House battle at relative standstill, but watch generic ballot


After shuffling many of our House ratings, it’s pretty obvious that the race for the House remains locked in a battle of trench warfare, with little obvious movement on either side. Given that the Republicans start from a position of great strength — Democrats need to net 25 seats to take control of the House — the GOP remains a heavy favorite to hold the lower chamber of Congress.

Our new ratings show 195 safe seats for the Republicans and 156 safe seats for the Democrats, with 14 likely Republican seats and 13 likely Democratic seats. That leaves 23 leaning Republican and 20 leaning Democratic with 14 toss-ups. Only 57 of 435 seats — 13% — are in the leaning or toss-up categories.

Our last full House update projected a Democratic gain of six seats; this time, we’re projecting a Democratic gain of four seats. Given the nature of the House — where many races are likely to be close to the end and where reliable polling is relatively scant, compared to Senate or gubernatorial races — our projections are bound to change, perhaps dramatically, in the last seven weeks before the election. But the center of gravity in this race — two sides locked in a close contest — is clear, and it favors the Republicans.

Parsing the generic ballot

Democratic Congressional Campaign Committee Executive Director Robby Mook said a couple weeks ago that the race for the House was “in a neutral environment,” but that if the generic House ballot was at Democrats plus five points, “we have a good shot.” We think the Democratic advantage in the generic ballot would have to be higher, but this was a relatively clear-eyed — and, for Democrats, potentially off-message — assessment.

The generic House ballot, a polling question that asks whether a respondent plans to vote for the Republican or Democratic candidate in his or her local U.S. House race, is an imperfect but useful tool in measuring the direction of the House race.

Chart 1 shows the average generic ballot on Sept. 19 for the past five national elections. In 2006 and 2008, Democrats had already assumed a fairly big lead (six points or so), which was indicative of their sizable seat gains in those two cycles, and Republicans had also opened a slightly smaller lead at this point two years ago (4.5 points), which eventually ballooned into a 9.4 percentage point generic ballot edge in a year in which they picked up 63 House seats. Meanwhile, 2002 and 2004 saw smaller leads for Republicans and Democrats, respectively, at this point in those cycles; Republicans ended up making modest additions to their majorities those years.

Chart 1: House generic ballot and final results in recent House elections

Notes/Sources: For 2006-2010, the author used the RealClearPolitics average as reported on the website on Sept. 19 of those years. For 2002 and 2004, there was no running average on that date, so the author averaged the five polls taken before or through Sept. 19 to calculate the average. Popular House vote differentials are from Vital Statistics on American Politics 2011-2012. Years are shaded blue or red to indicate which party gained seats during those cycles.

Granted, the generic ballot is not exactly the most precise measure — as is clear from the chart, the final average is typically off by several points from the actual combined national House vote — but Democrats have made some recent movement in the generic ballot, perhaps related to President Obama’s polling bump from his recent convention; they now lead by 2.2 percentage points in the RealClearPolitics average (the full trend for the year is available at that link; it kind of looks like a roller coaster, albeit one for little kids given the small range). Other averages have the Democratic lead in the three-to-four-point range.

How predictive is the total House vote of the number of seats won? Not perfect, but certainly more predictive than it used to be.

In the 24 House elections held between the end of World War II up to the 1994 Republican Revolution (1946-1992), Democrats won an average of 52.6% of all votes cast in those House elections but captured 58.6% of the seats. Meanwhile, Republicans won 46.1% of votes cast but only 41.3% of the seats. So Democrats — probably in part because of the very Democratic Solid South — typically won more House seats than their performance in the total national House vote would merit, and Republicans won fewer.

After 1994 — when the Solid South accelerated its turn to the Republicans — that Democratic advantage went away, as demonstrated in Chart 2.

Chart 2: Comparing total House votes to number of seats won, before and after 1994 Republican Revolution

Source: Vital Statistics on American Politics 2011-2012

In the nine elections from 1994 to 2010, both Republicans and Democrats did slightly better than their overall vote would indicate, but Republicans won about two percentage points more total seats and Democrats won about one percentage point more (these numbers don’t add up to 100 because of rounding and votes for third party candidates).

While we don’t have a big sample size, it seems that the total House vote does tell us something about how many seats a party will hold after the election in this era. To be clear, a 2.2 percentage point lead for the Democrats in the generic ballot at this point is not very meaningful, but if that number inches up a few more points points, that might foretell bigger gains.

Models confirm the conventional wisdom

Last week in the Crystal Ball, we showcased some presidential forecasting models. Taken on average, they forecast a tiny two-party popular vote majority for the president (50.2% for Obama, and 49.8% for Romney). There are models for the House, too.

A model by Crystal Ball contributor Jim Campbell of the University at Buffalo-SUNY shows a Democratic gain of close to 10 seats. Campbell’s forecast is based on the net difference of the number of Democratic and Republican “seats in trouble.” These are seats in substantial jeopardy of being lost by a party. Two forecasts are produced using the “seats in trouble” index, one also taking into account the number of seats Democrats currently hold and a second taking into account the president’s approval rating. The equations, which will be fully explained in the October edition of PS: Political Science & Politics, predict Democrats to gain about eight or nine seats.

Another model, published this week at the political science blog The Monkey Cage, showed Democrats gaining just one seat. It uses a combination of national factors and district-level statistics to makes its forecast (more details are available at the link).

While neither model precludes the possibility of a Democratic takeover, their overall conclusions are pretty similar — modest Democratic gains, but continued Republican control.

Rating changes: Going out on some limbs, and staying out on some others

It’s getting to be moving time in the House, and given that the Crystal Ball will not have any toss-ups by Election Day, we’re going to stick our neck out a little bit in some places, with the caveat that we might have to revisit these races later.

