Tag Archives: 2014 midterm

No, That’s Not The Message Voters Sent Yesterday

I’m working on about 3 hours sleep but wanted to give some initial thoughts regarding yesterday’s completely unexpected Senate and House results. At last check Republicans had picked up at least seven seats in the Senate, giving them a 52-43 margin (with two independents caucusing with Democrats.) However, with all the votes in, it appears that Republican challenger Dan Sullivan has ousted incumbent Democrat Mark Begich in Alaska, which would give Republicans 53 seats. Meanwhile, Virginia is headed to a state-mandated recount between incumbent Democrat Mark Warner and Republican challenger Ed Gillespie, and Louisiana will hold a runoff between Republican Bill Cassidy and incumbent Democrat Mary Landrieu in December. I suspect Cassidy will win the runoff, but there’s a good chance Warner survives the recount. So, conceivably Republicans could end up with 54 Senate seats which would be a net gain of nine.

In some ways the House results are equally impressive for the Republicans, given how few races were really up for grabs and the fact that Republicans went into yesterday already holding 234 seats. As of now it appears Republicans have netted 14 seats to push their majority to 243, but with results still pending in more than a dozen races that total could go up to as much as 19 seats gained. One has to go back to the 80th Congress (1947-49) to see a Republican House majority that big.

So, what are we to make of these results? To begin, it’s important to resist the inevitable tendency for pundits to overreach in their effort to discern “the message” the voters send yesterday. Already I am reading that the results indicate 1) a rejection of Obama,  2) a rejection of Democrats’ “war on women”  3) a rejection of Democratic liberal governance or maybe some combination of all of these. Some Democrats, not surprisingly, are suggesting that Republicans “bought” the elections due to backing from Superpacs.

The reality is that while this was a good night for Republicans, the results were driven by midterm election dynamics that political scientists have long documented. In this respect last night’s results were not unusual – nor were they even unexpected, at least based on fundamentals-driven forecasts. The most important point to remember is that the electorate in a midterm is different than what we see in a presidential election year, a point I made repeatedly last night. I haven’t seen turnout figures, but I’m guessing turnout was about 40%, down about 18% from 2012’s presidential election. More important than the size of the turnout, however, is its composition: yesterday it skewed older, whiter and more affluent than the electorate of 2012, and these are all attributes associated with a greater propensity to vote Republican.

More generally, the President’s party almost always loses House and Senate seats in a midterm – this is as close to a covering law that we have in political science. The magnitude of last night’s House losses by Democrats were surely attenuated somewhat by the fact that Republicans controlled so many seats, but nonetheless a net gain of 14-19 House seats by Republicans is well within the norm for a midterm election. On average, the president’s party loses about 28 seats in these midterms during the post-World War II era.

In the Senate Republicans did better, but not unusually so based on the fundamentals. As this chart indicates, political scientists who forecast the Senate race thought Republicans would pick up 8 seats based on the state of the economy, Obama’s approval ratings, and the prevailing view among most voters that America was headed on the wrong track.

Yes, this election was in part a referendum on Obama, but exits polls indicate that fully 45% of voters didn’t factor Obama’s performance into their vote at all, while 19% said their vote was meant to express support for him, so this can’t be viewed as a wholesale rejection of his presidency. More generally, when the economy is weak, the president suffers from low approval ratings and people are generally dissatisfied with the state of the nation, we should not be surprised that a Republican-oriented electorate dumped on members of a Democratic president’s party in Congress.  Indeed, the greater surprise would have been if Democrats somehow held onto their Senate majority in the face of these fundamentals.

Of course, elections have consequences, and yesterday is no exception. To me, the most important is that these results are not likely to reduce polarization in Congress. Consider the Senate Democrats who were turned out last night. David Pryor was the second most conservative Democrat in the Senate, Mary Landrieu (who may yet hang on) the third most, Kay Hagan fourth and Mark Begich 12th among the 55 Democratic/Independents Senators. It is almost certainly the case that the Republicans replacing them are not going to be more moderate, although I confess to not knowing enough about them to place them on an ideological scale with any great degree of confidence. Still, I’m fairly confident polarization is not likely to decrease during the next two years.

In looking ahead, my guess is that the Senate will become more unruly – not less so – during the next two years. This is partly because the Democratic caucus has shifted left with the loss of its more moderate members. But it is also because several conservative Senate Republicans – with at least one eye on a potential 2016 presidential run – will view this election as an opportunity to push conservative policies designed to appeal to the party base. As David Mayhew reminds, for legislators the payoff is more often in the position taken than in the legislative results. Similarly, I see no reason why Obama is going change his ideological leanings as a result of last night’s “shellacking” redux. Presidents – like any politician – are not infinitely malleable when it comes to ideology. They have core beliefs that guide their conduct, although in Obama’s case those beliefs sometimes appear frustratingly opaque. Rather than do a governing about face, Obama is likely going to accommodate Republicans when he can, but otherwise wield that veto threat to block Republican initiatives, just as Gerald Ford, the two Bushes and Bill Clinton did when they faced an opposition-controlled Congress. We may see legislation passed, but only where both parties see it in their own interest. In short, last night’s election is not likely to have affected the strategic calculus that has governed relations between Obama and Republicans to this date.

A final thought. As the table below indicates, we have cycled through almost every possible configuration of partisan control of our national governing institutions – a period of instability that testifies to the public’s apparent unwillingness to give governing power to one party or the other for any significant amount of time.

