Skip to main content
Don’t Just Look At Trump’s Approval Rating To Judge His First 100 Days

With President Trump’s 100th day in office approaching, analysts and pundits are taking stock of his performance. How’s he doing? The simplest, most common way to answer that question is to look at Trump’s approval rating (it’s low, but the bottom hasn’t fallen out). And that’s a perfectly fine measure by which to judge the White House; approval ratings tell us how Americans feel about Trump, and — eventually — they’ll do a good job at approximating his chances of re-election.

But presidential approval is not a great measure of the more immediate electoral issue: How will the president’s party do in the midterm elections? By all means, keep one eye on Trump’s approval rating, but keep the other on the generic congressional ballot — a longstanding poll question that pits a nameless Republican against a nameless Democrat in a race for the House.1

Approval ratings have historically jumped all over the place, and even on the eve of the midterms, they do a relatively poor job of forecasting the election results. The generic ballot, in contrast, has tended to be far more telling. And the generic ballot polls conducted since Trump became president suggest that House Republicans are on track to suffer a potentially major defeat in the 2018 midterm elections.

Democrats are in a historically great position according to the generic ballot. Typically, at this point in a midterm cycle — the first half of the year after the preceding presidential election — the party in control of the House is fairly popular. Most likely, they did at least OK in the last election (they still control the House, after all) and they simply haven’t had enough time to upset the electorate too much. Republicans in 2017, though, are one of the exceptions to the rule.

2018 39.9% 45.0% -5.1
2006 39.4 43.6 -4.2
1998 40.0 44.0 -4.0
2014 39.0 42.1 -3.1
2002 41.8 44.3 -2.5
1982 42.0 42.5 -0.5
1994 40.7 39.7 +1.0
1954 47.5 46.0 +1.5
2010 40.8 37.9 +2.9
1986 50.0 42.0 +8.0
1958 51.7 40.5 +11.2
1942 43.3 31.7 +11.7
1974 50.3 31.8 +18.5
1962 58.2 33.5 +24.7
1966 58.0 31.7 +26.3
House Republicans are in a historically poor position

Average of polls taken in January through June of the year before the midterm election. Some years are missing because polling data was not available.

Sources: Roper Archive, HuffPost Pollster, Surveymonkey

Even when Democrats were busy crafting and eventually introducing a fairly unpopular health care bill in the middle of 2009, they still held a 3 percentage point lead on the generic ballot for the 2010 midterm elections. This year, a combination of Trump’s unpopularity and House Republicans’ inability to pass any major legislation has likely contributed to Americans wanting to put Democrats back in charge.

Will that last? Isn’t it too soon to even be looking at these numbers? Sort of. The relationship between how the House majority party fares on the generic ballot at this point and how it does in the next midterm election is messy. Republicans weren’t polling well in the first half of 2013, for example, but went on to dominate the 2014 midterms. House Republicans could turn things around before November 2018.

But here’s the problem for the GOP: Midterms are almost always about the president. Voting for Congress in midterm years is essentially just a mechanism for passing judgment on the White House. In 2018, unlike 2014, the House and presidency will be controlled by the same party. And when we look at the generic ballot numbers this early in the cycle while also accounting for who holds the White House, the generic ballot has forecasted midterm House results fairly well.

Clearly, there’s a strong correlation (+0.78) between how the White House party is doing on the generic ballot at this point and how it ends up doing in the midterm election.2 In fact, the party in the White House consistently underperforms the generic ballot at this point.3 That is, things usually get worse for the party in control of the White House as the midterm elections get closer. For an extreme example of this trend, we can look at 1965, when President Lyndon Johnson’s Democratic Party was up 26 points on the generic ballot. But as Johnson pushed for a liberal agenda, and as mass protests riots dominated the news and the Vietnam War raged, Democrats went on to win the House popular vote by just 3 points in 1966.

History suggests that the Democrats’ current 5-point advantage in the House vote would, on average, become an 8-point win for them in the 2018 House popular vote. Although it’s difficult to gauge how many seats Democrats would gain if they won the House popular vote by 8 points, that margin would probably give them at least a 50 percent chance of gaining 24 Republican seats in the House and the majority.

Of course, it’s far from guaranteed that Democrats would win the House vote by exactly 8 percentage points. Like with a president’s approval rating, generic ballot margins tend to undergo a reversion toward the mean, just as the president’s party’s standing tends to decline. The fact that Trump is already so unpopular may mean that he won’t become more unpopular as the midterm approaches. Therefore, the Democrats’ lead in the generic ballot may not expand further, instead holding steady the way it did in 1974 amid the Watergate scandal. It’s also possible that some event (e.g. a terrorist attack) interrupts the normal historical trend, and the White House Party gains ground on the generic ballot, as Republicans did after 9/11.

Still, while you’re reading dozens of 100-day analyses this week, leave some room for the generic ballot. Democrats’ 5 percentage point lead is probably one of the worst signs for Trump and the GOP. It shows that Trump’s unpopularity is translating down-ballot and hurting his party. When you combine the generic ballot numbers with high Democratic enthusiasm, there’s no sign that Trump’s Republican Party has some unique ability to “defy midterm gravity.” The GOP’s House majority is in danger.


  1. Some pollsters ask whether a voter prefers Congress to be controlled by Democrats or Republicans.

  2. This finding is in line with past work by Joseph Bafumi, Robert S. Erikson and Christopher Wlezien, who showed that the generic ballot taken in the beginning of the year of the midterm election also has a strong predictive value for the eventual midterm results. (We’re, of course, looking at numbers a year earlier than they did.

  3. Bafumi, Erikson and Wlezien found the same thing.

Harry Enten is a senior political writer and analyst for FiveThirtyEight.