Tag Archives: election forecast

Why I’m Not Scratching My Head About The State Of The Election Today

Steve Lombardo, a former Romney adviser in 2008, writes this in his Huffington Post column today: “Any serious observer of the presidential election has to be scratching his/her head. In mid-September Obama was on track for reelection because Romney, at that point, had been deemed unacceptable by a vast segment of the electorate. Now, in mid-October, the President is dazed, staggered by a near knockout in the first debate and a subsequent Romney surge that seems to have the Governor on a winning trajectory. The problem is that neither scenario accounts for unplanned events.”

Except that’s not true.   This race is essentially tied not because of “unplanned events”, but because of perfectly predictable events; as we get closer to Election Day, more and more voters are behaving almost exactly as political science forecast models suggested they would.    As I said again and again, although Obama was outperforming the economic fundamentals in many of the swing state polls during September, that did not mean that this election was not going to tighten as more voters begin tuning into the race.

But pundits persist in insisting that rather than responding rationally to the state of the economy, voters are instead a fickle bunch who are largely ignoring the economy.  Consider this observation today by another pundit who, citing a Gallup poll titled “U.S. Economic Confidence Best Since May, tweeted this: “I give up trying to make sense of this election.” Evidently he doesn’t understand how the race could be tied if voters’ economic confidence is on the rise.  Again, however, if you look at the actual poll, (and not just the title), it’s pretty clear why Romney and Obama are running neck-and-neck.

The reality is that despite the positive trend recently in voters’ economic outlook, the poll actually shows that more Americans continue to be pessimistic about the state of the economy.  Their collective confidence may be the best since May – but that’s not saying a helluva lot.  And that pessimism is largely driving today’s polling numbers.  Gallup’s Economic Confidence Index is based on the combined responses to two questions, the first asking Americans to rate economic conditions in this country today, and second, whether they think economic conditions in the country as a whole are getting better or getting worse.  The Index is then computed by adding the percentage of Americans rating current economic conditions ((“excellent” + “good”) minus “poor”) to the percentage saying the economy is (“getting better” minus “getting worse”), and then dividing that sum by 2. So, an index above zero indicates that more Americans have a positive than a negative view of the economy; values below zero indicate net-negative views.  As you can see, the index is still strongly negative.  So, is it surprising that as Americans increasingly tune into the election to consider which candidate they will support that Obama’s “lead” has eroded?  I don’t think so.

Look, I’m not saying everything is unfolding in this election precisely as anticipated – every campaign has unexpected twists and bumps in the road.  But if you told me back at the start of September that this election would be essentially tied with less than three weeks to go, I would have thought you were reading my posts!

Which lead us to tonight’s debate.  I’ll spare you a rehashing of the ubiquitous “Five Things Candidate X Must Do To Win”  comments, and instead simply remind you to listen to what the candidates say more than watching how they say it. Candidates’ posturing and body language is a vastly over rated phenomenon, in my view.  While I expect Obama to come out more energized, and to be more pointed in his critique of Romney’s policy platform, I don’t think there’s a lot of room left to change the campaign narrative on either side.  This election is about where it should be, and if history is any guide, we aren’t going to see Obama move more than 2 polling points – at most – in either direction.

I will, of course, be live blogging the event. We’ve upgraded our blogging program to make it easier to comment (so I am told!), so I hope you’ll join in.  Remember, this is the most crucial event in the campaign so far – I know this because most of the talking heads told me so.  And they can’t be wrong.

I’ll be back on at about 8:45.  See you then…. .

The 2012 Election: Polarized Voters – Or Polarized Choices?

