How low can Trump’s approval rating go?

Historically, net approval declines by about 20 points from the 100-day mark to the midterms

If it's Friday, it's time for the Chart of the Week! Although this week, you technically get three new charts (and four total), so maybe we need to rename this Charts of the Week... anyway:

On Monday, someone posted on the Strength In Numbers chat wondering if President Donald Trump's approval rating, now the lowest on record for the 100th day into a president's term, will continue to fall between now and the 2026 midterms.

The answer to this question, of course, depends on what Trump does over the next 18 months. And since I have no special knowledge about his plans, policy or political or otherwise, I don't feel comfortable making a direct prediction of what the president's approval rating will be on November 3, 2026. Frankly, making predictions more than a month out right now seems like an exercise in futility.

But what we can do is model historical presidential approval ratings for the first two years in a term and come up with a baseline for where Trump's rating "should" be on each day of his presidency, if he's tracking the historical trends. Readers can use these trends to benchmark Trump's performance over the next two years.

So let's dive right in...

Historical presidential approval ratings

First things first, here's the data I'm using. It consists of a record for every president's net approval rating on each day of their presidency, up until the 652nd day, subtracted from their initial rating. That tells us how much support each executive has lost over time, taking into account the percentage of adults who approve of them and the percent who disapprove. And why 652 days? Well, because that's the length of time between January 20, 2025, when Trump was inaugurated as president for a second time, and November 3, 2026 — the next midterm elections.

Because the charts are a bit unwieldy (they look like I just threw spaghetti at my computer screen), I have split the presidents into two charts based on when they were elected. The first chart shows approval for presidents elected between 1952 and 1992:

And this chart shows presidents' net approval ratings to the first midterm from 2001 to 2025:

There are two immediate takeaways from these charts. First: Hey, what happened 240 days into George W. Bush's presidency? Oh, okay, we'll have to exclude that line from the model I run in a second... But the increase in Bush's approval after 9/11 is very striking to see on that chart. Question for the SIN community: Is the rally around the flag after 9/11 the largest and quickest increase in president or prime minister approval ratings in history, regardless of country?

And second, we see that presidents generally lose support over the first two years of their term, for the most part regardless of what they do. The explanation for this from political science is that every action a president takes ends up marginalizing someone, and as they shift policy in their ideological direction away from the median voter, presidents marginalize increasing shares of the public.

Some presidents do lose a lot more support, however: Gerald Ford, for example, loses the most (net) support for a president ever after he pardons Richard Nixon for his role in Watergate and the subsequent cover-up. Jimmy Carter almost passed Ford up for a hot second in 1978, but recovered ground after the Camp David Accords.

Remember, these graphs are showing you change in net approval ratings over time. You can look at their actual net ratings over time at the data portal.

From here, we could simply take an average of the decline on every day, and apply that to Trump's rating today. I'm just eyeballing it today, but at -10 today, it looks like Trump would be at -20 to -30 by November 2026. We can be more precise than that, though!

Polarization blunts decline, but not by much

One thing to take into account, however, before I set up a model, is that approval ratings have tended to decrease less over time in the polarized era. This is because the higher rate of strong partisanship decreases the pool of persuadable voters who react to events and conditions during a president's term. Higher polarization generally means both a higher floor for approval ratings and a lower ceiling.

There’s no set date for when the extreme polarization in mass opinion we see today "kicks in," so to speak, but a rough definition is somewhere after 2008. So I'm going to take the above data, chop off the first 100 days of a term (since that's already happened for Trump's second term), and set up a model to predict the decrease in approval ratings as a function of (a) the number of days since the 100th day of the term and (b) whether or not the president in question was elected after 2008. I remove George W. Bush's term from these calculations because external shocks on the scale of 9/11 are not expected over the next 2 years — and if one happens, this calculus goes out the window, anyway.

The chart below shows the results from this model. You can read the lines as telling you "If public opinion reacts to Trump like it did to Trump 1, Biden and Obama, his net approval should fall another 22 points by November 2026 — on average, and ignoring uncertainty in the model." The line labeled "Not Polarized" tells you how we should expect, on average, Trump's approval to sink over the next 550 days if the public reacts to him in the same fashion they reacted to George H.W. Bush, Clinton, Reagan, etc.

One interesting note is that approval ratings for presidents elected in the polarized era tend to fall quickly and then level off over time. Maybe this reflected the shrinking supply of people that can change their mind about him, after the big batch of persuadable moderates change their opinion. And yes, this can be modeled with the logit curve.

Since he’s at -9 today, the historical model would put Trump around a -30 rating for the midterms.

However, the uncertainty intervals on these predictions are pretty large, on the order of about 40 percentage points for net approval. The model assigns about an 80% probability to Trump's approval rating being between -5 and -55 on November 3, 2026.

Subjectively, I would be pretty shocked if Trump ended up at -50 by next November. To get there he'd need to be at a 70% disapproval rating, which just seems borderline impossible in our era of high polarization. And while Biden's approval fell by 40 points by his first midterm, he also started out a lot higher than Trump did..

In summary, this exercise tells us a lot about historical patterns in approval and about the limits of history in forecasting public opinion. If I were to guess, Trump's approval rating won't ever go below 30% unless he (a) causes another great depression or (b) breaks political norms or institutions in a manner significantly more serious than the Jan. 6, 2021, insurrection.

If your subjective Bayesian priors for those probabilities are high, then you can push your prediction of Trump‘s approval closer to 30-35. Those predictions do not seem unreasonable to me.