideological services to constituents. So having more incumbents is a big help.
* The playing field is skewed Republican, and has been for a long time: We suspect it's hard for people to accept just how skewed congressional districts are toward Republicans. Our 2012 report explains this in far more detail, but Democrats recalling their wins in 2006 and 2008 may not see things the same way we do. Those House victories were possible because many voters who voted for Romney in 2012 voted for Democratic House candidates in those elections. They are very unlikely to split their tickets again. The combination of the 1994 and 2010 elections have hardened partisan divisions in our country to the extent that Democrats' ability to win the support of such voters is greatly diminished. That means the underlying skew toward Republicans of more than 40 House seats cannot be ignored or wished away.
* We assume a 50%-50% national two-party division, but users can adjust that national division and the projections will change accordingly: Our analysis of the 2012 elections suggests that the underlying national preference in House races was a 52% to 48% advantage for Democrats. Our projections for 2014 are based on an assumed 50%-50% two-party year, but it's possible that will not turn out to be the case.
This last comment relates to the most common specific criticism of the report in DailyKos comments - that it is "way too Republican friendly," as Stephen Wolf wrote, or that our ratings represent a "GOP wave scenario."
To be clear, FairVote is a nonpartisan organization, and (to reiterate) our projections are based on a formula that applies equally to both major parties. If FairVote's projections seem to unfairly favor Republicans, it may be because we see 2012 as a "mini-wave" year for Democrats, one with a 52%-48% Democratic edge. Our district competitiveness ratings are based on 2014 being a 50-50 year between the parties, meaning that Democratic incumbents will on average earn 2% less of the vote than they did in 2012 and Republican incumbents will earn 2% more.
One cool feature of our projections, if you open up the spreadsheet, is that it allows you to manually adjust how well you think Democrats will do overall in 2014. If, for instance, you predict that 2014 will be another 52% Democratic year, you can see exactly how much better Democrats are likely to do than in an even year (not very much better, incidentally).
We're not going to address every criticism of one of our specific district ratings here, though we do stand by all of those ratings. Let's take a look at one as an example, though: our projection of Rick Nolan's MN-8 as a tossup, which prompted Stephen Wolf to exclaim "what the hell." MN-8 has a partisanship of 50.8% Democratic - that is, it supported Barack Obama at a two-party rate of about 51% in 2012 relative to his national average. While it's true that Democrats hold onto their "turf" better than Republicans these days, that's still a very closely divided district.
In order for the race to move into "Lean D" territory, Rick Nolan would need to have a stronger electoral history in the district. That's not really the case, though - he only has one election worth of data to back him up, in which he beat incumbent Chip Cravaack by a respectable but not overwhelming 9% in a Democratic-favoring year. Our model significantly de-weights the projected influence of first-term incumbents on a district's partisanship, simply because first-termers tend to be the most vulnerable - they are the least tested and have the fewest incumbent advantages. As a result, we project Nolan to receive about 52.4% of the vote - more than Democrats would expect in an open race, but not enough to put the race out of the tossup column.
FairVote welcomes further discussion of our projections and the methodology by which we produced them. And we urge users to download the spreadsheet and explore it, as that's where you really can see how congressional elections work.
Finally, keep in mind that we believe our report exposes the bankruptcy of winner-take-all elections as an effective instrument of representative democracy in 21st century America. Using this exact same model, we will make our 2016 projections within weeks of the 2014 election - after what we suspect will be a highly accurate record this cycle.
Regardless of any quibbles about individual races, our projections indisputably demonstrate two key facts about the 2014 election: 1) the vast majority of races are predetermined years in advance of the election, and 2) the Republican Party has a major structural advantage over Democrats in a balanced or even slightly Democratic year.
Maintaining the rules that have resulted in these facts is a statutory decision - far easier to change than other much-discussed election reform proposals such as a constitutional amendment to overturn Citizens United. For information on our fair voting solution to uncompetitive and skewed congressional elections, check out our interactive map at fairvoting.us.