Tag Archives: 2012 presidential election

The State of The Race: It Was The Best Of Times. It Was The Worst Of Times.

Today saw two separate and seemingly contrasting assessments of the state of the 2012 presidential election.  First, CNN released its latest presidential poll, and it showed Barack Obama leading Mitt Romney among a sample of registered voters by 52%-47%.  Later, Fox News released its own survey showing Obama up among registered voters by 9%, 49%-40%, over Romney.  At the same time as these generally favorable polls for Obama were released, political scientist James Campbell, in this guest column at Larry Sabato’s Crystal Ball, weighed in with his own assessment of the election that painted a distinctly different picture.   Campbell analyzed two sets of numbers: second quarter GDP numbers in the election year, and GDP numbers for the last 10 fiscal quarters prior to the election in presidencies starting back to 1956 through the Obama administration, and correlated them with elections results.  He found that by both measures the economy under Obama ranks near the bottom of the pack – a finding that, if history holds, does not bode well for Obama’s reelection chances.  As Campbell writes, “If the election becomes largely a referendum on President Obama’s economic record, it is hard to see how he can win. As I have been setting it out in this essay, the record really speaks for itself. It is dismal, and this is without noting the 42 consecutive months and counting of an unemployment rate over 8%.”

How do we reconcile polls results such as those contained in the CNN and Fox surveys that show Obama ahead with the much more pessimistic forecasts by political scientists, such as Doug Hibbs, Campbell and even Alan Abramowitz, based on economic fundamentals?  Bert Johnson and I touched on this topic in our latest professor pundits’ video, but I want to elaborate some of our comments here. In a scholarly article written  in 1993, Gary King and Andrew Gelman examined why trial heat polls like CNN’s and Fox’s can be so volatile in the months preceding a campaign despite political scientists’ claim that election results are largely dictated by fundamental factors that do not change appreciably during the campaign season.   The answer, they suggest, is that many voters who are surveyed several months before an election often really haven’t given much thought to who they will vote for, even though they respond to the question sincerely.   That is, these survey respondents over estimate their own certitude regarding for whom they are likely to vote.  But their answers this far out are often driven by short-term events, such as media coverage of a candidate’s “gaffe”, or more general perceptions regarding who is “ahead” in the race.  These short-term factors, however, are not the ones that ultimately drive the decision for most individual voters.  Instead, that decision will be rooted in the fundamentals, such as economic growth as measured by GDP, as interpreted through voters’ own partisan and ideological lens.

Although their research is dated, I don’t think King and Gelman’s conclusion are any less applicable today, which is why I caution against paying much attention to trial heat polls such as CNN’s and Fox’s at this stage of the campaign. This is even more so for state- level trial heat polls, which is why I am skeptical of efforts to predict the Electoral College vote this far out, particularly since many pollsters’ likely voter screens at the state level aren’t particularly accurate as yet.   More generally, I don’t believe that the relative electoral fortunes of Obama and Romney are fluctuating as much as many voting models based on polling data suggest.

Now, as I have said many times before, when we get closer to the election these polls will become increasingly accurate and they should (fingers crossed!) converge with what the forecast models based on the fundamentals tell us should happen.   Indeed, by the final weeks of the campaign, the polls will be much more precise than our fundamental-based forecast models.   As of now, however, the fundamentals cast a gloomier outlook on the election for Obama than do today’s trial heat polls.  How gloomy? Here, according to Campbell, is where Obama’s 2nd quarter GDP numbers rank among presidents dating back to 1952:

As you can see, only Carter had a worse election year second quarter.  The story is not much better if we look at the average GDP growth for the last 10 quarters; here the GDP figures for Obama are better only than George H.W. Bush’s and the Nixon-Ford quarters.

