Presidential Approval on the Path to the Midterm: The Pattern is That There is No Pattern

KEY POINTS FROM THIS ARTICLE

— One wouldn’t expect the president to get much of an approval boost from the nation’s 250th anniversary, particularly as he handles it in a highly partisan way.

— The president’s approval numbers, much like gas prices, are a little better lately, but still worse than they were before the attack on Iran.

— There is not a consistent pattern for postwar presidential approval from July 1 to the midterm, and in recent years a president’s approval hasn’t changed much over the last four months of the campaign.

Approval trajectory from July to November

There was perhaps a world in which President Trump might have used the looming semiquincentennial to bolster his, and his party’s, fortunes for the fall. Or at least tried to, perhaps by presenting himself as something other than a hard partisan figure.

Though in some ways it’s ancient history, President Gerald Ford used the 1976 bicentennial as part of a strategy to dig himself out of a deep hole against challenger Jimmy Carter in that year’s presidential race. As longtime political analyst Michael Barone recounts in his brilliant book Our Country: The Shaping of America from Roosevelt to Reagan, Ford and his team “concentrated heavily on building a positive feeling not only about the president but also about the country,” exemplified by a campaign song, “I’m Feeling Good About America.” (That’s also the name of our Center for Politics documentary about the 1976 campaign, free to watch on YouTube). Ford still lost, but it ended up being a very close election.

It may be that Republicans find a way to dig out of their current political hole, but a sunny, optimistic president rallying the country over the course of a happy, celebratory summer doesn’t seem to be in the cards. This president cannot even countenance helping his party achieve an easy win, as he has both denigrated and thus far declined to sign a bipartisan housing bill, a matter of national concern in an election whose major theme is “affordability.”

Events marking the 250th anniversary of the signing of the Declaration of Independence have turned into yet another partisan event, with the president apparently planning to hold what amounts to a campaign rally on the Fourth itself. The World Cup, hosted partially by the United States, has excited international visitors and provided that rarest of all things on social media: organically heartwarming content. But that’s not really a story that involves the president.

None of these things seem likely to actually hurt the president in terms of his approval rating. But the president—not for his sake, but for his party’s sake—needs to be improving his numbers. As we noted a couple of weeks ago, if Democrats actually win the Senate this year—they’re still underdogs to do so—Trump’s numbers being worse than they were in 2018 would likely be a big explanation why.

It is possible that some of the downward pressure on Trump’s approval because of gas prices is being relieved to a modest degree. According to AAA, the average national price of a gallon of regular gasoline is currently about $3.85, down from about $4.30 a month ago, but also markedly higher than about $3.20 a year ago. The conflict in Iran that led to the gas price spike remains unstable, as there seems to be an awful lot of shooting going on amidst a ceasefire.

The gas price situation—bad but not quite as bad as it was a month ago—also describes the president’s approval: Per the Silver Bulletin average, Trump is at 39.4% approve/57.5% disapprove, or about -18 net. That’s a couple points better than his second-term low point of about -20 in mid-to-late May, but also several points worse than -13, which was Trump’s net approval right before he launched attacks on Iran.

Republicans remain in danger of Trump being underwater in approval in states with key Senate races. For instance, the New York Times/Siena University poll of Texas released Tuesday morning showed a 47%-47% tie in that state’s Senate race. Trump was at just 44% approval/53% disapproval among likely voters. Republican Senate nominee Ken Paxton was at 47%, so Trump’s approval is not necessarily a ceiling for Republicans. Across six states that feature key Democratic Senate targets recently polled by NYT/Siena (Alaska, Iowa, Maine, North Carolina, Ohio, and Texas), Trump’s approval rating was 43%, but Republican Senate candidates were getting 47% of the vote. That means Republicans are at least currently showing the ability to run ahead of Trump in what is collectively a right-of-center group of states, but their task would of course be easier if Trump’s numbers were better.

If Republicans can’t rely on the president himself to do things that could plausibly help his approval rating, could they at least hold out hope that Trump’s approval will be higher in November than it is now?

