Subscribe now

For the better part of two years, the story about the Democratic Party’s image has been simple and unflattering: voters don’t like it. The party’s favorability rating has been underwater in nearly every major poll since the middle of 2024, and the party took another hit during last fall’s shutdown fight. Our new April 2026 Strength In Numbers/Verasight poll confirms this broad pattern: Democrats are underwater at 45% favorable to 48% unfavorable, a net rating of -3.

But Republicans are doing much worse. Our survey finds the Republican Party at 39% favorable and 55% unfavorable — a net of -16, with 42% of Americans saying their view of the GOP is “very unfavorable.” That number is 12 points higher than the equivalent figure for Democrats (30%).

We originally designed this survey to unpack a puzzle of party popularity. Last summer, our analysis of favorability polling found many left-leaning Democrats were critical of their party, dragging down its overall favorability rating even though many of these critics were still planning on voting for the party. We wanted to know whether this was still the case. It is, and we asked some specific questions to voters about what the party would need to change for its rating to improve.

But our polling also shows that this puzzle — Democrats being more unpopular than Republicans because of intra-party disaffection — has now resolved itself. When we ask voters whether they think Democrats or Republicans possess certain character traits, Democrats win on nearly every one we tested. Independents dislike Republicans roughly three times as much as they dislike Democrats. Even voters who say they have an unfavorable view of the Democratic Party often plan to vote for Democrats anyway. And when Democrats’ own voters complain about their party in their own words, the complaint is not that Democrats are too liberal or “weak and woke”, it’s that they’re not fighting hard enough, particularly against Donald Trump.

Here are the headline findings:

• Party favorability: 45% favorable / 48% unfavorable for Democrats (net -3); 39% / 55% for Republicans (net -16). The GOP’s “very unfavorable” number (42%) is 12 points higher than the Democrats’ (30%).

• Character vs strength: Democrats lead Republicans by double digits as the party that is tolerant (D+26), respects democratic institutions (D+16), cares about people like you (D+14), is honest (D+12), and looks out for the middle class (D+11). Republicans lead on willingness to fight (R+4), clear messaging (R+8), getting things done (R+7), and strong leadership (R+11).

• 2026 House vote intent (among all adults): In a modified 2026 congressional vote question that tests vote intention and party disapproval at the same time, 47% of adults say they’ll vote for a Democrat, with 29% doing so happily, while 18% are unhappy with the party but voting Dem anyway.

• Unhappy but still voting Democratic: Among voters with an unfavorable view of the Democratic Party, 16% say they’ll vote Democratic in 2026 anyway. Among the “somewhat unfavorable” slice, 41% still plan to.

• Democrats’ biggest complaint about Democrats: In their own words, many say leadership is weak and not fighting hard enough against Donald Trump.

The toplines and crosstabs for this poll can be found on the Strength In Numbers polling portal.

Subscribe now

Both parties are underwater with U.S. adults

Let’s start with the topline. 45% of U.S. adults have a favorable view of the Democratic Party — 15% very favorable, 30% somewhat, against 48% unfavorable, split 18% somewhat and 30% very. Six percent don’t know. The Republican Party’s numbers are worse on both sides of the ledger: 39% favorable (16% very, 23% somewhat) to 55% unfavorable (13% somewhat, 42% very). Six percent are unsure.

Two things stand out. The first is how close Democrats are to breaking even. A net -3 is within the poll’s margin of error and the best public-image reading we’ve measured for the party in any Strength In Numbers/Verasight wave this cycle.

The second is the asymmetry in intensity. 42% of Americans say their view of the Republican Party is very unfavorable — 12 points higher than the same number for Democrats. These are groups of voters that really detest the other side and are unlikely to be moved by campaign messaging or news coverage about the current state of politics. Democrats have an edge, in other words, with devoted voters.

Partisans like their own brand. Independents hate Republicans more.

Split the numbers by party ID and the reason Republicans are doing so poorly in the topline really becomes clear. Democrats rate their own party at net +60, while Republicans rate theirs at a net +61. This is a (predictable) reversal from 2025: there is no evidence here that Democratic voters are disenchanted with their own brand in some way that Republicans are not.

Instead, the real asymmetry is among independents.

Self-identified independents (with leaners pushed toward the parties) rate Democrats at net -9. That’s meaningfully negative, but also not the worst number I’ve ever seen. By contrast, independents rate Republicans at net -30.

