Will Right-Track/Wrong-Direction Numbers Help Trump?
Right now, Hillary Clinton is doing well in the polls. She leads in national head-to-head surveys as well as those in swing states like Pennsylvania, Virginia, Florida and Colorado. Public opinion could still shift towards Trump -- we’re three presidential debates, one vice presidential debate and over two months away from the election. But as of now most of the horse race polls are looking good for the former secretary of state.
That being said, one polling question – about the direction of country – continually produces results that might scare Democrats. The number is pretty simple: Pollsters ask Americans whether they think the country is going in the right direction or if it’s on the wrong track, then tabulate the results.
Right now, 28.8 percent of Americans think that the country is going in the right direction, and 64.6 think it’s on the wrong track. Moreover, the majority of Americans have felt the country is on the wrong track for almost all of the Obama presidency. Taken at face value, these numbers might seem bad for Clinton -- she has clearly tied herself to the current administration, and most Americans are unsatisfied with how things are going.
So I decided to investigate this direction-of-country number further and see what effect it might have on the election. Specifically, I examined whether the direction-of-country polling numbers have been helpful in predicting election results and dug into how these numbers should be interpreted this time.
Is This Number Predictive? And How Do We Interpret It?
In theory, the first question should be straightforward. One could simply collect the polling on the direction of the country heading into past elections, control for other known relevant factors (e.g. the economy, the president’s approval rating) and see if it has a statistically significant effect on the outcome. But the paucity of historical data makes this task more difficult.
The archives at Polling Report show NBC News and the Wall Street Journal asked this question at various points from 1995 until now. That gives us data on the 1996 through 2012 elections -- only five data points. ABC has data on a similar question starting in the early 1970s (it asked if the country is “seriously” off on the wrong track) but the polling is still somewhat sparse; combining the data sets only yields information on nine presidential elections (they didn’t poll on this question during presidential election years in the 1970s). Gallup and Pew have also asked voters if they’re satisfied with the way things are going in the country, but with this data alone (please email me if you have other sources) it’s difficult to get more than nine data points.
That being said, the small amount of data available seems to show that when Americans believe the country is on the right track, the incumbent party gets more votes.
This graph shows the net direction-of-country number heading into past presidential elections on the horizontal axis (the percentage responding “wrong direction” subtracted from percentage responding “right track”) and the incumbent party’s popular vote margin on the vertical axis.
In 1980, 1992 and 2008 (points in the lower left-hand corner) voters thought the country was on the wrong track and the challenger took the presidency. And if voters were happy with or only somewhat dissatisfied with the direction of the country, the incumbent party candidate typically won the popular vote. It’s hard to project what Clinton’s vote share would be with this limited data (there’s no data in the region of the graph where a -36.2 percent net direction-of-country rating would fall), but it’s easy to see how this could be interpreted as a bad sign for Democrats.
But if we dig into the numbers more deeply we find that the direction-of-country number isn’t all that helpful in predicting elections. Specifically, most of the information that right-track/wrong-direction provides is already baked into a more commonly used predictor -- presidential job approval.
This graph shows that when voters think the president is doing a good job, they also tend to have a more positive view of the direction of the country.
It’s possible to think up some causal mechanism linking these numbers -- some way that one feeling drives the other -- but I think the better explanation is that these numbers picked up on the same set of feelings and judgments. Specifically, net presidential job approval often works as a summary statistic of the political mood of the country. The electorate’s overall feelings about the economy, foreign policy and other key issues get rolled into this number, and I suspect that the direction-of-country number also tracks shifts in that summary judgment.
To test this idea, I ran a few regressions. I tested whether direction-of-country had a statistically significant effect on the overall election results after accounting for presidential approval. It’s dicey to make strong inferences on such a small data set, but the results showed (after accounting for presidential approval) that direction-of-country did not have a statistically significant effect on the incumbent’s popular vote margin. Additionally, the RCP average for President Obama’s approval rating was a good predictor of RCP’s average for direction-of-country (correlation of 0.72) after the post-inaugural honeymoon (extending six months after Election Day 2008) ended.
All of these results come together to suggest that the direction-of-country number may not be so useful for predicting election results. Pollsters haven’t been asking the question for many election cycles, and presidential approval -- an older, more thoroughly studied metric -- seems to capture movements in these same feelings.
Is This Time Different?
Still, it’s worth asking: What if this time is different? That is, is there some reason to believe that direction-of-country has more predictive or explanatory value this cycle than in the past? The best evidence I could find for that is summarized in this graph:
The graph shows two clusters of points. The denser cluster of blue points shows the RCP average for presidential approval mapped onto the RCP average for direction-of-country during Obama’s presidency (the honeymoon phase is again excluded). The more spread out set of green points shows how direction-of-country ratings from NBC News and The Wall Street Journal mapped onto Gallup ratings of presidential approval from the mid-1990s through the late 2000s. Each data set has the line of best fit drawn through it.
There’s one key pattern here -- the line of best fit for each data set is almost parallel, but the line for the Obama era is shifted downwards. In other words, for multiple administrations, an increase of one point in presidential job approval has led to an increase in a little less than a point in direction-of-country polls. But in the Obama era, voter attitudes on the direction of the country have shifted downwards by about 14 points.
Put more concretely, if Bill Clinton or George W. Bush registered a zero net job approval, about 15 percent more Americans would say the country is on the wrong track rather than going in the right direction. But when Obama is at around a zero net job approval, that gap is closer to 30 percent. This is an imperfect way to compare these numbers (more explanatory variables could be added; the dates on the Gallup and NBC often don’t match exactly; some of the variation might be explained by comparing RCP averages and individual polls; etc.), but the gap here seems hard to ignore.
It’s not clear to me why this gap formed. But data suggest that this gap shouldn’t alter our estimation of the race too much. Will Jordan of YouGov examined the data from a recent Economist/YouGov poll where 61 percent of respondents said the country was on the wrong track. He found that 22 percent of voters said the country was on the wrong track and that they blamed Obama, Clinton or congressional Democrats and an additional 22 percent said they blamed some in both parties. Overall, 29 percent said the country was headed in the right direction, and 10 percent said it was on the wrong track and blamed Trump or congressional Republicans. In other words, 44 percent blamed either the Democrats or both parties (who might be disposed to vote for Trump) and 39 percent either blamed Republicans or believed the country was on the right track (who might be disposed to vote for Clinton).
That’s not a huge gap. And if less than half of voters believe the country is on the wrong track and lay some blame on Democrats, this direction-of-country number doesn’t seem to portend a radical shift against Clinton.
Again, the state of the race can change in the next few months. But for now, the direction of the country doesn’t provide much reason to doubt the central narrative of the race so far -- that is, that although the fundamentals (e.g. the economy) suggest a close race, Clinton has been able to lead by a sizable amount because voters tend to view her more favorably than Trump.