To gear up for your return to work, try Le Grand Continent . The ideas and analyses you won’t find anywhere else. But you’ll see them everywhere soon. In your inbox starting at 8 euros per month
The first striking element of this second term is the absence of the traditional “state of grace.” In the data collected by Ipsos for Reuters, Donald Trump began his second term with 47% approval against 41% disapproval, a net positive of six points. This level, very low compared with previous presidents, was nevertheless slightly higher than at the start of his first term. But very quickly, the dynamic reversed: by February 2025, disapproval exceeded 50% and rose throughout the first half of the year as the tariff war was put in place. By the summer of 2025, however, the approval managed to stabilize around the 40% mark.
The second phase of retreat was triggered at the end of February 2026 by the start of the war against Iran. The conflict did not generate any surge of national unity in public opinion, contrary to what the theory of “rallying around the flag” during wartime would suggest. Worse, Donald Trump’s popularity curve gradually slipped toward 35%, and neither the February State of the Union address nor the commemorations of the United States’ 250th Independence Anniversary in July shifted the trend: in the latest Ipsos measurement conducted from August 21 to 24, 2026, only 33% of Americans approved of Donald Trump’s actions, against 65% who disapproved. Some poll aggregators give slightly more favorable results for the president, but the overall trend of a gradual decline is clear whatever indicators are used.
The comparison between Donald Trump’s current popularity and that of his predecessors is, in fact, harsh. Two and a half months before his first midterms, he reached 42% favorable opinion in 2018, nine points higher than today. Even more striking, at the same stage of his mandate, Joe Biden, despite inflation, stood at 41% approval, a level of support also notably higher than Trump’s current standing (see Graph 2). As for Barack Obama, George W. Bush, and Bill Clinton, their popularity in August preceding the first midterms was 45%, 68%, and 43%, respectively.
This trajectory toward historically low levels contradicts a hypothesis often advanced in the wake of Donald Trump’s reelection: that his 2024 victory signaled the birth of a durable conservative majority. On the contrary, poll trends seem to indicate that voters gave Donald Trump a mandate more limited than some Republicans imagined. Frustrated with Joe Biden’s handling of inflation and the cost of living, they punished the Democratic Party. But there is no indication that they embraced the entire cultural, institutional, and economic program carried by the Trump administration for a year and a half. The backlash now visible in American public opinion could thus reflect the limits of Trump’s 2024 victory.
Independents and Hispanics: the sociology of disengagement
These national figures, however, obscure trajectories that vary greatly across voter categories. By comparing the average of the first three surveys of 2025 with the three conducted in the summer of 2026, one finds that the sharpest declines are concentrated in two blocks of voters who are crucial for the Republican camp ahead of the midterms. On the one hand, independents are disengaging the most: their approval of Donald Trump’s actions fell from 42% to 23% in a year and a half, a 19-point drop, consistent with record lows measured by other institutes among this pivotal electorate. On the other hand, there is also a pronounced retreat among Hispanic voters, where the president’s popularity fell from 39% to 24%, a 15-point decline. The deterioration is also notable among those without a college degree (37%, -11 points) or among people earning less than $50,000 per year (32%, -14 points). All of these categories are particularly exposed to inflation and, in 2024, had contributed to Donald Trump’s election.
Conversely, the erosion is much more contained among those over 55 (42%, -7 points), among non-Hispanic Whites (43%, -8 points), and — as will be discussed — within the Republican electorate (81%, -4 points). The Democrats, for their part, were already near the floor by 2025 and had little room to fall (4%, -2 points), as were African Americans (16%, -2 points).
Finally, a recent territorial modelling shows that the president’s popularity does not exceed 50% in more than a handful of particularly conservative states: 50% in North Dakota, 51% in Wyoming, Idaho, and the Dakotas, 53% in Oklahoma with a best-case of 54% in West Virginia. In contrast, in several swing states that will be decisive for Senate control, Trump’s support base is now very weak: 44% in Ohio, 41% in Wisconsin, 40% in North Carolina and Michigan, 38% in Nevada, Minnesota and Pennsylvania, 37% in Georgia, and 35% in Arizona.
“It’s the Economy, Stupid – again
Ipsos surveys also allow breaking down Trump’s unpopularity by issue. This helps nuance certain explanations often put forward to account for the president’s current difficulties, including the Iran war, the Epstein affair, or immigration policy.