Many voters who dislike Democrats still plan to vote for them

If you look at the voting behavior of Americans by their view of the parties, you’ll see why I was originally so dismissive of the discourse about this polling in 2025. While the Democratic brand has mostly recovered since then (at least in our polling), there are still a lot of voters who rate the party poorly. A plurality of them, however, say they’ll vote Democratic, and that was the case in 2025, too.

We asked every respondent to pick the statement that best described their 2026 House vote intent. Seven options ranged from “I support the Democrats and will probably vote for them” to “I’m unhappy with the Democrats, but will still vote for them” to “I probably won’t vote regardless.” (I would take the exact result here as a rough measure of voting intention, since the question asks about how people feel about the party more broadly too, and is very different results-wise from this survey’s generic ballot question (D+7).

Adding the two Democratic buckets across all adults gets you to 47% who say they’ll vote for a Democrat in 2026 — 29% enthusiastically, plus another 18% unhappy but voting Democratic anyway. The two Republican buckets together come to 34%: 24% who plan to vote Republican regardless, and 10% who are unhappy with Democrats specifically and will cast a Republican ballot because of it.

Break the question out by party, and those old stressors on the Democratic Party brand show up again. Republicans are comparatively much more consolidated in the committed lane than Democrats are.

But, as I argued last year, this has little to no bearing on election results. Few true swing voters say they would vote Republican because they’re unhappy with Democrats. Very few self-described Democrats say they would vote red because they’re unhappy with the party; instead, that’s all coming from the right:

Another way to look at this is to break down the 2026 vote question by the favorability rating of the parties:

Of the 735 respondents who say they have an unfavorable view of the Democratic Party, 16% say they’ll vote for Democratic candidates next year anyway — “unhappy, but will still vote for them,” while an additional 25% are committed to Democrats and like what they’re doing. with the unhappy-but-voting-anyway group). Just 35% plan to vote Republican.

A “somewhat unfavorable” rating of the Democratic Party is not a Republican vote. For a meaningful share of American voters, the favorability rating question is a way to register complaints against a party you are committed to.

Understanding the parties’ brand weakness

Favorability rating is a common question in polling, but it suffers from being overly general. What we really want to know is what voters think defines each party's brand.

So we also asked respondents to tell us how well each of a selection of traits describe either party. As expected based on the favorability numbers, Democrats mostly win on the positive traits, while voters say Republicans possess more of the negative ones.

Voters say Democrats are more tolerant, democratic, and likely to care about people like them than say that of Republicans. 62% of Americans say the trait “tolerant of different types of people” describes the Democratic Party well, against 36% who say so of the GOP — a 26-point gap. Democrats also lead on respects democratic institutions (55% vs 39%, D+16), cares about people like you (50% vs 36%, D+14), prioritizes people like me (47% vs 35%, D+12), honest and ethical (48% vs 36%, D+12), and looks out for the middle class (48% vs 37%, D+11).

But Republicans win on qualities associated with strength, though by smaller margins. 55% say the GOP has a clear message, against 47% for Democrats (R+8). 48% say Republicans have strong leadership, against just 37% for Democrats (R+11) — the largest single GOP advantage on the list. Republicans also edge Democrats on “gets things done” (46% vs 39%, R+7) and “willing to fight for what it believes in” (63% vs 59%, R+4).

On the two explicitly negative items we asked about, Republicans come out worse on both. 56% say the Republican Party is “too extreme in its positions,” against 46% who say the same of Democrats. And 53% call the GOP “out of touch with ordinary Americans,” against 51% for Democrats — effectively a tie.

As I wrote back in February, the Democratic brand is not predominantly woke, but weak. Respondents to our survey associated the Democrats with traits like honesty and caring about the working class, but they are seen as weak and not particularly effective. The Republican brand, by contrast, is a strong brand that a majority of the country finds extreme.

Zooming in now on political independents — who are the actual persuasion targets in elections — the same pattern holds. If anything, on most items it’s even more pronounced.

Independents give Democrats the edge on tolerance, honesty, caring about ordinary people, respecting institutions, and looking out for the middle class. Republicans hold their advantage on strong leadership, clear messaging, and getting things done — though the margins shrink compared to all adults. And independents view the Republicans as “too extreme,” just like the general public does.

That helps explain why Democrats’ -3 favorability is running 13 points ahead of the GOP’s -16. Independents don’t love Democrats; they just trust them more and feel better represented by them. And that’s despite the large gap in the leadership questions.

One more look at the “strong leadership” gap before we move on. We asked respondents to name, in their own words, a political figure whose values they share. Among Republican respondents, Donald Trump and JD Vance dominate — no one else is anywhere close.

