Four polling organizations, one week of fieldwork, three surveys
Between July 23 and 27, 2026, four reputable polling firms—Quinnipiac, AP-NORC, CNN/SSRS, and Economist/YouGov—conducted surveys across the country during the same time frame. Four different questionnaires, four different samples, four different methods. And four results that align: a 32% approval rating from Quinnipiac, 33% from AP-NORC, 34% from CNN/SSRS, and 34% from Economist/YouGov. When disparate survey methods yield the same result, it’s no longer a measurement anomaly. It’s a fact.
The Survey Period: July 23–27
I describe this moment as a building whose plumbing is being inspected: the walls are still standing, no one disputes them, but the pressure gauges tell a different story. The pressure drops from floor to floor. The Quinnipiac poll, released on July 29, set a record low for that institute during Donald Trump’s two terms in office. AP-NORC, released on the 30th, hit the lowest point of the second term in its own series. CNN/SSRS matched the lowest reading recorded at the end of the first term. YouGov, which released its own analysis on August 1, recorded a net rating of −28—approval minus disapproval—a record low in its series.
Three internal records, one convergence
Each pollster compares its figures to its own historical data, never to that of its neighbors. That is precisely what makes the convergence so striking: three independent series, calibrated differently, are all simultaneously falling to their lowest points. The Republican base has not crumbled—the declared partisan base remains massive, as we’ll see later. It has lost momentum in three areas at once: independents, non-MAGA Republicans, and the enthusiasm of its own base. And the three leaks are not in the same place and cannot be fixed with the same tool. This entire post hinges on that distinction.
What “record” means—and what it doesn’t mean
The word “record” demands immediate discipline, because it is the exact point where accuracy is usually lost. The 32% from Quinnipiac is an all-time low in Quinnipiac polls, spanning two terms. It is not the lowest figure in U.S. presidential history: Newsweek points out that Harry Truman dropped to 32% in December 1952 and Richard Nixon to 24% before leaving office. The comparison sheds light without drawing conclusions: it marks a low point, but it does not predict how far the decline will go or whether there will be a rebound.
The Short Memory of Poll Series
YouGov adds a useful point of reference: at the same point in his first term, Donald Trump’s net rating was −9. Today, it stands at −28 in the same series. The difference between the two terms is therefore not merely a commentator’s impression; it is a gap measured by the same tool, at the same relative dates, using the same method. That’s the kind of data that interests me: data that compares an administration to itself.
The Honest Use of Precedents
Historical precedents are guiding lights, not fixed tracks. December 1952: 32% for Truman; 24% for Nixon before he left office—two documented benchmarks, two contexts, no universal rule. The host of the Fast Politics podcast mentioned a “benchmark” for outgoing presidencies; I prefer Newsweek’s quantified and dated benchmarks to any symbolic line. A floor has never dictated a trajectory: it merely indicates how deep one is when one begins to dig or climb back up.
963 registered voters, 1,165 adults: why the numbers don’t add up
Before interpreting the data, we must count accurately. Quinnipiac surveyed 963 registered voters via live calls to landlines and cell phones, with a margin of error of ±4.1 points, including the sampling error. AP-NORC surveyed 1,165 adults through its AmeriSpeak probability panel, with a margin of ±3.7 points. The CNN page, however, does not list either the survey population or the margin of error: the figures of 1,225 adults and ±3.2 points appear only in editorial summaries of the poll, and I treat them as such. YouGov does not publish either the sample size or the margin of error on its analysis page—I refrain from declaring a “significant” gap based on that information.
Populations that cannot be compared line by line
Registered voters are not adults in general; an online panel is not a series of telephone calls. That is why Republican approval stands at 76% in the Quinnipiac poll and 78% in the CNN poll: two methods, two populations, no statistical scandal. We report the margin; we do not average it. The temptation to fabricate a single figure—“about 77%”—would create a fictitious precision that neither pollster has measured.