I want to pick up a discussion I began in this post at the Economist’s Democracy in America site regarding political science election forecast models. At this point in the presidential race, less than 100 days before the election, some of the fundamentals that political scientists incorporate into these econometric models are set, and we are going to see more of their predictions come out in the next several weeks.  As I noted in the earlier post, with the release of the 2nd quarter GDP figures last Friday, all the pieces are in place for Emory political scientist Alan Abramowitz to issue his “Time for A Change” forecast.  It predicts that Obama will receive 50.5% of the major party vote in November, making him a very slight favorite to retain the presidency.  But as I discussed, Abramowitz has tweaked his forecast model, adding a “polarization” variable that he describes as follows: “For elections since 1996, the polarization variable takes on the value 1 when there is a first-term incumbent running or when the incumbent president has a net approval rating of greater than zero; it takes on the value -1 when there is not a first-term incumbent running and the incumbent president has a net approval rating of less than zero.”

Based on this description, the polarization variable will come into effect in 2012, since Obama is a first-term incumbent.  This matters, because the addition of the polarization variable reduces Obama’s projected vote share by more  than 2%.  That is, under the “traditional” pre-polarization Abramowitz model, if I’ve done my calculations correctly, his forecast would have Obama winning 52.7% of the major party vote – a more comfortable margin!  However, as Abramowitz explains, his traditional model had overstated the winning candidate’s margin of victory in the last four elections.  The reason, he argues, is an increasingly polarized electorate that is less likely to cross party lines to vote for the other party’s candidate even when the “fundamentals” suggest they should. The polarization variable brings his predictions more in line with the actual results of the last four elections.

But is the variable justified, or is it simply an ad hoc retrofit that has no substantive basis?  In previous posts I have argued that there is little evidence that the public has become more polarized in recent years, even as political elites have grown increasingly partisan.  Abramowitz, however, disagrees. As he and his  co-author Kyle Saunders write in this article “The argument that polarization in America is almost entirely an elite phenomenon appears to be contradicted by a large body of research by political scientists on recent trends in American public opinion.”  Moreover, they argue that this increase in polarization tends to make citizens more engaged in politics – not less. As evidence, they point to National Election Studies (NES) data that show a steady increase, from about 50% in 1952 to about 75% in 2004, in the number of Americans who perceive a strong difference between the two parties.  There was a similar increase, from just under 70% to well over 80% during this time span in the number of Americans who care who wins the presidential election.   This is evidence, Abramowitz argues, that voters are increasingly polarized, and he uses it to justify tweaking his traditional model.

The problem with the Abramowitz/Saunders’ claim  of a more polarized America, as Mo Fiorina has pointed out, is that it confuses a polarization in voter choices with polarization among voters themselves.   Thus, we should not be surprised that an increasing number of Americans perceive a difference between the parties.  There is a difference, but it reflects a process of party sorting  – not a growth in ideological extremism.  Simply put, party labels have become increasingly aligned with partisan ideology; there are fewer conservative Democrats or liberal Republicans today.  So quite naturally more people perceive a difference between the parties today than they did a half-century ago – but this does not mean the public is more polarized along partisan lines.

Similarly, if the choices for president are viewed as more ideologically extreme, it makes sense that more voters are going to express an interest in who wins.  Indeed, one reason voter turnout was so high in 2004 and again in 2008 is that that many voters viewed the candidates as espousing ideologically distinct views on the major issues.  In 2004, as Abramowitz’s own data shows, George W. Bush was a very polarizing figure.  It is therefore no great surprise that a record number of voters cared who won the presidential election that year, and that turnout was up.  A similar argument can be made about the 2008 presidential race.

So, if the American public is not increasingly polarized, contrary to Abramowitz’s claims, does that invalidate his inclusion of a polarization variable in his forecast model for the post-1992 presidential elections?  Is Obama more likely to get 52.7% of the popular vote rather than 50.5%? Not necessarily.  Abramowitz’ explanation may be faulty,  but the polarization variable may still work if more voters are viewing their choices in the presidential election in polarized terms.    Even if this is true, however, I’d like to see more evidence suggesting that the public’s perception of the ideological differences between the presidential candidates grew significantly higher in 1996, and has remained higher ever since, as Abramowitz’s new forecast model suggests.  Moreover, I’m not sure this is leading an increasing number of voters to cast ballots contrary to what the fundamentals would indicate – it may be instead that they see the two candidates’ issue stances regarding the fundamentals  as distinctly different.  Thus, in 2012, Romney and Obama are espousing two very distinct views for addressing the lingering effects of the Great Recession.  Many moderate voters may not buy entirely into either view but they don’t have the luxury of mixing and matching elements from both candidates’ programs.  Instead, they must choose – even if they don’t necessarily fully embrace either choice.