Democrats, however, are much more split over which political figures embody their values. Among Democratic mentions, no single name gets anywhere near the share Trump commands. Left-leaning candidates such as Bernie Sanders, Alexandria Ocasio-Cortez, and Zohran Mamdani lead the pack, with more traditional voices including Barack Obama, Pete Buttigieg, Kamala Harris, and Gavin Newsom coming in close behind.

This compounds a problem for the party. It’s no wonder that so many voters think it has weak leadership when it doesn’t even have a clear leader.

What Democrats’ own voters want: for them to fight harder

So if Democrats have a “weak” problem, what can they do about it? The most revealing question in our whole poll is probably the following open-ended survey item. We asked every respondent to describe, in their own words, a recent thing the Democratic Party did that upset them. We hand-categorized the substantive answers into the buckets below.

Among Democratic respondents, the single most common complaint about their own party is that it is too weak, too cautious, or not fighting hard enough — 30% of substantive answers fall into that bucket, and another 17% are still mad about the party caving in last fall’s shutdown fight without winning an extension of ACA premium subsidies (a few also cite the failure to extract concessions on ICE). That runs ahead of every substantive policy concern, including on immigration, the economy, Israel/Gaza — you name it.

And we don’t have to guess who Democrats want their party to fight. Of the 103 Democratic respondents whose answers fell into the “fight harder” bucket, more than half named Trump or his administration directly — “not standing up to Trump,” “letting Trump run amok,” “strongly worded letters are not action,” etc. A handful said they wanted the president impeached. One person said the party was “bending the knee” on the One Big Beautiful Bill. While a small single-digit percentage of Democrats said they wanted their party to moderate on cultural issues, the dominant ask is visible, sustained opposition to the president and his party.

“What upsets me the most right now about the Democrats right now is that they don’t have a clear message. I used to consider myself a strong democrat and I really don’t anymore as a result.” — Democrat

“Weak leadership. Lack of focus.” — Independent

“The party leadership consistently takes positions that are opposed to the vast majority of Democratic party voters … they try to skirt by by being ‘not as bad as Trump’.” — Independent

“They have failed to present a unified front to counter Republican insanity.” — Democrat

Independents land somewhere different. Their complaints cluster around general dissatisfaction and messaging — the sense that they don’t know what Democrats stand for or what the party is actually trying to do. Two specific themes show up that don’t appear in the Democratic column: The first is a lingering Biden grievance: independents in our open-end repeatedly cite the party “lying about Biden’s health,” keeping his campaign going “far too long,” and “anointing” Kamala Harris as the nominee without a competitive process. The second is a leadership vacuum complaint — variations on “lack of leadership” and “I don’t know what they stand for.”

Some independents do echo the Democratic “fight harder” critique, but it’s much smaller in volume (11 mentions vs 103 among Democrats). These voters chastised Democrats for not impeaching Trump, “not pushing hard enough back and folding under pressure.” Independents are not asking Democrats to move left or right. They’re asking the party to look like a coherent operation with someone in charge.

Another thing worth highlighting is that the Republican immigration column is largely just retrospective grievance. Roughly half of Republican respondents who mentioned something about immigration cited the Biden-era immigration surge — “letting millions of illegals into the country,” “open the borders with biden,” “they failed to secure the border” — rather than any recent Democratic action.

Most of the rest of Republican responses on immigration fall into one of two themes: that Democrats prioritize “illegals over Americans” (the modal phrasing on the right), and that Democrats are now actively obstructing ICE — sanctuary policies, blocking deportations, “stop funding ICE.” A smaller cluster folds the SAVE Act / voter ID into the bucket, framing Democratic opposition to it as pro-noncitizen-voting. But mostly, Republicans were just upset about “open borders” under Biden.

On the shutdown bucket, note that Democrats and Republicans are mad at Democrats for opposite reasons. Democrats are upset that the party “caved” on recent shutdowns, especially the one in fall 2025 over health care funding. Democratic voters wanted their party to keep fighting for health care, but they reopened the government without winning an extension of ACA subsidies (or, more recently, on any immediate concessions on ICE). Republicans, on the other hand, are are upset that the party caused the shutdown in the first place by refusing to fund DHS and holding up TSA paychecks.

It is notable that independents don’t really care about the shutdowns.

Democrats are not unpopular. They’re unsatisfying.

What all of this suggests is that Democrats do not have the problem many political narratives say they do. The party’s core weakness is not that voters see it as elitist or too extreme; it is that too many voters, including their own, see Democratic politicians as unmoored, passive, and ineffective. Republicans, by contrast, still project the kind of strength and clarity that voters often reward — though their extremism is a huge drag on votes.