Note that Campbell doesn’t actual present a forecast in this article.  In past years, however, he has presented a forecast model in early September based on GDP 2nd quarter growth and Gallup Poll trial heat polls from Labor Day, and I expect he will do so again in a matter of weeks. In previous elections this model has done quite well predicting the winner’s share of the two-party vote. Like all forecast models, however, it is not foolproof; in 2008 Campbell was the only political scientist to my knowledge who forecast a McCain victory (with 52.7% of the two-party popular vote).  Ironically, Campbell blames his failure to forecast Obama’s victory on one of those “black swan” events that political scientists always say can throw a forecast off, but which never seem to appear.  In 2008, however, if Campbell is to be believed, it did appear in the form of the Wall St. meltdown. As Campbell wrote in his election post-mortem (gated): “Unlike any election in modern history, the 2008 campaign had been derailed by an entirely unanticipated and unpredicted external event.”    But, given this unprecedented fiscal meltdown, why did the other forecast models successfully predict Obama’s victory and come much closer to estimating his share of the two-party vote?  Campbell says, in essence, that they were lucky:  “Under the circumstances, I think that whether a forecast was close to or distant from the vote this year is largely a matter of good or bad luck.  Nothing in any of the models would have either directly or indirectly anticipated the effects of the timing of the Wall St. meltdown.  Nothing.”

Not all forecasters would agree.  As I noted in an this earlier post discussing Doug Hibbs’ forecast model, Hibbs has no use for models like Campbell’s that incorporate trial heat polls or presidential approval ratings because they are essentially adding “noise” the forecast, since both approval and trial heat results are themselves predicated, at least in part, on the economic fundamentals.

So where does that leave the race?  In my view, the fundamentals that have generally proved reliable in the past suggest that come November, this race will be much closer than the margin suggested by today’s CNN and Fox polls.  Although you will undoubtedly hear pundits proclaiming in the wake of these latest polls that we are seeing a potential turning point in the race, with Obama beginning to stretch his lead, the King and Gelman study reminds us that this type of polling volatility is par for the course and likely does not provide a very reliable preview of November’s outcome.

The 2012 Presidential Election Is Most Like This Previous Election

One of the recurring exercises in any presidential election year is the effort by pundits to find the most appropriate historical analogue among previous presidential elections.  What is deemed “appropriate”, however, is usually determined by whether the historical antecedent favors your preferred candidate.  This election cycle has been no exception.  Democrats see strong parallels between the current race and the 2004 contest between John Kerry and George W. Bush.  Although Kerry actually led in some early polls, the election remained quite close until Bush, the first-term incumbent running for reelection, pulled it out with slightly more than 51% of the two-party popular vote. Not surprisingly, Romney supporters don’t believe 2004 has much to say about this election. Instead, they point to 1980, when Ronald Reagan trailed Jimmy Carter in the Gallup Poll for much of the post-Labor Day period only to surge ahead in the closing week, as the most fitting historical parallel.

In this vein, Greg Sargent, writing in yesterday’s Plum Line column for the Washington Post, took issue with the 1980 comparison for several reasons.  The first has to do with the difference in candidate qualities; Sargent writes:   “Reason one: Obama is a better and more likable politician than Jimmy Carter was, and Romney has not proven himself to be Ronald Reagan.”  As evidence, Sargent quotes long-time GOP strategist Ed Rollins who notes that Romney is no Reagan: “There’s no question that on his best day, he’s not a Ronald Reagan…. Traditionally incumbents don’t do as well in debates as challengers for the simple reason that challengers have to stand on the stage and look like an equal. Romney can do that, but Obama is good. He’s likeable. Carter was never likeable.”

If asked, I think many pundits would agree with Rollins’ characterization of Carter; he was the moralizing, preachy politician who – with a smugness bordering on arrogance – castigated Americans for their “crisis of confidence”.   It was small wonder that he was viewed as so much less likable when compared to Reagan – the eternal optimist who spouted homilies about America as the shining “city on a hill”.  Not surprisingly, Reagan pulled ahead after their one and only debate in late October, largely because voters had a stronger personal affinity with  Reagan.  The lesson?  Reagan’s sunny optimism trumped Carter’s dour moralizing.

