Robert Erikson is professor of political science at Columbia University and an expert on public opinion and elections. He answered a few questions via email about his presidential forecasting model and his new research on presidential “coattails” in U.S. House races.
Q: Let’s start with your own forecast of the presidential election. What goes into it, and what does it predict?
Chris Wlezien and I have a forecast based on two factors: economic data and polling data. Our economic measure is based on a tally from The Conference Board’s Index of Leading Economic Indicators (LEI) through the beginning of the election year. This index gives us a measure of how well or poorly the economy has done over the current presidential term, as well as a sense of how the economy is likely to change in the months before the election. A stronger economy should benefit the incumbent party.
For polling data, we rely on the incumbent party candidate share of the two-party vote in the polls at the time of our forecast. One advantage of our model is that the economic data allow us to make relatively early predictions, even though polling data early in the campaign — say, in the first quarter of the election year — are not very predictive of the November result. As the campaign goes on, however, polling data become more predictive of the Election Day vote, as we have shown in “The Timeline of Presidential Elections.”
This year, our forecast in both June and August predicted that Hillary Clinton would receive 52 percent of the two-party vote. At that time, the economic indicators were largely neutral with respect to the election, but Clinton had a lead in the polls, which is why the forecast favored her.
Q: That brings us to what this might mean for congressional races. Do presidential candidates have “coattails” in down-ballot races? Does Clinton’s lead mean that she’ll help other Democrats win?
That’s a more complicated question than many people realize, as I show in a recent article. There are actually a couple of ways that the presidential race can affect down-ballot contests.
Let’s take coattails to start. This is the idea that the winning presidential candidate can sweep into office fellow party members in down-ballot races. There are multiple reasons this might happen. People could vote a straight-party ticket out of enthusiasm for the presidential nominee. It could also be that people’s vote for president and lower-level races are affected by the same factors, such as the state of the economy, which ties their fortunes together.
To estimate the effect of presidential coattails, I looked the U.S. House vote in the 17 presidential election years from 1948 to 2012. For House candidates, I find that it does help for their party’s presidential candidate to be winning. For every percentage point that a presidential candidate gains in the two-party vote, their party’s down-ballot candidates gain almost half a point themselves. If Hillary Clinton’s lead holds, that would appear to benefit Democrats this year.
Q: But there’s another way that the presidential contest can affect down-ballot races?
Coattail effects can be tempered by what’s known as “balancing.” The idea here is that in down-ballot contests, some voters may tilt toward the opposite party of the candidate they believe will win the White House. They may do this to balance the ideological makeup of government. If the president’s going to be a Democrat, for instance, maybe a Republican Congress can help keep them in check. This could be a sort of hedge against policies that might stray from the ideological middle.
In my analysis, I estimated balancing by looking at the effect on the House vote of the pre-election odds that the Democrat would win the presidency. This serves as a proxy for how sure voters will be that the Democrat will win. As the odds go up, the Democratic House vote declines. Controlling for other factors, this suggests that the more certain voters are that a Democrat will become president, the more likely they are to cast ballots for Republicans in House races. (It works the same way when Republicans are favored.) Being seen as a certain winner rather than a certain loser actually penalizes down-ballot candidates by about 3.5 points.
With Clinton currently favored (if narrowly), this would mean that some voters may cast ballots for Republicans down-ticket in the hopes of electing a GOP Congress that would moderate a (nother) Democratic president.
Q: It sounds like coattails and balancing could both occur, but cancel each other out?
Yes, in fact, there are certain landslide elections since World War II where there were strong coattail effects that were hard to detect because they were canceled out by balancing voters. That happened in 1956, 1972, 1984, and 1996.
The ideal for a party is to win the presidency by a large margin — but in an upset that voters don’t see coming. Coattails would still be strong, but there would be much less balancing because voters wouldn’t be certain which party is going to win the White House. That’s the way a party gains in Congress from coattails without suffering the penalty from balancing.