More Americans see the GOP as extreme, out of touch, and worthy of intense dislike. That is why Democrats can be underwater on their favorability and still in a stronger electoral position overall.

The strategic implications here are straightforward. Democrats do not need to reinvent themselves ideologically nearly as much as they need to convince voters they can act with purpose and deliver on their promises. Their own supporters are not begging for moderation so much as urgency; independents, too, have fewer specific ideological qualms with the party as they do personal germane criticism. They are not demanding a lurch left or right so much as evidence of leadership, coherence, and fight.

In a political environment where neither party is broadly beloved, voters must know you stand for something — and for standing up for it, too. The Democrats have made a lot of progress on these numbers over the last year. But a perception of weakness is still its biggest one.

Subscribe now

The April Strength In Numbers/Verasight poll surveyed 1,514 U.S. adults online between April 10 and April 14, 2026. The margin of error for the full sample is ±2.6 percentage points. Full methodology and crosstabs are available at gelliottmorris.com/poll.

If you made it this far: Strength In Numbers is subscriber-supported. If this work is useful to you, upgrade to a paid subscription to keep independent polling alive — and to get the full crosstabs and Tuesday Deep Dives.

Subscribe now

On April 7, President Trump posted on Truth Social that Iran’s “whole civilization will die tonight,” capping a week of increasingly unhinged posts about the war in Iran (in another, the president told Iran’s leaders to “Open the Fuckin’ Strait, you crazy bastards, or you’ll be living in Hell. … Praise be to Allah!”). The posts have drawn sharp criticism from political and media figures across the political spectrum, including prominent right-wing voices who backed Trump in 2024. Tucker Carlson called the threats against Iran’s civilian infrastructure a war crime and now says he regrets helping elect Trump, while Alex Jones, Megyn Kelly, Marjorie Taylor Greene, Theo Von, and Tim Dillon have also spoken out.

In Congress, Rep. John Larson has introduced 13 articles of impeachment against Trump, with more than 85 House members publicly backing either impeachment or invoking the 25th Amendment. All of which raises the question: how much of the general public wants Trump impeached? If even his most right-wing supporters are breaking away, support among the broader public is presumably pretty high.​​​​​​​​​​​​​​​​

A new Strength In Numbers/Verasight poll conducted April 10-14, 2026 finds 55% of U.S. adults say the House should vote to impeach Trump. 37% oppose, and 8% are unsure. A surprising percentage of both Republicans and Trump’s own 2024 voters say they would support impeachment if a vote were held today.

That net +18 verdict puts Trump in the neighborhood of the numbers Richard Nixon saw at the peak of the Watergate scandal in August 1974 — more on that comparison below. The toplines and crosstabs for this poll can be found on the Strength In Numbers website.l

Subscribe now

Who wants Trump impeached?

Our new poll shows that 55% of U.S. adults support the House voting to impeach Trump, while 37% oppose and 8% are unsure.

As for the president’s overall approval rating, there is a strong intensity gap in responses to our poll. Overall, 45% of all adults say they strongly support impeachment, while only 30% say they strongly oppose it. That is a 15-point intensity gap in favor of impeachment — the people who want Trump out are both more numerous and more committed than the people who want him to stay.

Support for impeachment extends well beyond the Democratic base. The chart below shows support and opposition to impeaching the president for major demographic groups in our new survey:

My big takeaway from this chart is that all but three groups support impeaching Trump: Republicans, Trump’s 2024 voters, and seniors (who oppose impeachment by 4 points, 47-51). On the other side, independents (including leaners) split 50-28 in favor, and non-voters, who may have slightly voted for Trump over Harris in 2024, back impeachment 53-25.

But also of note is that 21% of Trump’s own 2024 voters now say he should be impeached. That’s roughly one out of every five of the people who put him back in office.

Share

Support for impeachment is historically high, by modern standards

The 55% figure is unusual by modern impeachment-polling standards.

After January 6, 2021, ABC News/Washington Post found 56% wanted Trump impeached and removed from office. Other polls showed similar numbers: The Pew Research Center had it at 54%, and Gallup at 52%.

For comparison’s sake, during the Ukraine impeachment in fall 2019, Fox News had impeachment and removal at 51% and Gallup at 52%. Bill Clinton’s peak removal number in January 1999 (which failed) was just 33%.

And support for impeaching Trump today is only a few percentage points lower than it was for Richard Nixon in 1974: And at the height of Watergate, days before Nixon resigned, Gallup found 58% wanted him removed. Trump is in “Nixon resignation” territory with these impeachment numbers (and his approval rating overall).