Indeed, it is clearly economic and social issues that constitute Donald Trump’s true Achilles’ heel. On economic management, approval fell from 42% in January 2025 to 26% in August 2026, while disapproval rose from 36% to 66% over the same period. On living costs, the collapse is even more brutal: only 21% now approve of the administration’s actions, while 70% disapprove, whereas at the time of Trump’s inauguration for his second term those figures stood at 35% and 41% respectively. And while these affordability issues are by far the area where Americans express the greatest displeasure, they are also the topics they consider most important.
Levels are less catastrophic and the erosion is less pronounced on immigration, with 38% of Americans currently approving of Trump’s actions in this area, versus 46% at the time he returned to the White House, a relatively modest 8-point drop in a year and a half. The approval levels are similar for security issues, with 34% approving the president’s actions in this domain. This situation is all the more striking as it forms an inverted mirror of Trump’s first term, when public opinion praised his economic actions but was highly hostile to his immigration and security measures. Similarly, the share of opinion that approves action in the face of Iran is low (25%) and has declined since the start of the conflict (33% in March), but the levels and trends are less worrying than on economic issues.
The survey details show how much the independent electorate, which helped propel Donald Trump to victory in 2024, has turned away from the president. Only 28% of independents approve Trump’s action on immigration and 24% on insecurity, with even lower figures about the Iran conflict (15%), the economic situation (15%), and the cost of living (12%). On these last two topics, disapproval exceeds 70% (71% and 75% respectively), with independents expressing discontent with the Republican administration’s handling of these social issues.
A loyal MAGA base, but a moderating Republican electorate drifting away
The question that most directly conditions the November vote remains: what will become of the president’s electoral base?
On this point, Ipsos data cut both ways. On one hand, the loyalty of the electorate identifying with the MAGA movement remains massive and unwavering, with a rate of approval of Donald Trump ranging from 98% to 97% since taking office. But this stability masks an internal transformation within the Republican Party: the core MAGA Republicans represent an ever-smaller share of the conservative electorate. The proportion of Republicans who identify to some degree with this movement fell from 62% in February 2025 to 48% in July 2026, while those who identify as “strongly” MAGA dropped from 36% to 22%.
Conversely, it is precisely among Republicans further away from the MAGA movement that presidential approval has declined: it fell from 76% in February 2025 to 68% in July 2026, with a marked acceleration since late 2025.
The midterms, a referendum on the president’s actions?
These high levels of unpopularity, however significant, are not by themselves sufficient to predict the November election, as United States midterm elections are by no means reducible to a single national indicator: incumbents’ strength and popularity, House seat distribution, Senate battleground states, and the Democratic Party’s image could neutralize part of the national climate unfavourable to Republicans. Yet it remains true that the president’s party almost always loses seats in midterms: since 1862 and the emergence of the American two-party system around Republicans and Democrats, 38 of 41 midterm elections have resulted in a loss for the president’s party in the House of Representatives.
This regularity lies at the heart of an old debate in American political science. According to political scientist Edward Tufte, midterms are a moment for voters to assess the presidential record and the administration’s economic performance, effectively functioning as a referendum on the president: since voters cannot sanction him directly, they do so through the candidates from his party.
Since popularity polls began, the correlation between the president’s approval rating and the magnitude of his party’s defeat in midterms has been fairly clear. But the 2022 election also showed the limits of the model: with Joe Biden at roughly 40% favorable, a level that suggested a Democratic landslide, the anticipated Republican wave in Congress faded to a gain of only nine House seats—a disappointing outcome given the president’s unpopularity.
This purely referendum reading must therefore be nuanced. Another scholar, James Campbell, recontextualized this theory in a broader model that combines two mechanisms. The first is the “presidential coattail” effect: some of the less politicized voters who helped secure the presidency do not rejoin the political fray two years later, when the stakes seem less decisive. This coattail effect of the winning presidential candidate on his party’s congressional candidates was well known to political leaders before it was formalized by scholars, as illustrated by a 1930s interchange between the New York Democratic leader Hymie Schorenstein and a young candidate anxious about a local election:
– Son, said Hymie, have you ever seen a ferry come to the dock?
– Of course,
the young attorney answered.
