Trump's job approval is in potential "blue wave" territory

Unless there's a war, presidents always lose House seats in their first* midterm

I am aware that we talk about polls a lot here. I think polls are important and worth talking about, but I acknowledge this can be alienating. One comment I get frequently online is that all these polls and presidential approval ratings don't really even matter. People say this to me for a variety of reasons, taking positions ranging from "polls are always wrong, who cares?" to "President Trump doesn't care about polls, he's going to do what he wants."

I get it, polls schmolls. What about the voters? And what about policy? Those are important too! But I think there are a couple of very good reasons to pay close attention to public opinion: First, you should care about polls because we live in a democracy, and what people say should matter to our leaders. As those living in authoritarian regimes are quick to remind us, the only way for opinions to matter is to make them matter. When you can’t vote with a ballot, vote with your attention, energy, anger, enthusiasm, and actions. Polls are just another way to hear those emotions.

But second — and more germane to what we do here at Strength In Numbers — approval ratings also map pretty neatly onto political and electoral outcomes in the real world. So this week's Chart of the Week is a way-too-early look at the 2026 U.S. House midterms, via way of the relationship between presidential approval in May the year before a midterm and eventual midterm seat loss for the party of the president.

How much does it really matter that Trump is unpopular?

---

Presidents lose in the midterms. A bad approval rating makes things worse.

According to my average of polls, 43.5% of Americans approve of the job Donald Trump is doing as president today, while 51.8% disapprove — for a net rating of -8 (when rounding).

We know this is a bad rating because we have historical benchmarks to compare Trump to. -8 is, in fact, the worst approval rating for a president in their first(-ish) term since 1935 — other than the president's rating in his initial 2017 term. Some presidents were doing worse at this point in their second terms, but since Trump did not serve his terms consecutively, I do not think that is a better benchmark. Still, I acknowledge some potential measurement error here.

Historically, a -8 rating only happened when presidents were presiding over tough economic times or scandals. Reagan was at -8 during the early 1980s recession, for example, and Biden was at -8 before peak inflation in 2022-2023, when it fell much further.

To put the impact of a -8 rating in concrete terms, we can look at what a rating this early in the typical presidential term says about the president's performance in their first midterm elections. The chart below shows each president's approval rating in May of the year they were inaugurated, with the number of seats their party lost in the U.S. House of Representatives in the next year's midterms:

Note two things. First, in every case except 2002, the president's party lost ground in the House.

Second, generally speaking, the lower a president's approval rating is in May of the year before a midterm, the more House seats their party tends to lose in that election. If we look back at recent history, a significantly negative approval rating has often preceded substantial midterm losses. For instance, in the lead-up to the 2018 midterms during Trump's first term, his approval was also in negative territory (hovering between -10 and -12 points by May 2017 in various averages), and the Republican party lost 43 House seats, flipping control to the Democrats.

A net approval of -8 is indeed on the lower end of the historical spectrum, putting President Trump's party in a potentially precarious position for 2026 — equal to or worse than their position in 2018. While every election cycle is unique, and many other factors will come into play, a net approval of -8 points this far out suggests that the conditions could be pointing towards significant losses for the president's party, potentially echoing the "blue wave" of his first cycle."

The precise number of seats forecast here is not the point; A better prediction would certainly require us to look at which seats were up for the taking, who was retiring, who was running, etc — and that is the job for a proper forecast model. You can see there is quite a lot of uncertainty in the prediction from a simple linear model predicting midterm seat loss with approval ratings data 440 days before the midterms. Economic conditions, major legislative achievements (or failures), and unforeseen domestic or international crises can all shift public opinion.

A bad rating hurts in Congress, too

What this early data does underscore, however, is that the presidential approval ratings can carry significant weight. They are a barometer not just of public sentiment, but a fairly reliable leading indicator of tangible electoral consequences.

Low public support can have political consequences, too. Presidential power is often exercised through the strategy of "going public," in which the president appeals directly to the people to pressure lawmakers to support his agenda.

However, when a president is unpopular, this strategy loses its teeth. Members of Congress, particularly those in competitive districts or from the opposing party, feel less compelled to align with an unpopular president. In fact, association with a low-approval president can become a liability, making them more resistant to the president's agenda to safeguard their own electoral future.

Consider President Biden's experience in his first term. As his approval ratings declined from their initial highs, particularly during the withdrawal from Afghanistan and amidst rising inflation, his administration faced significant headwinds in passing key components of its "Build Back Better" agenda. His diminished public standing made it more challenging to unify his own party (especially with narrow margins in Congress, and red-state Democratic senators bucking the party line) and provided little incentive for Republicans to cooperate.

While legislators are not necessarily keenly aware of their constituents' views on specific policies, they understand when they are angry. A president with a -8 net approval rating simply doesn't have good leverage.

Although it doesn't look like Trump is directly influenced by public opinion, the public can still pressure the people around him. And regardless of their actions, the president himself is undoubtedly getting polling data from other channels too. Why else would he have a pollster on payroll?

So here’s my answer for those saying polls don’t matter, wondering if Trump's current unpopularity truly influences electoral and political outcomes down the line: Quite straightforwardly, yes — history suggests it matters a great deal.