But note our poll is not completely apples to apples: we asked about the House voting to impeach, a lower bar than the “impeach and remove” language most national pollsters have used historically. But even accounting for that, the April 2026 number sits at or near the high-water mark of modern impeachment polling, and well above the Ukraine and Clinton readings.

Would a Democratic House impeach Trump?

None of this means Trump will actually be impeached in the current Congress. A Republican House will not impeach a Republican president. The White House has already called the Larson resolution “pathetic”, and Republican leadership has shown no interest in bringing it up for a vote.

Impeachment requires 218 votes. There will not be enough votes to impeach the president in the 119th Congress — but there might be in the 120th.

Democrats need a net gain of just a handful of seats to flip control of the House of Representatives, which would give them enough for the simple majority required to impeach Donald Trump. And the political environment is about as favorable a backdrop for a wave election as either party has seen since 2018 or 2010.

If Democrats win the majority in November, they will walk into their first session in January 2027 with a public mandate for impeachment already in hand. Of course, the Senate would still need to vote to convict the president for the impeachment to have any effect in the real world (this poll does not speak to support for removal, especially in key states), and that is an uphill battle.

But for the first time since Trump returned to office, the polls are indicating Americans support impeaching their president — for a third time. That is itself a serious indictment against his presidency.

Subscribe now

The April Strength In Numbers/Verasight poll surveyed 1,514 U.S. adults online between April 10 and April 14, 2026. The margin of error for the full sample is ±2.6 percentage points. Full methodology and crosstabs are available at gelliottmorris.com/poll

Related Articles

Subscribe now

President Donald Trump has now been in office for about 15 months, and his numbers keep getting worse. Our latest Strength In Numbers/Verasight poll, conducted April 10–14, finds just 35% of U.S. adults approving of his job performance, with 61% disapproving — a net approval rating of -26. That’s a new low in our poll, down from -23 last month and a steep fall from (an already poor) -16 when we first began tracking this question in May 2025.

This deterioration has been driven largely by his handling of the economy and prices. Trump’s net approval on prices and inflation has fallen to -46 — the worst rating on any single issue in the history of our poll, and a stunning 6-point drop from March’s already record-low -40. Nearly three-quarters of Americans (72%) now disapprove of the way Trump is handling prices.

Here are the poll’s headline findings:

Headline poll findings

• Job approval: 35% of U.S. adults approve of Trump’s job performance; 61% disapprove (net -26, down from -23 in March)

• Generic ballot: Democrats lead Republicans 50% to 43% among registered voters, a 7-point margin

• Direction of the country: 55% say things are going poorly and major changes are needed, a new high. Just 8% say things are going well

• Prices: Trump’s net approval on prices/inflation has fallen to -46, the worst rating on any issue we have ever recorded

• Border security: The one bright-ish spot — border security bounced back to net +1, making it once again the only issue where Trump is not underwater

• Iran: 64% of Americans say the war in Iran has not been worth the cost, up from 58% in March. 48% say the U.S. should never have gone to war in the first place

More details from this month’s survey follow.

This is the first of several articles releasing data from the April Strength In Numbers/Verasight poll. Subscribe to Strength In Numbers to get them in your inbox.

Subscribe now

Democrats have led the generic ballot in every poll we’ve conducted

On the generic congressional ballot, Democrats lead Republicans 50% to 43% among registered voters, a 7-point margin. Democrats have now led in every single Strength In Numbers/Verasight poll since we began fielding in May 2025, with margins ranging between +4 and +10 points.

The April margin is slightly wider than March’s D+6 among registered voters, though within the range of normal month-to-month variation. The consistency of the Democratic lead is more notable than any individual month’s margin: across 11 monthly polls (we skipped December 2025), Democrats have never trailed, with margins ranging between +5 and +10 points.

Trump’s approval on prices hits a record -46 — and he’s underwater on everything else, too

Trump’s issue-by-issue approval ratings tell a grim story for the White House. His net approval on prices and inflation has fallen to -46, worsening every single month of 2026: -31 in January, -35 in February, -40 in March, and now -46 in April. Just 26% of Americans approve of the way he is handling prices, while 72% disapprove. More than half of all respondents (56%) strongly disapprove.

And here’s a zoom in on the change from March to April. Trump has lost ground on most issues, and is underwater with the average American on almost every single one:

The one exception is border security, which bounced back to net +1 after dipping to -3 last month. Even this “best” issue is essentially a coin flip — 49% approve and 48% disapprove. But it is, once again, the only issue where Trump is not in negative territory.