Quite possibly. For one thing, presidential approval in the last five midterms has generally been a little higher in the exit poll than it was in pre-election polling averages (the only exception among the last five midterms was Barack Obama in 2010). This is probably tied into the fact that different pre-election polls measure different groups of people, and all of those polls get lumped together in polling averages. There are polls of all adults, polls of registered voters, and polls of likely voters, with each of those successive groups becoming smaller: not all adults are registered voters, and not all registered voters are likely voters. The exit poll, meanwhile, is designed to measure the opinions of the actual November electorate. Even in polling now, Trump’s numbers tend to be worst in polling of all adults. The Silver Bulletin average shows polling for just all adults (Trump -23 net approval) compared to polls of just registered or likely voters (-17). So there is a difference there, although once we get closer to the election, more registered voter surveys will add likely voter screens, which should help Democrats to some degree in polling this year because their voters are expressing higher enthusiasm for voting (as one would expect for the nonpresidential party in a midterm). We saw an extreme example of how this might work in a national Marquette Law School House generic ballot poll released recently: Democrats had just a 2-point edge, 47%-45%, among registered voters, but that expanded to 8 points among likely voters. Republicans still have an apparent enthusiasm problem, although some of that may just improve organically as the election approaches. One thing that may have helped the GOP juice some enthusiasm and salvage some purple and red-state races in 2018 was the contentious confirmation hearings over now-Supreme Court Justice Brett Kavanaugh. That contributed to why what turned out to be an erroneous NPR report on Tuesday saying that Justice Samuel Alito was retiring caused such a stir, but NPR quickly retracted the report. A court fight this fall could change the contours of the election, but as we are writing this now, there is no court vacancy to fill.

Where Trump’s approval goes from here is anyone’s guess: He has shown an ability to regenerate in the past from low points, and it shouldn’t surprise anyone if his numbers are a little better in November than they are now. However, it’s also possible that Trump, who is reaching the midpoint of his second and final term in office, enters a period of drift where his approval rating doesn’t really get better and perhaps even gets worse.

In considering Trump’s possible trajectory, we wanted to see if there was any sort of consistent pattern in how presidential approval changes in midterm years over the last four months of the election. We used the Silver Bulletin historical approval averages going back to the 1946 midterm, seeing where a president was on July 1 compared to the day of the general election in early November.

The polling history is shown in Table 1.

Table 1: Postwar presidential approval, July 1 vs. November election

Note: *Ford’s first approval rating in the average came on Aug. 19, 1974, not July 1 (when Richard Nixon was still president). Nixon’s final approval rating when he left office was only 24%.

Source: Silver Bulletin, Crystal Ball research

The sample size, 20 midterm elections, is not large, and there’s also not a consistent pattern. It is true that in the majority of the elections (12 of 20), the president’s net approval got worse from July to November. But in seven instances, the president’s approval improved, and in one instance (George W. Bush in 2006), there was no change at all. There also are several other years where the change was so minor either way that there probably was not any real change. So it’s a mixed bag.

Interestingly, Bush in 2002 registered the only truly substantial change from July 1 to the midterm over the past few decades, and it was actually a big decline. Bush benefited from a huge “rally around the flag” effect following the Sept. 11 terrorist attacks, and his approval was still very good throughout 2002, although his numbers fell over the course of the campaign season (with noticeable drops among both independents and especially Democrats, per Gallup’s tracking). Bush was an unusually aggressive midterm campaigner for his party’s candidates in that midterm. As Center Director Larry J. Sabato wrote in Midterm Madness, our book on the 2002 election, Bush decided to “put his considerable store of post-September 11 chips on the line,” using his accumulated political capital as part of a successful effort to modestly expand a small GOP House majority and retake the Senate after it had flipped to a 51-49 Democratic edge following Vermont Sen. Jim Jeffords’s mid-2001 caucus switch from Republican to Democratic. This probably had the effect of making Bush a more partisan figure, contributing to an overall approval decline. But his approval at the end of the campaign was still quite good, so he was a net plus for Republicans that year. Though by 2006 Bush certainly ended up becoming a huge liability for Republicans.

Again, that big shift in Bush’s approval over the course of the 2002 campaign is more of the exception than the rule, as other recent midterms have seen the president’s approval change only slightly. Barack Obama’s net approval dropped slightly in both 2010 and 2014, directionally suggestive of the weak midterm performances for Democrats in both years. Joe Biden’s approval improved a bit, which likely helped Democrats to some degree as they held the Senate and only modestly lost the House (and court politics may have aided Democrats that year following the Supreme Court overturning Roe v. Wade in late June). And Trump himself hardly budged from July to November in 2018.

As is often the case with modern presidential approval ratings, expecting little-to-no change is a better bet than expecting big change. Particularly as the president doesn’t change much himself.