Subsamples: The Area of Greatest Caution
The more a sample is broken down, the wider the margin of error becomes. AP-NORC documents this: ±6.0 points for its Republican subsample, ±5.4 for the Democratic one. Quinnipiac does not publish margins by subgroup, which automatically makes its partisan breakdowns less reliable than its overall figure. When the AP reports Republican approval of the war at around 61% compared to 71% in June, it itself notes that this decline, by that margin, remains slight. This rigor on the part of the AP regarding its own data is precisely the standard this post sets for itself throughout.
Independents: From 34% to 22% in One Month
The first drop—and the sharpest. On June 24, Quinnipiac reported a 34% approval rating among independents, compared to 59% disapproval. On July 27, in the same poll series: 22% approval versus 67% disapproval. Twelve approval points evaporated in a month among the voters who decide close races—and eight disapproval points gained in the same period.
YouGov’s net rating: a different tool, but the same trend
YouGov confirms the trend with its own tool: a net rating of −53 among independents, compared to −4 at the start of the term. Forty-nine net points lost since January, in a sample entirely distinct from Quinnipiac’s telephone survey. The same firm measures other converging declines: 33% support among 45- to 64-year-olds versus 63%, a net rating of −29 where the start of the term showed +15; and non-college-educated whites—the coalition’s historic base—have averaged a net rating of −6 across three waves, compared to +27 in the first three weeks of the term. Two surveys, two populations, one trend.
A data-driven assessment, not a projection
The Associated Press interviewed Mikal Baker, a 47-year-old independent from Georgia, as part of its poll: “I don’t see the benefit… I can’t say it will ever be over with.” I quote this man because he exists, he is named, and his statement captures exactly what the number measures: not anger, but disengagement. The disaffection among independents need not be dramatized. It has been measured—twice—using two different methods.
Non-MAGA Republicans: from +84 to +12
A second trend, the least discussed—and, I believe, the most significant. YouGov distinguishes, among Republicans, between those who identify as MAGA and those who do not. The latter overwhelmingly approved of the president at the start of his term: a net rating of +84. Today, they stand at 55% approval versus 43% disapproval—a net of +12. Seventy-two net points lost within the coalition itself, among voters who vote Republican without wearing the MAGA hat, who fund local campaigns, and who hold sway in suburban counties.
A self-identified group, a necessary caution
The “MAGA” label is a self-identification, not a membership card: the boundary between the two groups shifts with the political climate, and YouGov does not publish a margin of error by subgroup. I therefore remain cautious about the exact scale of the movement. But the trend, measured by the same polling firm across several successive waves, is clear and consistent: the quietest segment of the coalition is the one cooling off the fastest.
Why This Number Is Strategic
A lost independent voter can be offset by an over-mobilized base; that’s the classic midterm arithmetic. But a non-MAGA Republican who shifts from enthusiastic support to lukewarm approval isn’t switching sides—they’re changing their behavior. They contribute less, campaign less, and sometimes vote—or not at all. This isn’t a defection; it’s an evaporation. And almost no one talks about it, because this segment doesn’t protest, doesn’t shout, and doesn’t generate media buzz. The YouGov data is, to date, the only public evidence of this cooling off.
Enthusiasm, not approval: 19% versus 38%
Third sign of waning support—and the most strategic one. The July 29 CNN/SSRS poll asks a question distinct from approval ratings: Republicans’ enthusiasm for the remainder of the term. The answer: 19%, down from 38% in the spring. Half of the reported enthusiasm has vanished in a single season, even among voters who continue to approve of the president. In the same CNN poll, strong approval drops to 15%—a series low—while strong disapproval rises to 50%, a record high. The opposition is on fire, while the base is cooling off: this is the very definition of a mobilization gap.
What intensity measures—and what it doesn’t
Declared enthusiasm is not actual turnout: it is an indicator of attitude, never a tally of ballots, and I refuse to turn it into a prophecy. But Tim Malloy, the Quinnipiac analyst, draws from his own data a well-documented and frequently cited warning: “A warning sign is blinking red for the GOP. With turnout so crucial in midterm elections—which many voters tend to ignore—the Democrats seem more motivated to show up than Republicans.” The warning has a name, a source, and a methodology. I’m reporting it; I’m not declaring it.