Deportations, immigration, and crime — the issues the administration presumably wants to be judged on — are all in the range of -10 to -12. Better than the kitchen-table issues, but nowhere near positive.

The country’s top problems are the issues where Trump is weakest

When we ask Americans to name the single most important problem facing the country, 34% say prices and inflation — the same issue where Trump’s approval is at its worst. Another 14% say jobs and the economy, 12% cite elections and democracy, and 8% name health care — rounding out the top four.

When we expand to let respondents name their top three problems, 57% include prices/inflation, 40% name jobs and the economy, and 34% cite health care. These are precisely the issues where voters trust Democrats more.

On the question of which party voters trust more to handle each issue, Democrats lead on 8 out of 12 issues, including all of the top voter priorities. Democrats hold their biggest leads on health care (D+22), government funding and social programs (D+22), education (D+18), and elections and democracy (D+13). They also lead on prices/inflation (D+11) and jobs and the economy (D+10) — the two issues voters rate as the country’s biggest problems.

Republicans lead on border security (R+13), crime and public safety (R+3), immigration (R+3), and deportations (R+2).

When we ask voters which party would do a better job handling their own top issue, Democrats lead 48% to 36%, with 17% unsure. That’s a 12-point advantage on the issues people care most about.

A quick note before more: this poll exists because readers pay for it. Strength In Numbers has no billionaire backer and no ad sales team — just subscribers who believe the public deserves honest, transparent data about what Americans actually think. If that’s you, please consider upgrading to a paid subscription. You’ll get the full crosstabs, the Tuesday Deep Dives, and the satisfaction of knowing you helped keep independent polling alive.

55% say things in the country are going poorly — a new high in 2026

The share of Americans who say “things are going poorly and major, disruptive changes are needed” reached 55% in April, a new high in our polling. Just 8% say things are going well. Another 34% say things could be going better.

Together with Trump’s approval numbers, the picture is one of slow but steady erosion in public confidence.

Most Americans say the Iran war is not worth the cost

Our poll also asked about the war in Iran, and the results are bleak for the administration. A commanding 64% of Americans say the war has not been worth the cost to the United States, including 51% who say it was “definitely not worth the cost.” Just 29% say it has been worth it. Opposition has grown since March, when we asked a similarly framed question and found 58% called the war a bad use of taxpayer dollars.

With the U.S. and Iran having recently agreed to a temporary ceasefire, we asked respondents which view came closest to their own. The plurality — 48% — say the U.S. should never have gone to war in the first place. Another 13% say the U.S. did not achieve its goals but ending the war is still the right move. Just 16% say the U.S. achieved its goals and the ceasefire is the right move, and 12% say the U.S. should have continued fighting until it achieved its goals.

Adding those two categories together, 61% either opposed the war entirely or believe it failed to achieve its objectives, while only 16% view it as a success.

We also found that 40% of Americans say they do not understand the main reasons the U.S. went to war in Iran in the first place, with 18% responding “not too well” and 22% saying “not well at all.” Only 28% say they understand the reasons “very well.”

The trend is not the GOP’s friend

Trump’s numbers have now declined in every wave of our poll this year. And while his support bounced occasionally in 2025, it has never increased two months in a row. Overall approval has dropped from 40% in January to 35% in April. His net rating has fallen from -18 to -26.

His drag on the Republican Party puts it in a dangerous place for November’s midterm elections. Democrats have led the generic ballot in every poll we’ve conducted since last May. And no wonder; Trump is underwater on every issue except border security. His approval on prices — the issue voters care about most — has fallen from -31 to a record -46. A 55% majority of Americans — a new high — say the country needs major, disruptive change.

None of this guarantees a Democratic wave in November. Trump’s approval rating could always recover, the war in Iran could end cleanly, and gas prices can fall. But the trend is not his friend, and recovery looks increasingly improbable. The fundamentals a president typically needs to protect his party’s congressional majorities — a growing economy, a public that feels secure, a sense that the country is on the right track — are all moving in the wrong direction at once. If the April numbers are anywhere close to where things stand next fall, Republicans should be preparing for significant losses, at all electoral levels.

Over the next couple of days, Strength In Numbers will release more data from our monthly survey with Verasight. Subscribe today to get those fresh polling dispatches delivered straight to your inbox.

