Home > News > Only one poll number right now tells us anything meaningful about the 2016 election
187 views 7 min 0 Comment

Only one poll number right now tells us anything meaningful about the 2016 election

- July 10, 2015

The current president’s approval rating is one key factor to consider. (Kevin Lamarque/Reuters)
Although the 2016 election is nearly 500 days away, the nation — or at least the nation’s political junkies — remain hungry for news about the presidential campaign. But what news is there? Polls. So far — in 2015 alone! — no fewer than 57 polls have asked voters to choose between hypothetical nominees Republican Jeb Bush and Democrat Hillary Rodham Clinton.
So it’s useful to be mindful that “trial heat” polls conducted now have zero ability to predict the winner of an election that’s 16 months away. Polls don’t even tell us much about primaries and caucuses that are six months away.
Remember this, then, when digesting last week’s widely reported Gallup finding that, after nine months in which a roughly equal number of Americans identified as members of each party, Democrats again outnumber Republicans in the electorate. That puts Democrats “in a favorable political position as the 2016 campaign is getting underway,” wrote Gallup’s Jeffrey M. Jones. The numbers are “flashing a warning for Republicans,” declared a headline here at The Post.
Yes, it’s true that more Americans are identifying as Democrats, and not just in Gallup’s surveys. When you take HuffPost Pollster’s current averages from all published polls (as averages are more reliable than any single data point) you find that the share of Americans identifying as independents has gone down, while the numbers of Democrats — and to a lesser extent, Republicans — are going up.
But what — if anything — does that mean for an election that’s nearly a year and a half away? The figure at below left displays the relationship between party identification in the electorate (or “macropartisanship,” as we political scientists like to call it) in the year before a presidential election year and the Election Day outcome. For each of the 16 presidential elections extending back to 1948, the Democratic presidential candidate’s share of the popular vote is plotted against the Democratic Party’s average affiliation advantage in the year before the election year.
The result? Party identification more than a year in advance of the election predicts nothing about how the election will ultimately turn out. (To be wonky about it, party identification insignificantly predicts the opposite of the final election outcome. A similar analysis, which I don’t show here, finds no relationship between change in party identification the year before an election and the election’s ultimate result.)
Sharp-eyed proponents of the hypothesis that partisanship predicts elections might point to a positive relationship over, say, the past four elections (as did the Upshot’s always perceptive Nate Cohn when I posted this figure on Twitter last weekend). But to predict outcomes accurately, we need data from more presidential campaigns. Four elections do not a reliable trend make.
So what’s a political junkie to do? Watch this number: President Obama’s approval rating. As political scientists Robert Erikson and Christopher Wlezien explain in their masterful volume “The Timeline of Presidential Elections,” presidential approval ratings reflect many of the factors that affect how Americans vote — especially their view of the economy. Erikson and Wlezien show that if voters approve of the current president — even as far out as 200 days before an election — that’s a good indication that the incumbent party will win.
Right now we’re more than twice that far away from Nov. 8, 2016, but even so the president’s approval rating today is still a significant predictor of which party will win next year. You can see that in the right-hand graph, which plots the popular vote for the same 16 presidential elections against the average net approval rating of the incumbent president recorded from June through September of the year before the election year — that is, right now. In this graph, since we’re predicting the Democratic share of the vote, signs on approval ratings are reversed if the incumbent is a Republican. George W. Bush’s terrible approval rating in the summer of 2007 is signed as positive; Dwight Eisenhower’s fantastic rating in the summer of 1955 is signed as negative.
The relationship between presidential approval and election results 16 months later is statistically significant (at p = .04 if you’re counting) and as we would expect, slightly stronger in election years when the incumbent is actually running for reelection (in an analysis not shown here). Summer presidential approval accounts for 23 percent of the variation in election outcomes in the following year. Removing two influential outliers — the 1964 and 1992 elections — from the analysis does not change the finding. The results suggest that the incumbent party becomes the bare favorite to win when its president’s net approval rating is slightly positive in the summer of the year before an election, at +4. Currently, HuffPost Pollster estimates President Obama’s net rating to be just the opposite, at -4, making the Republicans narrowly favored to win the 2016 election by a predicted margin of 51 percent to 49 percent.
Of course many other developments will emerge that affect next year’s outcome. And given that these predictions are based on very few elections, they have large margins of error. But at this early date, they’re far superior to speculations based on trial-heat polls or polls about voters’ party identification. In fact, compared to any other polling data regularly covered by the media right now, Obama’s approval rating is the only number that history demonstrates tells us something meaningful about which party may win next year’s presidential election.