Alas, Rollins has the story completely backward, as Mo Fiorina reminded us earlier in this this New York Times editorial which I’ve discussed before.    Rather than being disliked, on a personal level Carter was held in far higher regard than Reagan in the eyes of most voters – indeed, despite his low job approval, Carter had the highest-rated personal qualities of any Democratic candidate in the period from 1952-2000, and Reagan had the lowest of any Republican presidential candidate. As Fiorina summarizes: “The Ronald Reagan of October 1980 was not the Reagan of “morning again in America” in 1984, let alone the beloved focus of national mourning in 2004. Many Americans saw the 1980 Reagan as uninformed, reckless, and given to gaffes and wild claims. But despite their misgivings about Reagan, and their view that Carter was a peach of a guy personally, voters opted against four more years of Carter.”

As is often the case, the prevailing view of Carter as unlikeable likely reflects a post-hoc rationalization for why he lost the election, as well as eight years of a largely successful Reagan presidency.  But that’s not how the two candidates were viewed in the run-up to the 1980 election – something Rollins evidently forgets.

But perhaps there’s a second good reason why this year isn’t like 1980? Sargent, again citing Rollins, argues: “Reason two: The electorate is far more polarized now. Rollins notes that a last-minute shift was enabled by the larger role Dem swing voters played at the time. ‘There was a big swing element in the Democratic Party — blue collar Democrats,’ Rollins noted. ‘It’s smaller now.’

Again, I think this analysis is at least in part wrong; as I’ve argued in many previous posts (see here, for example) there’s not much evidence that the electorate is more polarized now than in previous elections.  What has changed is that the choices have become more polarizing.  This may make it less likely that voters of one party will consider voting for the opposing party’s candidate.  But it does not mean that there are not enough undecided voters left to preclude a small shift toward Romney in the closing weeks of this race – and in a tight race, a small shift will be enough.

Is that likely to happen?  I don’t know.  But if it does, it almost surely will have little to do with which candidate is more likeable and instead will turn on more deep-seated voter concerns regarding which candidate is better able to resurrect a moribund economy.  In this respect, 2012 has a lot in common with 1980 – and with 2004 – and with most of the presidential elections in the post-World War II era.

3:45 p.m. I’ve added a link to the Gallup data showing Carter leading Reagan during most of the post-Labor Day general election campaign until Reagan surges near the end.  Note that the race was quite tight for most of this period, according to Gallup.

An Electoral Landslide For Obama?

Michael Tomasky wrote a provocative online piece yesterday in the Daily Beast in which he speculated that Obama may, in fact, be on the verge of winning an electoral landslide.  Tomasky wrote, “The secret is the electoral college, and the fact is that the more you look at it, the more you come to conclude that Mitt Romney has to draw an inside straight like you’ve never ever seen in a movie to win this thing…. the paths to 270 are few.”   In looking at the key battleground states, Tomasky concludes that, given current polling, it is very unlikely that Romney will win enough of them to secure an Electoral College majority.  “In other words, Obama can lose the big Eastern four—Ohio, Virginia, North Carolina, and Florida: all of ’em!—and still be reelected. And barring some huge cataclysm, he’s not losing all four of those states. If he wins even one—say Virginia, the smallest of the four—then Romney has to win Colorado, Iowa, and New Hampshire; all possible, certainly, but all states where he has been behind, narrowly but consistently, for weeks or months.”

