Will 2014 Midterms Be a Wave Election?

By Sean Trende - August 28, 2014

‹‹Previous Page |1 | 2 |


As a final way to get at this, I constructed an index that rewards parties for picking up seats and defeating incumbents, and penalizes them for losing incumbents and failing to take advantage of a favorable map by defeating senators in states with favorable partisan leans. You can see the results in the following chart:

Chart 1: Index of Magnitude of Senate Losses

The 2014 result is a weak wave, but it looks a lot more like waves than the non-waves. (There are wave elections that are more toward the bottom of the chart, but they all involved a substantial House victory for the prevailing party, something that isn’t an issue this year.) Moreover, the six-seat cutoff is meaningful. If the GOP picked up five seats and defeated two incumbents, its index would be just a half-point higher than the non-wave year of 1950.

So to a certain degree, I think the question of why we don’t see a wave today involves a false premise: We see evidence of one right now. I suspect our expectations for a 2014 wave were simply set too high during the Obamacare rollout.

But I do think there is a substantial case to be made for Republican gains beyond seven seats. To begin, the RCP averages currently show Republicans ahead in seven races. Democratic incumbents are well below 50 percent in states like Alaska (46 percent) and Colorado (46.3 percent), and there are still a huge number of undecided voters in Michigan.

That last point is the key one, I think. I’m not a big believer in the “undecided rule,” which postulates that undecided voters break against the incumbent. This has been shaky at the presidential level for years, and appears to be increasingly shaky at the sub-presidential level.

But I’ve noted before that over the past few election cycles, Senate races have tended to follow presidential approval ratings rather closely. In fact, you can predict the 2006 races based upon presidential approval, incumbency, and candidate quality rather well using similar data from the 2010 and 2012 elections. I noted that it works less well in races with unusually popular incumbents (such as Joe Manchin and Olympia Snowe) and in Greater Appalachia, where President Obama’s job approval is below the “normal” Democratic performance in the state.

I took the recent polling in the competitive Senate races where both horse race numbers and job approval numbers were included. Not every race is included, because not every race has polls with both presidential job approval and horse race numbers. I plugged those numbers back into the regression equation we derived back in the winter, and compared the projected Democratic performance with the actual Democratic performance:

Table 3: Projected Democratic Performance vs. Actual Democratic Performance

Things are actually falling about how we’d expect them to fall, given the president’s job approval. Democrats are performing unusually well in the Greater Appalachia state of Kentucky, and Jeanne Shaheen is running unusually well for an incumbent (though the most recent WMUR/UNH poll gives some pause there). But beyond those races, Democrats are performing more or less as the model suggests they should: About two points better, on average. Remember, the model doesn’t project two-party vote share, just raw performance for the president’s party -- to account for this, my cutoff for victory in my model was reduced to 48 percent. The fact that Democrats are overperforming the model should offer Democrats some comfort, but it could also be the case that these races just haven't fully engaged yet, and that Democrats are benefitting from an opponent that isn't yet fully defined.

The upshot of this is that in many of these states, Democrats have mostly exhausted those voters who approve of the job the president is doing. This leaves Democrats with an undecided electorate that overwhelmingly disapproves of the president. As just one example of this: Democratic polling firm PPP polled Arkansas and found that the president’s job approval was 13 percent among undecided voters. Add in the fact that many of these pollsters are still surveying registered voters, and it draws a picture for Democrats that isn’t encouraging.

Note that if the presidential job approval theory of Senate elections holds true, this could help explain our initial observation as to why waves tend to break late. As of August, both parties have mostly nailed down their bases, along with Independent “leaners.” In normal years, the undecideds will be more evenly divided, and the races won’t break heavily one way or the other. But in wave years, the undecided voters will tend to be overwhelmingly composed of voters who disapprove of the president. The late movement isn’t really movement at all; it is unengaged voters who are previously disenchanted with the president tuning in and making up their minds. (Note also that PPP found that undecided voters in North Carolina tended to approve of the president. This could explain why Hagan is underperforming expectations right now: She hasn’t nailed down her base.)

Before we turn to the arguments against the GOP wave, it is worth noting one other thing. The Democrats’ drop-off problem -- where Democratic candidates in special elections underperform President Obama -- hasn’t gotten any better. In fact, it has gotten worse. You can see this in the following chart.

Chart 2: Democratic Performance in Special Elections vs. Obama 2012 Vote Share

The black line -- we’ll call it the “break even” line -- represents what we’d see if Democrats matched the president’s 2012 performance in special elections. A candidate above the line over-performed the president; a candidate below the line under-performed the president. As you can see, the trend line for 2013 is below the break-even line, representing a drop-off of about six points. The 2014 trendline is below the 2013 trendline, representing a drop-off of an additional two points.

While I seriously doubt that this trend will continue full force into the midterm elections, even a moderate Democratic drop-off would lead to a GOP wave. For what it is worth, the median drop-off among Democrats in our sample above right now is about three points; if you drop Kentucky, it is four points.

The counter-arguments are very real, if not overwhelmingly compelling. For example, it is true that there is no Democratic equivalent this cycle to Rick Santorum or Blanche Lincoln. But I’m not sure how relevant that is. In 2010, Lincoln was the only incumbent Democrat who seemed beyond repair at this point. Would we conclude that 2006 wasn’t a wave if she had opted to retire, rather than to lose badly? The light-blue states in 2010 looked as if they were headed toward tight races, much as they do today. The improving economy and deficit should help Democrats, but the president’s job approval is nevertheless below where it was in 2010.

The unpopularity of the Republican Party is probably the best reason to suspect a wave might not arrive. In particular, voters who dislike both the president and the Republican Party might opt to vote Libertarian in unusually large numbers, allowing a Democrat to win with about 46 or 47 percent of the vote. This probably happened to some extent in Montana, Indiana and Missouri in 2012. Or, they could stay home, as they also probably did in 2012 (for this, look to see if the president’s job approval in the exit polls is significantly above what the public polls suggest).

Perhaps this will make the difference between a GOP wave and a ripple. But we should also remember that we saw this exact same argument in 2010 (as well as an argument that the growing non-white population would prevent a 1994 repeat). The GOP’s position has deteriorated further since then, to the point where its core voters might not be as enthusiastic as they were in 2010. So this factor might still matter, but the truth is, we don’t have a whole lot of empirical evidence to suggest that it will.

As I said, you can evaluate these arguments for yourself and reasonably come to opposite conclusions. The job approval model might not hold in 2014, the GOP’s unpopularity might have finally dropped to the point where it is seriously weighted down, and (perhaps most intriguingly) the Democrats’ Bannock Street Project to boost base turnout could bear fruit. Regardless, the fact that we haven’t seen a massive wave yet isn’t particularly interesting. If a wave does materialize, we won’t have to look much further than evidence that is right in front of our noses today to explain why. 

‹‹Previous Page |1 | 2 |

Sean Trende is senior elections analyst for RealClearPolitics. He is a co-author of the 2014 Almanac of American Politics and author of The Lost Majority. He can be reached at Follow him on Twitter @SeanTrende.

A Reply to Rush Limbaugh
Mona Charen · December 9, 2014
Against the Wave: Three Winning House Democrats
David Byler · December 14, 2014
Obama, Congressional Dems Show Cracks in Unity
Jim Kuhnhenn · December 7, 2014
Democratic Louisiana Senator Mary Landrieu Defeated
Melinda Deslatte and Bill Barrow · December 6, 2014

Sean Trende

Author Archive

Follow Real Clear Politics

Latest On Twitter