The question of a better quality of life, posed by the same institute
Malloy summarizes a second finding from the same survey: to the tried-and-true question—“Are you better off than you were a year ago?”—the recurring answer is no. Quinnipiac reports that 49% of registered voters say they are in a worse financial situation than a year ago, 71% rate the economy as “not so good” or “bad,” 59% believe it is getting even worse, and 88% rank inflation among the serious problems. Enthusiasm doesn’t die from an opinion—it dies from a grocery receipt.
The MAGA core hasn’t budged—and that’s the real problem
Here’s the counter-indicator, and it’s massive: MAGA Republicans post a net score of +89 on YouGov. Nothing has changed. Overall Republican approval remains at 76–78%, depending on the poll. A coalition whose core remains unchanged is not a coalition that’s collapsing, and anyone who writes otherwise is distorting the data. I’m saying this bluntly because half of this week’s analyses have claimed otherwise anyway.
An engine running without torque
But let’s flip the image: an engine can run at full throttle without transmitting any power to the wheels. That “throttle” is the core support at a net +89 in the YouGov series. The torque is what reaches the road: enthusiasm at 19% in the CNN poll, independents at 22%, and non-MAGA voters at +12. The scenario that worries strategists isn’t the defection of the core base—it’s that the intact core base is becoming the only part that’s functioning, and it’s imposing its demands on the rest of the machine.
The Closing Loop
The more the core holds firm, the more it defines the message; the more it defines the message, the more the other three channels dry up; the more they dry up, the more indispensable the core becomes—and the more its relative weight increases in every decision. It’s a feedback loop, not a conspiracy: no one designed it, but everyone is caught up in it. A party can certainly win elections using this mechanism when overall voter turnout is low. It may also discover, one Tuesday in November, that the engine’s power was never transmitted to the wheels.
Why Three Divisions Require Three Incompatible Messages
Let’s revisit the three leaks one by one, along with their respective remedies. To win back an independent with a 22% approval rating, you need a results-oriented message: falling prices, wages, an end to conflict. To win back a non-MAGA Republican who has gone from +84 to +12, you need governmental normalcy, predictability, and institutional calm. To remobilize a base whose enthusiasm has fallen from 38% to 19% in the CNN poll series, on the contrary, you need adrenaline, a fight, and a sense of identity. Three locks, three keys—and only one hand to hold them all.
The Asymmetry of Remedies
Thematic approval ratings confirm the impasse
Quinnipiac’s issue-by-issue breakdowns tell the same story in detail: 34% approval on the economy, 34% on foreign policy, 28% on Iran, 40% on immigration, and 37% on trade. No issue exceeds the 40% mark; so there is no obvious ground on which to shift the conversation without friction. YouGov, for its part, measures a net rating of −29 on the handling of global trade. When all the issue-based avenues are difficult, the choice of message is no longer a communication choice: it’s a trade-off between voter bases.
You can’t reassure a core base and an independent with the same sentence.
“Boxed in”: A journalist’s phrase, not a verdict
On July 31, speaking on the Fast Politics podcast, NOTUS journalist Evan McMorris-Santoro summed up the month with a single image: the Republican candidates are “boxed in”—stuck. They can’t distance themselves from the president, he said, nor can they align with him, because that’s toxic for the general electorate. He adds that the month is ending “in absolutely dire straits” and that there is “no end in sight” for these negative numbers. These are his words, dated, recorded, and timestamped—at 1 min 31 s and 2 min 01 s into the episode. I’m reporting them for what they are: a journalist’s assessment, not a measurement.
The nuance that the recap omitted
The same guest, in the same episode, slipped in a caveat that almost no one picked up on: “It’s a weird time with polls—what to believe, what not to believe.” ” A journalist who publicly doubts his own material—that was just as worthy of being quoted as his soundbites. And his mention of a Republican summer plan centered around July 4—celebrations that were supposed to kick off the season—is based on anonymous strategists cited verbally: a reported statement, never an established fact, and I classify it as such.