At first read, Tomasky’s logic seems persuasive.  After all, Romney might squeak out a victory in one or two of the battleground states.  But is it realistic to expect him to win the “big four”  – “all of ‘em!” – and the additional battleground states he needs to claim victory?  Consider the state of the Electoral College map right now.  Obama is likely starting from a baseline of some 179 electoral votes, compared to 131 for Romney.  If we add the “leaning” states to each candidates’ column, Obama moves to 247, while Romney is only at 191.  That leaves 100 electoral votes across eight states still in play.  Let’s say that Tomasky is right and that Obama is not going to lose all of the eastern  “Big Four”.   Since Obama is up almost 5 points in state polling in Ohio,  let’s assume Obama will win that, putting him at 265 electoral votes, only five short of the majority he needs to win. That would mean that of the remaining  seven battleground states, Romney would need to win six: a seemingly daunting task.

The problem with this type of analysis is that it implicitly treats the outcome in each state as an independent event.   But they are not independent;  the factors that influence how well  Romney does in Florida – say, voters’ perception of the national economy – will also affect his performance in the other battleground states.  So if in the last weeks the undecideds break his way in one battleground state, they are likely to do so in all of them.  And it won’t take a “cataclysm” to push Romney over the top – he’s within 3% in six of the eight battleground states based on the RealClearPolitics aggregate polling right now. This is not to say that local factors don’t matter at all – they do.  And in a close election, they could be decisive.  But national factors also come into play here, and that means it is not as improbable as you might think from reading Tomasky’s analysis that one candidate might end up sweeping almost all the closely contested races.   Put another way,  if Romney wins the national popular vote, it is likely he’s going to win enough of the battleground states to claim victory in the Electoral College as well.  The same holds, however, for Obama: if most of the remaining undecideds decide he deserves  more time to right the economy, he might very well coast to an Electoral College victory.  But he will likely do so primarily on the basis of a national tide – not local currents.

Are Voters Rational? Bain, Three-Legged Jackasses, and Viewer Mail

Another day, another poll indicating that the devastating Bain attack ad and the “go for the jugular” targeting of Romney’s tax returns  are, as yet, still not resonating nationally with likely voters. According to a USA Today/Gallup poll in the field from last Thursday through Sunday,  “by more than 2-1, 63%-29%, those surveyed say Romney’s background in business, including his tenure at the private equity firm Bain Capital, would cause him to make good decisions, not bad ones, in dealing with the nation’s economic problems over the next four years.”  The persistent polling results belie the oft-tweeted claim that the Obama ad campaign has “swift boated” Romney by turning his supposed strength into a weakness, much as the Bush campaign allegedly did to John Kerry’s Vietnam record in 2004.   It may be, however, that the ad campaign is resonating more strongly in specific swing states, like Ohio where Obama has been outspending Romney so far. Even here, however, the evidence is far from conclusive; the Pollster.com trend line for Ohio polls has Romney trailing Obama by 2.5%, 47.2%-44.7%.  On June 5th, just before the Bain attacks, Obama led Romney there by 3.3%.

In light of the evident failure of the Bain attacks to register so far, several readers have posted comments inquiring about the relative impact of “campaign effects” and election “fundamentals” more generally.  In that vein, Will asks, “So what, if anything, SHOULD I pay attention to that doesn’t have to do with the economy? If all that matters is the economy, a force which remains outside the control of either presidential candidate, would you say campaigns themselves are essentially useless?”

Let me be clear.  When it comes to influencing the presidential vote, campaigns are not “useless”.  In fact they do matter – but not in the way that journalists and partisan pundits often assume.   To begin, they are important for rallying the base and getting out the vote more generally.  One way they do this is by framing real world events in a way that appeals to their partisans – “Romney’s vulture capitalism contributed to the financial collapse” – but which also might convince independents to vote for a particular candidate.  Now, these sound like very important effects, just as many reporters would have us believe.  Indeed, as a counterfactual, let us imagine a world, as Carlisle Rainey contemplates, in which only one side bothers campaigning.  What, as JTTX asks me in a twitter comment, would our forecast models say about the relative impact of campaigns then?  The answer, I think, is that the side doing the campaigning would have an advantage (although perhaps not as large as one might think!)

