A new poll from the Des Moines Register shows that 73 percent of Iowa Republicans have a very or mostly favorable view of Wisconsin Rep. Paul Ryan, making him the very, very early front-runner for the 2016 Iowa caucuses. How much predictive value do early polls have?
Some, but only for the primaries. Studies of the usefulness of early polls are mixed. Former New York Times blogger Nate Silver took on the topic back in 2011, reviewing polling data from 1972 through 2008. Considering primary polling averages between the months of January and June in the year before the election, Silver found that early front-runners had a reasonably strong tendency to eventually win the nomination, especially if the early lead was sizable. Five of the seven candidates to poll at 35 percent or higher in early polls became the party’s candidate. This data doesn’t quite apply to the Des Moines Register’s recent poll, though, which was a favorability rating rather than a head-to-head voter preference poll. The Explainer is unaware of any data showing that high favorables more than two years before the Iowa caucuses portend eventual victory in a presidential primary.
While early primary polls have some predictive value, it’s impossible to call the general election this far in advance. In their book The Timeline of Presidential Elections: How Campaigns Do (and Do Not) Matter, political scientists Robert Erikson and Christopher Wlezien reviewed general election trial-heat polls from 300 days out until the election itself for the years 1956 to 2008 (excepting 1968, for which polls began later in the cycle). Erikson and Wlezien found that polls conducted 300 days before the election have virtually no predictive value. You should probably start paying attention to the 2016 general election polls in April of that year, because the predictive value of polls rises sharply between 300 and 210 days before an election. There is another sudden rise in predictive value around the time of the political conventions, after which the polls settle out and begin a slow and steady climb toward accuracy just before Election Day.
It’s not clear why primaries are easier than general elections to call far in advance. One possible explanation is that primary voters are committed partisans. They’re more likely to be following the races well in advance, making them less likely to change their minds as the election draws near. General elections draw far larger and more diverse crowds, including many people who don’t begin to pay attention until a few months, weeks, or even days before the election. Another possibility is that primaries are more personality-based than general elections. At primary time, the issues over which the general election will be contested are not yet clear, forcing voters to pick the candidate who is most personally appealing. It’s difficult for a candidate to change the public’s view of his or her personality, even with hundreds of millions of dollars in the war chest. In the weeks before the general election, the key issues have crystallized—data suggests, for example, that voters begin to consider the economy in their voting preferences during the second half of the election year—making it more likely that voters will change their minds with shifting political and economic conditions.
The recent Iowa poll makes it clear how open the field is to an unknown candidate. Mike Huckabee, Rick Santorum, and Rick Perry trail Paul Ryan fairly closely in favorability rating. All of them are well-known political quantities. Historically speaking, when an early leader fails to capture the nomination, the eventual victor has often come from nowhere. Jimmy Carter polled at just 1 percent in early 1975. Barack Obama’s surge in 2008 was somewhat less dramatic, but he was trailing far behind Hillary Clinton in early 2007.
Got a question about today’s news? Ask the Explainer.