The Basket and the Eggs
His most accurate analogy comes later in the episode: “all those eggs are in one basket… they don’t have a backup plan.” Another opinion-based statement. But one grounded in actual figures: this is precisely the thesis that the data in this post allows us to test. The difference between quoting and endorsing is the backbone of this text.
The party has no fallback option: 29% for congressional Republicans
If the “Trump” variable deteriorates, the natural reflex would be to promote something else: Congress’s record, the local brand, or the district candidate. Yet Quinnipiac measures approval of congressional Republicans at 29%, compared to 63% disapproval. The fallback brand is lower than the main brand. A party whose flagship product stands at 32–34% and whose substitute product stands at 29% has no backup option: it has only one option, and that is the very meaning of the term “systemic risk” used here.
The alternative isn’t popular either
This clarification is essential, because it puts a stop to the triumphalism on the other side: Democrats in Congress have a 23% approval rating in the same poll, six points below their Republican counterparts. No one is popular; this is a landscape of widespread distrust, not a shift in affection. What the data show is not a transfer of political support—it’s a difference in motivation between two equally disillusioned electorates. A crucial distinction: you don’t win these kinds of elections by winning people over; you win them by getting your base to turn out on a Tuesday in November.
Silence as a Data Point
One final figure from the same series deserves attention: 10% of registered voters surveyed by Quinnipiac no longer express an opinion on the president, compared to 5% a month earlier. The number of those who remain silent has doubled. I’m careful not to read any emotion into this—the data doesn’t say why these people are staying silent. But growing disengagement is neither support nor opposition: it’s a reservoir of volatility, in either direction, and both parties would do well to treat it with equal seriousness.
Gas Prices as a Political Variable
Why is the pressure easing now? July polls point to a specific factor: the cost of living, and above all, fuel. In a CNN poll, approval of the administration’s handling of gas prices has dropped to 21%, and 74% of respondents report at least some discomfort related to rising gas prices, up from 63% in March; 71% believe the president has not gone far enough to lower current prices, up from 58% a year earlier. In the AP-NORC poll, 72% consider it “extremely or very important” to prevent fuel prices from rising, up from 67% in March, and about four in ten adults now cite gas prices as a major source of stress, up from three in ten in February.
Three Voters, Three Perspectives
The AP puts faces to its columns of numbers, and I’m quoting them exactly as they spoke. Robert McClary, a 52-year-old Republican from Ohio, addresses the president: “You’ve kind of gotten yourself buried in this mess.” Ryan Sloan, 39, from Florida: “I just wish I’d had more transparency.” ” These are examples, not a sample. But their three tones—reproach, a demand for explanation, and Baker’s disengagement mentioned above—exactly mirror the three divides that the percentages measure.
The Cost of Loyalty
This is where the most important quote of the month comes from, gathered by the Associated Press from Jennifer Norton, a 71-year-old Republican from Crossville, Tennessee, on a modest income, annoyed by the price of roast beef: “I remember Mr. President Trump is a brilliant businessman.” She grows impatient, she says, then comes to her senses. This voter isn’t leaving her camp. She’s paying to stay there, and she says so in her own words. No poll captures better than this quote what a coalition under pressure looks like: loyalty remains, but the cost of loyalty is rising. No shame, no awakening, no blindness—nothing of the sort is in the source.
“I remember Mr. President Trump is a brilliant businessman.”
What the aggregate shows and what the three polls do not: D +6.4
Let’s switch tools, because the November election won’t be decided by presidential approval ratings. The generic vote aggregate compiled by Silver Bulletin—a weighted average of dozens of polls, adjusted for polling house effects, prioritizing samples of likely voters, and weighted by each pollster’s rating—gives the Democrats a +6.4-point lead as of July 31. The trend is telling: +7.1 in early June, a dip to +5.9, then a rebound to +6.4. While the president’s approval rating plummeted to record lows, the Democrats’ lead in Congress did not widen. It fluctuated.
What the aggregate includes—and what it smooths out
The recent polls included in this average range from D +7 in the Fox News and Quinnipiac polls to D +11 in the Emerson poll, with D +8 and D +10 elsewhere: the variation among polling firms is real, and it is precisely the aggregate’s job to smooth it out. But smoothing is not a poll. A poll has a margin of error. An aggregate has no sampling margin of error: it inherits the biases of its constituent polls and its weighting choices, and it does not automatically translate into seats—districting, incumbents, and candidates do the rest.