But I don’t believe that counterfactual  is the correct way to think about campaign effects and, like John Sides, I  don’t think most reporters think this either.  The reality is that most journalists are interested in the net impact of campaigns in the real world, not a hypothetical one. However, because they cover campaigns on a daily basis, they tend to think those effects are quite large.  Most political scientists disagree.  They believe that in the world in which we live, campaign effects, such as the Bain ad or Romney’s tax returns, are much more limited for at least three reasons.

First, of course, both sides are struggling to frame the election in ways most favorable to their candidate.   So voters aren’t just hearing about Bain – they are hearing about unemployment on Obama’s watch as well.  There are dueling messages out there that voters hear and respond to.

Second, by the time campaigns are in full swing, most voters have already made up their minds based on longstanding predispositions guided by, for instance, an affinity to one party or the other.  Current polling suggests that up to 90% of voters are already committed to either Romney or Obama.  Many of the remaining undecideds are, at this point, not paying much attention to the campaign at all.  Given this, it’s no wonder Bain hasn’t moved the polls.

Third – and I can’t emphasize this enough – reality in the form of those “fundamentals” is the grist for candidates’ campaign mills.  Why is Obama targeting Romney’s record at Bain rather than, say, his foreign policy comments?  Why is Romney touting his business credentials?  It’s because the fundamental issue driving this race is the economy.   You can’t simply substitute a “virtual reality” through the skilled use of clever campaign visuals when voters have independent sources of information by which to assess what the ads are saying.

This brings me to Avery’s query. In taking issue with my notion of “collective rationality”, he notes:  “It was interesting to me that one of the key assumptions of the model you were propounding is that people sort of “fact check” political ads against reality. Given what I know about people, and what I’ve been learning in reading about behavioral econ and political psychology, it seems like people are often pretty bad at fact checking and very good at cherry-picking, especially when it comes to things they already have a well-formed opinion on like “Democrats are better than Republicans” or vice versa.”  Avery goes on to suggest that in a world populated by people exhibiting these cognitive tendencies, “collective rationality” is a misnomer.  Instead we should see the cognitive blind leading the cognitive blind. (Think Rush’s “ditto heads”, or Kos’ “Kossacks”.)

Avery is right, of course, in that individuals’ typical cognitive processes often cause them to engage in selective exposure to campaign ads in a way designed to reinforce existing attitudes.  (As evidence, see almost any partisan blog!)  That doesn’t sound like the world I describe consisting of rational voters who objectively sift through the campaign ads to choose the “best” candidate.

But I am making a less heroic assumption regarding voters’ rationality.  By collective rationality, all I ask is that most voters can distinguish one candidate from the other ideologically, and choose the one closest to their own political views.  And most studies suggest voters are very good at doing this, particularly in an information-rich environment that is a presidential campaign.  So, Joe Sixpack may not understand the intricacies of off-shore tax shelters, and he may stubbornly insist that Obama is not a U.S. citizen, despite overwhelming evidence to the contrary.  But generally speaking, he knows that Obama is a stronger supporter of an activist government, greater business regulation and higher taxes on the wealthy than is Romney.  He knows this from talking with others on Facebook (ok, that would be Joe Brandy Snifter), hanging out at the ball field and maybe even glancing at the evening news – that’s where the “collective” process comes in.   And in deciding how to cast his vote, after absorbing this information, Joe Sixpack chooses – in the context of an economic downturn – which set of views is closest to his.   I  think most voters do this quite well.

The reason why the best forecast models are so accurate is that they essentially distill this process down into a few key variables that measure “reality” – the state of the economy, perhaps presidential approval, whether the nation is at war – and assume the rest of it – the campaign ads, etc. – play off this reality.  Now, as Stuart correctly notes (and as I have said many times before), when reality does not clearly differentiate one candidate from the other, any of these other factors – debates, the V.P. choice, – even the Bain ad! – can conceivably make a difference.   Journalists, who must file a story every day, have to focus on something, and so they emphasize this aspect of the horse race, as viewed through the daily prism of campaign stops, messaging, gaffes, etc.   But if you insist on accepting the horse race metaphor, then at least realize this:  the candidates aren’t always riding equally talented horses.  Sometimes one candidate starts on Secretariat, while the other is riding a three-legged jackass.