Picture2As long as individual Senators and Representatives believe that their electoral fortunes rest in part on the popularity of their party’s “brand name” among voters, and as long as the parties’ governing coalitions appear evenly matched even as they grow increasingly polarized in views, it remains the case that each side will usually conclude that it is not in their interest to compromise. And so we appeared destined to cycle through still another governing configuration.

Next up: the 2016 elections. Let the campaigns begin!

The Election Forecast: It’s The Fundamentals, Stupid!

Political scientist Lynn Vavreck posted an interesting New York Times column a couple days back in which she noted that midterm elections – more so than presidential elections – are referendums on the president and his party. As she writes, “Instead of rewarding or punishing the incumbent president for his handling of the nation’s economy, in midterm years voters address the president more directly — by penalizing his party members, on average, but also by calibrating that punishment based on how the president is doing his job. Average approval ratings of the way the president is ‘handling the job’ explain more of the variation in seat loss than the economic indicators.”

What Vavreck is referring to is the well-known “midterm” loss phenomenon which I have written about on multiple occasions. Briefly, as political scientists have documented, the president’s party on average loses about 28 House seats and 4 Senate seats in midterm elections during the post-World War II era. There are several explanations for why this is the case, but Vavreck cites one – the “surge and decline” thesis that points to differences in the size and composition of midterm elections versus presidential elections as a primary reason why the president’s party does less well in off-year elections. Strictly speaking, of course, the surge-and-decline thesis is not necessarily a referendum on the president’s performance so much as a fundamental aspect of turnout differences. Historically, as this chart indicates, midterm turnout in recent years hovers at about 40% – these are the habitual voters who turn out every election, whereas in presidential elections we see more variation and usually higher turnout. (This data is from Michael McDonald’s site.)

 

In 2010, turnout was about 41% of eligible voters, but it was about 58% in the 2012 presidential election.

But, it is also the case that beyond simple turnout factors, midterms are also directly referendums on a president’s performance. From this perspective, how one votes in the midterm is a direct reflection of a voter’s attitudes toward the President. As I’ve discussed in previous posts, congressional candidates – as strategic actors – are very much aware of when the President is a liability. Under those conditions, they have an understandable desire to keep him at arm’s length, as we have seen in “purple” state Senate races in Colorado, Iowa and New Hampshire this fall.

In their forecast models, political scientists try to account for both the surge-and-decline and the referendum factors by including a midterm variable, but also a measure of the president’s approval ratings. However, we do not want to overstate the degree to which the midterm is all about the President. As this table from Pew shows, even during the current election cycle, only about 32% of voters see their midterm vote as driven by opposition to Obama, while 20% say it is a vote in favor of Obama. That means 45% of those surveyed say their midterm vote has nothing to do with the President! And, as the chart indicates – that’s not unusual; many people in past midterms say their vote is not a referendum on the president’s performance.

So, if not just the president – what is the midterm about? One factor is a general assessment of how well the country is “doing”, captured in questions asking whether the country is on the right track or not. Here we see that most people think the country is currently on the wrong track.

right track

Another influence is the relative preference for one party or the other, often measured in the generic ballot question.

generic ballot

On this measure Republicans also have a slight advantage. Finally, the economy does matter, even in midterm elections, and some measure of economic conditions is often included in fundamentals-based forecast models. For example, Michael Lewis-Beck and Charles Tien include a measure for changes in disposable income in their forecast model.

When we incorporate all these variables into a “fundamentals”-based forecast for today’s midterm elections, what do we get? Middlebury student Day Robins put together this chart summarizing the predictions made by several political science forecast models. (Note that some models only forecast House results, some only Senate, and some do both.  Note also that most forecasts do not estimate a probability of a party taking over a chamber.)Picture1
Note that these models are not driven by the same logic that drive predictions based on poll aggregation, such as those by Drew Linzer at the Daily Kos or Mark Blumenthal’s Huffington Post. Poll-driven models are likely to be very accurate – maybe more accurate than fundamentals-only polls – but they don’t tell us why the polls turn out the way they do. For that, we need to turn to political scientists. And, as Robins’ chart makes clear, the fundamentals – the fact that this is a midterm, the President is unpopular, Democrats have more exposed seats, the economy is sluggish and generally more people think the country is on the wrong track – does not bode well for Democrats in either the House or the Senate. How badly will they do? On average, the models indicate Democrats will lose 12 House seats, with the median figure 14. In the Senate, the models say Democrats will lose on average 7 seats, dropping them to 48, with the median at 8 seats. Both forecasts sound right to me. To make it official, I’ll go with a Republican pickup of 8 Senate seats and 14 House seats. That will give you all something to root against tonight!

No turnout machine, no matter how sophisticated, can make much of a dent when the fundamentals mean you are moving electorally against a strong partisan headwind.

Keep in mind that, as the table above shows, as midterms elections have become increasingly nationalized, the partisan tides affect all candidates to a much greater degree. And so I expect it to go today.  This is likely to be a predominantly nationalized election, one in which, taken collectively, the fundamentals favor Republicans more so than Democrats.

I’ll be live blogging from the “Karl Rove” Crossroads Grille tonight, while simultaneously trying to keep the Middlebury students from crying in their beer. I hope you can join in – it’s been a while since I’ve been able to do a live blog.

See you at 7:30 – by then they may have already called the Vermont races!