Three metrics, three interpretations
Generic voting is not the same as approval: Quinnipiac publishes the two separately—32% approval on one hand, and a generic vote of 48% for Democrats versus 41% for Republicans among registered voters on the other, with 11% undecided. Confusing these metrics is to draw a conclusion that the data does not support. The drop in approval ratings is dramatic; the Democratic legislative advantage, measured by the aggregate, remains real but limited. The two statements hold true together.
The rebound scenario, as described by those who envision it
Honesty requires giving full weight to the counter-hypothesis, and it is a strong one. Newsweek interviewed four prominent academics. David Stebenne of Ohio State points out that presidents often hit their approval rating floor in their sixth year and that late-term rebounds do occur. Matthew Dallek of George Washington University considers a rebound “certainly possible,” with a conceivable return to the low 40s—but only under three cumulative conditions: an end to the war, prices under control, and expanding employment. Meena Bose of Hofstra emphasizes that there are no guarantees one way or the other; Thomas Whalen of Boston University points out the opposite risk.
The Figures Marking the Recovery
The same Newsweek analysis provides the milestones along the way: Quinnipiac’s measure of economic approval fell from 42% in January to 34% in July; support for military action against Iran dropped from 40% in March to 34%. A rebound would need to reverse these two trends, either in that order or simultaneously. It’s not impossible—it’s conditional, and the conditions are dictated by the experts who believe in it the most.
Late Mobilization and Polarization
Two mechanisms must be added that the July lows do not capture. First, late mobilization: a base at a net +89 on YouGov can be reactivated in a matter of weeks if the campaign becomes existential again—dormant intensity is not dead intensity. Second, polarization itself: it caps declines just as it caps rebounds, and it brings disappointed voters back into the fold as soon as the opponent becomes the focus again. Finally, the variation among polling firms is real—some online panels show significantly higher numbers—but I haven’t verified these figures at the primary source, so I won’t cite them.
The lines this post refuses to cross
Three claims are circulating, and I reject them all. “The GOP has lost its base”: false; YouGov’s MAGA net rating is +89, and Republican approval exceeds three-quarters. “These numbers signal defeat”: no—approval ratings aren’t generic votes, generic votes aren’t seats, and none of this predicts the outcome in November. “It’s irreversible”: Three of the four experts cited by Newsweek describe explicit conditions for a rebound, and the aggregate generic vote has already narrowed once this summer before widening again.
Rejecting Psychological Analysis
I also refuse to psychologize: neither the president’s state of mind, nor a party’s supposed fear, nor the imaginary “awakening” of a Tennessee voter. The data says what it says—levels, margins, trajectories—and remains silent on the rest. Attributing emotions to ten million voters is the laziest shortcut in political commentary, and it creates precisely the misreadings that this post seeks to correct.
A Documented Absence, Without Speculation
One absence, however, is worth noting, because it is documented in the July 31 episode: the president’s team is not releasing its own polls in response, the guest journalist observed. In a previous election cycle, flattering internal figures would have been released within the hour. I note this absence. I draw no conclusions from it—neither embarrassment, nor confirmation, nor strategy. An absence is an absence: it is noted; it is not romanticized.
The president’s team is not publishing its own polls.
One lever, three locks
Here is what July 2026 has established, piece by piece, source by source. The stated Republican base holds steady: 76% in Quinnipiac, 78% in CNN, a net +89 among YouGov’s MAGA voters. And around this base, everything that wins elections is simultaneously eroding: independents at 22%, non-MAGA Republicans at +12, and grassroots enthusiasm at 19% in the CNN poll. The structure is still standing, but the pressure is dropping; the engine is running, but it’s no longer generating torque. Neither of these two statements contradicts the other: it is their coexistence that makes this a significant development.