Subscribe now

Methods statement: Verasight collected the data for this survey from April 10–14, 2026. The sample consists of 1,514 United States residents ages 18 and above. The data are weighted to match population benchmarks of age, race/ethnicity, sex, income, education, region, metropolitan status, partisanship, and past vote. The margin of sampling error is +/- 2.6%. Topline document prepared by G. Elliott Morris for Strength In Numbers. While Strength In Numbers had input on question wording, all other methodological decisions were made and carried out by Verasight to ensure independence of the data-generating process behind these results.

You can download a full topline file, key crosstabs where they are mentioned in this article, and methodology statement at the Strength In Numbers/Verasight polling portal. The paywalled section containing crosstabs and more interactive graphics has been updated with these numbers.

If you have any questions about this poll or the release, please email polling[AT]gelliottmorris.com.

Have a suggestion for next month’s poll? Leave it in the comments below.

If you’re a frequent reader of Strength In Numbers, I’m confident you will get a lot of value out of a paid subscription to this Substack. You’ll get access to all of my analysis, including weekly Tuesday Deep Dives, and you’ll be supporting independent data journalism — like this original polling.

Become a paying member today

And if you believe independent data on the needs and beliefs of the mass public is a vital component of a modern democracy, check out my book, and the precursor to this Substack, Strength in Numbers.

Related Articles

Leading off: A new YouGov poll shows MAGA and non-MAGA Republicans splitting on the Iran war, the economy, And especially inflation. That matters because non-MAGA Republicans are exactly the voters Trump needed to assemble his 2024 coalition, and he’s losing them on the issue they care about most.

On deck this week: It’s poll week here at Strength In Numbers! Toplines and crosstabs come out Tuesday, with more data on Wednesday, and a deeper dive into the Democratic Party’s favorability/perceptions issues on Thursday.

The latter will be part of this week’s recording of the Strength In Numbers podcast with me and David Nir — live and in-person in Washington, DC! We’d love it if you’d join us!

Our session will take place on Thursday, April 23, at 11:15 AM at the America Votes Summit, which is being held at the Walter E. Washington Convention Center (the address and directions can be found here). We will be breaking down the current state of the midterms using data, and also have an exclusive new poll explaining current sentiment toward the Democratic party. We’ll be taking your questions in an extended Q&A session!

The first 75 listeners who RSVP can attend free of charge. Just click below:

RSVP for the Strength In Numbers podcast

We really hope to see you there, and we’d love to chat with you after the show, too!

On with the data.

1. Non-MAGA Republicans are breaking on Iran

A new YouGov/The Economist poll out this week shows MAGA and non-MAGA Republicans splitting sharply on the Iran war. Remember their polling shows about half the party calls itself “MAGA.”

This reminds me of a chart that David at YouGov rendered for me (when I somewhat manically emailed him about the MAGA/non-MAGA divide) last month:

Note that there’s usually a pretty consistent divide between MAGA and non-MAGA Republicans, and the two groups tend to move toward Trump and away from him in similar numbers at the same time.

That’s not what’s happening on Iran. Non-MAGA Republicans are moving against Trump in big numbers, and MAGA voters are standing pat.

But Trump is particularly weak with non-MAGA Republicans on inflation. The gap between MAGA and non-MAGA approval is wider there than on almost anything else, at about 40 points on average over the last 3 months. And non-MAGA Republicans are by definition exactly the voters inside the GOP who matter most electorally. On the issues these voters say matters most to them, Trump is pursuing exactly the strategy that could help him.

This ties into an MS NOW interview I did on Saturday. Trump’s 2024 coalition was built on four pieces. The core is the roughly 30–35% of Americans who are MAGA on any given policy. But the base alone doesn’t win elections. Trump won by adding three other groups: non-MAGA Republicans who are negatively polarized against Democrats and would never vote for them; swing voters who soured on Kamala Harris for ideological or personal reasons; and voters who were simply fed up with the economy and wanted the other party in charge.

Looking at the polling, Trump has lost a little ground with his base and with the reluctant Republicans. But he’s losing real ground with the Harris-skeptics and the economy voters — and he’s losing it on the issue those groups say matters most.

This is why MAGA is not an enduring, winning political movement.

Share

2. What Strength In Numbers published last week

Tuesday’s Deep Dive made the case that consumer sentiment is low because “excess prices” — how far prices have overshot their pre-COVID trend — dominates how people think about their personal finances today, and that standard economic models miss this factor:

Thursday’s podcast previewed early findings from our April Strength In Numbers/Verasight poll and walked through the excess-prices framework and why neither party has a quick political fix to this problem:

Friday’s Chart of the Week followed up with eve mode modeling of consumer sentiment, this time around a new inflation-delayed excess prices measure I backtest back to the 1960s built around excess prices:

Also: On Friday, I wrote over at 50+1 with my colleague Mary Radcliffe about how “soft” measures of Trump’s support are eroding among Republicans. I don’t usually cross-post the 50+1 stuff here to SIN, but this article is on the same subject as this week’s roundup and think you should check it out!