The GDP Numbers Are In – Do They Say Obama Will Win?

The much awaited second quarter GDP growth figure came out yesterday and, while it wasn’t a disaster, neither was it particularly good sign for the economy – or for President Obama’s reelection chances.  The government’s first estimate (they often revise the figure as new data comes in) is that GDP grew at an anemic 1.5% – a .5% drop from first quarter growth, and only half of the growth rate experienced during the last quarter of 2011.   This downward trend line is not what an incumbent president wants to see heading into an election.

As you know from reading my previous posts, GDP growth is one way of measuring one of the key “fundamentals” – the state of the economy – that I have been arguing is far more important to the election outcome than the Bain controversy or Obama’s verbal “gaffes”.  But it isn’t the only factor influencing the election, and so we shouldn’t overstate its significance either.  Peter Cahill has gathered data on second quarter GDP growth in every election year dating back to 1948, and correlated it with the actual share of the major-party vote won by the incumbent presidential party’s candidate.

Think of the trend line as the “real” linear relationship between GDP growth and vote share. If we plug 1.5% into the equation defining that line, it predicts that Obama will get about 50.3% of the two-party vote come November based on 2nd quarter GDP growth alone.  However, as you can see from the graph, while higher GDP growth generally correlates with a greater vote share, the relationship is not perfect; GDP growth only explains about 36% of the variation in vote share.  So a lot of other factors are going to come into play.  What are they?  As I discussed in yesterday’s post for the Economist’s DIA blog, Emory political scientist Alan Abramowitz’ “Time for A Change” forecast model adds three additional variables to second quarter GDP growth: the incumbent’s net approval (approval minus disapproval) in the Gallup poll at the end of June, how long the incumbent’s party has held the presidency and – in a recent innovation – a “polarization” term that takes into account the increased polarization of the electorate since 1996.  With yesterday’s GDP release, all the figures are in place for Abramowitz to predict Obama’s share of the major party vote come November.

Drum roll please!

By plugging the relevant numbers into the Abramowitz forecast equation, it spits out Obama’s predicted share of the major party vote come November as 50.5% – not much different from our estimate based only on second-quarter GDP growth. Based on the confidence interval around this prediction, Abramowitz estimates that if history holds Obama has about a two-thirds probability of winning the election.

This is an estimate, mind you, based on data from a small number (16!) of previous presidential elections.  But I would argue that it is better than a guess – Abramowitz’s model has performed generally quite well in out-of-sample forecasts, coming within 1.5% of the actual vote about three-quarters of the time. On the other hand, I  wouldn’t bet my kid’s tuition, based on this one model, that Obama will be victorious.  In short, it is telling us what pretty much every other  indicator suggests: that this is going to be a very, very close election but that Obama can be considered a very slight  favorite.

So, does this mean the outcome is already in the bag, and that what happens from here on out doesn’t matter.  Not at all.  Campaigns do matter – see my previous posts here and here.  And in terms of consequences, in such a close election,  they arguably matter even more this time around.

Abramowitz’s model, of course, is only one of several constructed by political scientists and, as I discussed in my Economist post, his recent change to his model is not sitting well with everyone. (I’ll discuss this in a separate post.) By my count, there were more than a dozen econometric-based forecast models in 2008.  Although they were all, save one, able to predict Obama’s victory, they weren’t equally reliable in forecasting the actual popular vote (although they did pretty well in the aggregate). And, of course, as we get closer to the actual election,  none of them will be as reliable as simply aggregating the public opinion polls, which is what Sam Wang, Nate Silver and others will end up doing.  So why discuss them at all?  Because the best ones remind us that there is a context to this election which largely determines how it will turn out.  And right now that context is saying that this election may be too close to call.