Chart 3: Crystal Ball House ratings changes

Rep. Brian Bilbray’s (R, CA-52) performance in the June primary was weak enough that he has been on watch all summer. While the race appears to be a dead heat at the moment, we think Democratic turnout should lift challenger Scott Peters — so we’re pushing this race in the leans Democratic direction.

We’re also comfortable enough now to upgrade the chances of Rep. Mark Critz (D, PA-12). His Western Pennsylvania district is challenging for Democrats, but this John Murtha protégé proved his mettle in a tough primary race against fellow Democrat Jason Altmire in the spring, and he survived two tough races in 2010. We’re giving the benefit of the doubt to 2010 Democratic survivors like Critz, when prudent.

Rep. David Cicilline (D, RI-1) is lucky to be running in a very Democratic district in a very Democratic state. While he’s not well-liked because of his time as mayor of Providence, Democrats — especially in much of New England — appear to be strongly coalescing around their president and their party. That should be enough to get Cicilline across the finish line, so we’re moving him to leans Democratic. The same thing goes for Rep. John Tierney (D, MA-6), who has suffered from bad publicity over his family’s legal troubles. We’re keeping him in the leans Democratic column even though he, like Cicilline, is a weak incumbent.

In Nevada, we’re tilting NV-3 to Rep. Joe Heck (R), where challenger and state Assembly Speaker John Oceguera (D) has run a widely-panned race. This is not because of a recent SurveyUSA poll that showed Heck up 13 points; that poll seems Republican-leaning and we don’t believe Heck is leading by that much, but nonetheless we are shifting his rating. Heck joins Reps. David Rivera (FL-26), Allen West (FL-18) and Dan Benishek (MI-1) as highly tenuous Republican House favorites in our ratings.

Meanwhile, there are a few other races that we’re moving into the toss-up column. Rep. Mike McIntyre (D, NC-7) is the hardest target for Republicans in the redistricted Tar Heel State. If the GOP falls short of maximizing their gains there (three other Democratic-held seats are shaded in various hues of red in our ratings), it will be because the Blue Dog McIntyre hangs on. Rep. Mike Coffman (R, CO-6), meanwhile, saw his district become much more Democratic in redistricting, and he, like McIntyre, will try to hold on in unfavorable territory; Coffman’s district is a better target than that of Rep. Scott Tipton (R, CO-3), so we’re distinguishing the two by listing Coffman as a toss-up.

In Illinois, we’re sticking with lean/likely Democratic ratings for four Republican-held seats, but of those four vulnerable Republicans (Reps. Judy Biggert, Robert Dold, Bobby Schilling and Joe Walsh), only Walsh seems like a slam dunk for the Democrats. With toss-up IL-12 (held by retiring Democratic Rep. Jerry Costello) and leaning Republican IL-13 (held by retiring Rep. Tim Johnson), there’s plenty of action in the Land of Lincoln, with possible outcomes ranging from a net zero change in seats to Democrats netting five. Expect something closer to the latter (our ratings currently reflect a perhaps-too-rosy net plus-four for Democrats there, assuming they hold IL-12). One person who disagrees with our ratings? None other than Sen. Dick Durbin (D-IL) — he believes Tammy Duckworth (D) will defeat Walsh and that Bill Enyart (D) will hold IL-12, but that the others are “toss-ups.” We may end up agreeing with him as Election Day grows nearer.

Two seats moving off our list are AR-1 and NM-1, which are recent battleground seats in states that are trending blue (NM) and red (AR). Another is Rep. Pat Meehan’s (R, PA-7) seat; in a state where they lost five net seats in 2012, Democrats really only have one viable pickup opportunity: Rep. Mike Fitzpatrick (R, PA-8), who was elected for the first time in 2004, lost in 2006, and regained his seat in 2010.

Several Democratic-leaning seats get bluer in these ratings; one, however, is coming onto the list: National Republican Congressional Committee Executive Director Guy Harrison’s pet project this cycle seems to be defeating Rep. Ed Perlmutter (D, CO-7), who is facing Republican Joe Coors (brother Pete lost a Senate race to Democrat Ken Salazar in 2004). Coors, a member of the brewing family, has tons of money to spend, and Democrats were sufficiently spooked that the House Majority PAC, a SuperPAC that supports Democratic House campaigns, spent $500,000 there on ads, as has labor. Perlmutter remains a heavy favorite, but the race belongs on the list of competitive contests.

Reps. Michele Bachmann (R, MN-6) and Kristi Noem (R, SD-AL) also return to our list of competitive races, but we’d be surprised if either lost. Bachmann has certainly said enough outrageous things over the years to be in trouble, but she occupies the most Republican district in Minnesota. At first blush, saying “the most Republican district in Minnesota” may bring to mind William F. Buckley’s idea of a backhanded compliment (“the tallest building in Wichita, Kansas”), but Bachmann’s district only gave President Obama 43% of the vote in 2008, or 11 percentage points less than the state as a whole. Bachmann will always be a target, but it will take a Herculean effort to unseat her; Democratic businessman Jim Graves is giving it a go this time around, and his internal polling shows him close to the former presidential candidate.


As our House race charts show below, there are dozens of competitive races for the House, and many of them will be difficult to call, even right before Election Day. However, there is little indication that the majority of the closest races — the leaners and the toss-ups — are strongly moving in one direction or the other. A closely contested House race, with no wave building for one side or the other, is by default a good position for the incumbent party. But watch the generic ballot; if races start moving to Democrats en masse, the trend will probably pop up in that number.

Chart 4: Crystal Ball competitive House seat ratings

Notes: Full ratings for all 435 seats are available here. Bold type indicates that the current party that holds the seat is not favored to keep it.