The question that remains on the table
A party that has staked its entire electoral strategy on a single variable discovers that this variable breaks down into three parts, and that fixing each part damages the other two. That is the paradox of the summer: the problem isn’t that the lever is weakening; it’s that there’s only one lever for three locks—and the hand holding it cannot split itself in two. The only honest prediction can be summed up in one sentence: fourteen weeks is a long time, late mobilization does happen, polarization brings back the party’s base, and none of these numbers will vote in November in place of the voters. As for Ms. Norton, she has already explained why she’s staying. The price of roast beef may reveal the rest.
Methodology: Margins, Polling Areas, Populations, and Blind Spots
This post is based on four surveys conducted simultaneously or nearly simultaneously, from July 23 to 27, 2026. Quinnipiac: 963 registered voters, live calls, margin of error ±4.1 points, including plan effect. AP-NORC: 1,165 adults, AmeriSpeak probabilistic panel, ±3.7 points, Republican subsample at ±6.0. CNN/SSRS: methodology not displayed on the page viewed; field data being circulated as reprints are marked as such. Economist/YouGov: Neither the sample nor the margin of error is published on the analysis page—I therefore refrain from describing any subgroup difference as “significant,” and I present its averages across three waves as averages, never as single measurements.
What I Was Unable to Verify
Columnist’s Transparency Box
Editorial Stance
This text is an opinion piece: a reasoned, bylined view that assumes a specific interpretation—namely, that the Republican Party is structurally dependent on a single, deteriorating variable. It does not predict any election results, offer any psychological analysis, or endorse any specific recovery framework: the phrases “dire straits” and “boxed in,” as well as the image of the basket, are Evan McMorris-Santoro’s; the warning about mobilization is Tim Malloy’s.
Methodology and Sources
Seven sources, all actually consulted on August 3, 2026: two direct publications by polling organizations (Quinnipiac, YouGov), two polls reported by their sponsors (AP-NORC, CNN/SSRS), the source recording of the quoted statements, an analysis by named experts (Newsweek), and an aggregate with published methodology (Silver Bulletin). The starting point for the research was an editorial aggregation by AlterNet dated August 2; it was never used as evidence, and its approximations—including the AP-NORC poll’s sample population, which was incorrectly presented as registered voters—were corrected using the primary sources. Polls, aggregates, generic voting, and approval ratings are systematically distinguished.
Nature of the Analysis
Facts are sourced and dated; interpretations—the image of the building, that of the engine, the asymmetry in repairs—are my own and presented as such. Poll data is time-sensitive: the figures must be revalidated prior to publication if more than 72 hours have elapsed since August 3, 2026.
OP-ED — 76% Support, 19% Enthusiasm: The Paradox Trapping the GOP
Sources
Sources primaires et officielles
- Quinnipiac University Poll, communiqué n° 3964, 29 juillet 2026 — 963 électeurs inscrits, terrain 23-27 juillet, ±4,1 points ; approbation 32 %, sous-groupes, vote générique, citations Malloy.
- Associated Press, sondage AP-NORC, 30 juillet 2026 — 1 165 adultes, ±3,7 points ; approbation 33 %, données Iran et carburant, répondants nommés dont Jennifer Norton et Mikal Baker.
- YouGov, analyse du sondage Economist/YouGov du 25-27 juillet, 1ᵉʳ août 2026 — net −28, MAGA +89, non-MAGA +12, indépendants −53.
- CNN, sondage CNN/SSRS, 29 juillet 2026 — approbation 34 %, approbation forte 15 %, désapprobation forte 50 %, enthousiasme républicain 19 % contre 38 %.
- Fast Politics avec Molly Jong-Fast, épisode avec Evan McMorris-Santoro, 31 juillet 2026 — enregistrement source des citations « dire straits », « boxed in », « no end in sight », avec leurs horodatages.
Sources secondaires et analyses
- Newsweek, « Could it rebound? », 31 juillet 2026 — scénarios de rebond par quatre universitaires nommés, repères historiques Truman et Nixon.
- Silver Bulletin, moyenne du vote générique, mise à jour du 31 juillet 2026 — D +6,4, trajectoire +7,1 → +5,9 → +6,4, méthodologie de pondération publiée.
This content was created with the help of AI.