Even then, however, it is possible something dramatic might happen.  That’s why we run the race!

Meanwhile, keep  those comments coming!

Misery Loves Company: Incumbents, Inflation and Unemployment

I want to follow my last post regarding unemployment and reelection by discussing a few other economic measures that are often referenced as useful indicators regarding a president’s reelection chances.  One of the most frequently cited is the “misery index”, which is simply a combination of the unemployment and inflation rates.  Anna Esten has put together some more charts documenting the relationship between the misery index and the electoral fortunes of incumbent presidents dating back to Truman in 1948.  The following chart shows the misery index for each incumbent president in the October before the presidential election.  Red indicates the incumbent lost his bid for reelection.

Not surprisingly, incumbents (Ford, Carter and Bush I) lost in three of the four highest misery index years.   The exception is Reagan in 1984, but the explanation for his win becomes clear when you look at the index level four years previous, when he beat Carter.  Carter lost when the October misery index was a horrendous 20.27, the highest any post-World War II incumbent faced in the month before election.  In the intervening four years, however, it came down nearly 9 points, and voters rewarded Reagan for that drop.

Reagan excepted, it appears that incumbents are in dangerous waters when the October misery index is hovering near double figures in an election year.  Where does this put Obama?  As of this past May, the misery index, driven mostly by high unemployment,   was at 12.7 – clearly dangerous territory.  But, of course, we are a long way from October, 2012.  Moreover, one might be tempted to argue, citing the Reagan exception, that the trend in the index come next October is more important than the actual number.  However, the historical record does not necessarily support this. As the following chart shows, the misery index was going down in the third quarter leading up to the elections in 1980 and 1992, but not enough to prevent Carter and George H. W. Bush from losing their reelection bids. Truman’s experience in 1948 suggests that presidents need to see a significant drop in the index to benefit.  If you start in double figures, and only begin to come down as the election draws nigh, it may be too little too late.

Of course, inflation has not been a major concern during Obama’s presidency – as yet.  So we might think that it is the trend in the unemployment rate that will be most crucial come November 2012.  But again, history does not provide Obama much solace, as this chart showing third quarter changes in unemployment rates (actually, it’s the change from July’s rate to October’s) indicates.

As we can see, a drop in unemployment was not enough to save Carter, Bush I or Ford.  To be sure, Carter and Ford were combating “stagflation” – high  unemployment and inflation.  Jobs alone couldn’t save them.  Truman, on the other hand, presumably survived the jump in the jobless rate in 1948 because it was combined with a steep drop in prices, which sent the overall misery index down. But Bush lost mainly because of concerns about jobs, even though the job picture was actually improving.  It was Clinton, however, who benefited.

The more general conclusion is that perceptions about the economy seem to lag reality.  That means the window of opportunity for changing voters’ attitudes regarding Obama and the unemployment rate may be shorter than we think.

Before we write off Obama’s reelection chances, however, the obvious caveats remain:  First, it’s July 2011 – not July 2012.  Second, we are projecting results based on 10 data points whose historical relevance can be debated.  I hope I made clear in the last post the great uncertainty surrounding any effort to extrapolate from such a small data set.  On the other hand, we shouldn’t blind ourselves to the political reality of running for reelection when unemployment is among historically high post-war levels.  Simply put, the economic fundamentals mean Obama is facing a difficult reelection challenge no matter which of the current Republican candidates he faces.

I’ve focused on very basic economic indicators here.  Nonetheless,  they do provide a good shorthand assessment of the fundamentals that drive elections.  If I get the chance, however, I’ll present some additional economic indicators focusing on disposable income, and some more sophisticated analyses that have proved very useful in predicting past election outcomes.