How do conventions affect the horse race?
The polls are more unstable during the convention season than at any other time during the campaign, according to research by political scientists Robert Erikson and Christopher Wlezien. Typically, a convention creates a polling “bump” for the candidate who is nominated.
This bump tends to reflect growing support among party members for their party’s candidate. Those members of the party who emerged less than enthusiastic about the presumptive nominee start coming on board.
Indeed, conventions can move a variety of political attitudes, even ones not directly about the candidate. One of my favorite examples from 2012 is how much more positively Obama supporters came to view the economy after the Democratic National Convention. This graph, taken from Lynn Vavreck’s and my book “The Gamble,” shows the sharp increase among Obama supporters around the convention:
Why would there be a bump at all?
This is because the campaign’s information environment tends to shift during conventions. The presidential general election usually presents a fairly balanced environment, with roughly equal amounts of information about each candidate (in the news, in ads). But during the convention, the candidate being nominated will typically get more news, and more favorable news, than at other times during the campaign
How big is the bump?
Averaging over presidential elections from 1964 to 2008, the bump has been 5-to-6 points, according to both political science research and the tabulations of pollsters like Gallup. But as Thomas Holbrook notes, the bumps appear to have gotten smaller in recent years. Here is his graph:
Why would the bumps be getting smaller?
Holbrook notes two possible factors, both of which also strike me as plausible.
First, in 2008 and 2012 the conventions were held later in the summer and in consecutive weeks, whereas in previous years they were held earlier and were usually a few weeks apart. Holbrook notes one consequence: “that the convention messages end up overlapping and may cancel out each other.”
Second, there are more committed partisans and fewer swing voters in the electorate. There just aren’t that many voters for the convention to persuade.
Why are some bumps bigger than others?
The size of the bump, Holbrook argues in his book, tends to depend on how much a candidate is underperforming the fundamentals of the election — that is, factors like the state of the economy. Underperforming candidates get larger bumps, as the conventions (and much of the rest of the campaign) tends to bring the polls in line with these fundamentals. But bringing the polls in line with the fundamentals isn’t the same as pushing a candidate into the lead. For this reason, larger bumps don’t mean a likelier chance of winning.
Naturally, there are many claims about how the size of the bump depends on the specific events during the convention — the quality of the speeches, whether Clint Eastwood talks to an empty chair, and so on. These events could certainly matter, but there is rarely any solid evidence. I tend to treat all these claims as speculation.
What does all this mean for 2016?
Several features of this election might lead us to expect larger bumps. For one, as Philip Bump at The Fix has noted, there are more voters who have not chosen a major-party candidate at this point than in 2008 or 2012. This could reflect the long and competitive primaries in both parties.
According to Pollster’s estimates, about 80 percent of Republican registered voters currently support Trump, while about 82 percent of Democrats support Clinton. Typically, we would expect party loyalty to exceed 90 percent on Election Day. So it would seem as if conditions are ripe for both conventions to rally the faithful.
Which candidate is likely to benefit more? There is a decent chance it will be a wash, with maybe a larger bump for Clinton than Trump. Right now, Clinton has about a 3.5-point lead in the polls. This is just under what current forecasts suggest she will have on Election Day. For example, the forecasting site Pollyvote estimates that Clinton will win in November by about 5 points. So you could imagine that after both conventions, Clinton’s lead over Trump will be larger than it is now.
Of course, every election year has its idiosyncrasies — and clearly there is particular uncertainty about what the GOP convention will look like given Trump’s unorthodox candidacy.
But whatever transpires will be consequential. After the conventions, it becomes much harder for the candidate trailing in the polls to make up enough ground to win.