– Then you’ve noticed that when the ferry arrives, it carries debris, pieces of wood, garbage, etc.?
– Yes.
– Then, my boy,
replied Schorenstein, stop worrying. President Roosevelt is our ferry, we are the debris.
But this “coattail” phenomenon has a reversal: two years after the presidential election, the midterm contest is perceived as less decisive and turnout drops sharply, particularly among the less politicized voters. This leads to a contraction of the electoral coalition that had carried the president’s party into Congress. In fact, the correlation between the president’s score and the decline of his party at midterm exists: the larger the victory at the presidential election, the more the congressional coattail is felt, but the greater the collapse two years later.
As for the second mechanism highlighted by James Campbell to explain the president’s party’s systemic failure at midterms, it is indeed the referendum effect. But for him, it primarily affects independent voters. Two years after the presidential election, their choice is mainly based on their judgment of the president’s record, with the sanction vote tendency amplified by the propensity of dissatisfied citizens to vote in midterms more than those who were satisfied. This mechanism is particularly important in a system where polarization has made Democratic and Republican voters more homogeneous: the few voters still available between the two camps can swing the result.
Toward a severe Republican defeat?
The James Campbell model thus distinguishes two mechanisms.
The retreat of the presidential coattail explains why the governing party almost always loses in midterms, while the referendum aspect of independent voters determines the scale of that loss. Both lead to the contraction of the coalition formed two years earlier: disengaged voters abstain, while a portion of independents shifts to the opposition. Yet Ipsos data suggest that these two dynamics are at work today.
The first warning sign is the disengagement of independents, who had been one of the drivers of the 2024 Republican victory: their approval of the president’s actions has massively fallen. It is precisely among this electorate that Campbell’s referendum mechanism is expected to operate most strongly: less attached to the parties, independents are the most likely to translate their judgment of the president’s record into a vote against the candidates of the governing party. And on the issues that matter most to them—starting with the cost of living—the verdict on Donald Trump is especially harsh. The risk of a sanction vote from this decisive electorate is therefore very high.
More fundamentally, the problem of Trump’s fragile coalition in 2024 arises. His victory relied, even more than in 2016, on widening his appeal beyond the traditional conservative core by gaining ground among Hispanic voters and lower-income groups. It is precisely in these sociodemographic segments that his popularity has fallen the most. This does not necessarily mean they will vote en masse for Democrats, but even if their departure takes the form of returning to abstention, it remains another handicap for Republican candidates. In both cases, this dynamic would weaken a Republican coalition whose broadening seems less solid than imagined in the wake of the election.
These fragilities are compounded by a mobilization problem specific to midterms. As noted, turnout declines between presidential and midterm elections tend to eliminate a portion of the less politically engaged voters who had helped secure the presidency two years earlier. This phenomenon is traditionally most noticeable among voters satisfied with the president’s record, who are less likely to go to the polls than those who are dissatisfied. At this stage of the campaign, 53% of Democrats say they are enthusiastic about voting in the 2026 elections, versus only 44% of Republicans. At the same time, the growing share of Republicans who do not identify with the MAGA movement appears more critical of Donald Trump and could express this discontent by staying away from the polls. The differential mobilization could thus be an additional advantage for Democrats.
A final element could further worsen the scale of the Republican retreat: by personalizing political life and nationalizing the election issue, Donald Trump has made it harder for Republican candidates to distance themselves from his personality and record or to center their campaigns on local issues. This strategy could indeed mobilize the MAGA core whose loyalty remains unwavering, but it could also facilitate the sanctioning of a widely unpopular president by his party’s candidates.
The sole president’s unpopularity therefore does not fully capture the danger facing Republicans, because each of the mechanisms that powered Trump’s victory in 2024 now appears to be reversing: independents are turning away from the president, several recent segments of the Republican coalition are drifting away, Democrats enjoy an enthusiasm advantage, and Trump himself has made the choice to turn the midterms into a referendum on his person.
There are of course several weeks left before the election, and midterms are never wholly determined by the national climate. But if these trends persist, the question will soon be less about whether Republicans will suffer the usual midterm backlash and more about how far the 2024 coalition will retract. Perhaps this is where the real paradox of this second Trump presidency lies: never had the Republican Party been so closely identified with its leader, yet never had this identification exposed it to sanction alongside him as much as now.