FiftyPlusOne

“Soft” measures of Trump’s popularity are eroding among Republicans

Welcome to The Trendline, FiftyPlusOne’s weekly polling roundup. Subscribe to get it delivered straight to your inbox and get bundled access to our polling website…

Read more

5 days ago · 19 likes · FiftyPlusOne, Mary Radcliffe, and G. Elliott Morris

If you’re a frequent reader of Strength In Numbers, I’m confident you will get a lot of value out of a paid subscription. You’ll get access to all of my Tuesday Deep Dives, monthly polling data, and more.

Become a paying member today

3. Even more numbers!

A handful of things I read this week.

• The Economist Intelligence Unit’s latest democracy index suggests the global democratic recession may finally be easing — the scores of nearly three-quarters of countries held steady or improved last year, and the global index posted one of its biggest annual increases since 2012: Global democracy is in better shape than you think

• Arin Dube shows that four Republican-leaning states (Arizona, Florida, Missouri, Nebraska) raised their minimum wages to $13–$15 via ballot initiatives — and restaurant and retail employment didn’t take a hit: These Republican-leaning States Went Big on Minimum Wages

• 75% of U.S. adults say they’ve read all or part of a book in the past year, per Pew — and print still beats digital and audio: Do Americans read print books, e-books or audiobooks more?

• Charles Franklin’s direct comparison of Trump first-term vs second-term consumer sentiment: first 23 months of term one averaged 97.5 on the Michigan index; term two is averaging 55.5: Second Term Worse than the First

Share

One more thing. I’m looking for people to help me stress-test something I call the “2026 Senate Simulator” that I’ve been quietly building for a few weeks. It’s an interactive tool for exploring the Senate landscape -- part education, part fun, and not a full election forecast. You can tweak the underlying assumptions of the model (national environment, value of incumbency, state-level partisanship, etc.) and watch the Senate map shift in real time.

You can try the 2026 Senate Simulator here. Let me know what you like, what’s confusing, what’s wrong, etc. And let me know especially if you think something is missing that would otherwise make this better!

And that’s it for this week! Thanks for reading. Strength In Numbers will be back in your inbox on Tuesday.

Got more for next week? Email your links or add to the comments below!

Leave a comment

Related Articles

A gallon of milk costs about a dollar more today than it did in 2019. Eggs are running about 45% higher than they were five years ago, rent is up about 30%, and used cars are about 25% more expensive. Across the basket of essentials Americans actually buy every week — groceries, housing, transportation, the un-skippable stuff — prices sit roughly 10 to 15% above where the pre-pandemic trend said they should be, and they have basically refused to come back down.

That gap, between the price level you actually pay at the store and the level you’d be paying if 2020 had never happened, is what I’ve been calling “excess prices.” And statistically and in theory, it does a better job of predicting how Americans feel about the economy than any of the standard macro indicators we usually reach for.

That was the argument I made on Tuesday — that the reason consumer sentiment is in the basement, despite low unemployment and rising real wages and a near-record stock market, isn’t news coverage or social media vibes or voter irrationality. It’s the level of prices, not the rate of change. Inflation falling from around 9% to around 3% doesn’t undo the fact that the supermarket bill is now, and probably permanently, meaningfully higher than it was.

I was overall pleased with the reaction to Tuesday’s Deep Dive, and several economists sent me thoughtful notes about the work (with some cheeky criticism of the spaghetti line charts). There was also some criticism of the model, which offers up an opportunity to do some more explanatory work with this model. Here are the two fair objections I got, that I’ll address in this piece:

1. Overfitting. Maybe the model only “works” because it was built and tested in a low-inflation, post-1990 environment. The fix for this problem is to explore whether the same setup also predicts sentiment in another high-inflation regime — for example, the 1970s.

2. The polling variables might be downstream of the same biases pulling down consumer sentiment. My model uses a question from the University of Michigan’s Survey of Consumers that asks people who say the economy is bad why they say that. Specifically, it uses Table 7 of the survey, which reports the percent who say their finances are poor because “prices are higher.” The criticism of this variable is that maybe respondents are just parroting back the negative inflation coverage they’ve absorbed from cable news and social media. The fix here is to strip out the part of price perceptions that news sentiment can explain, and test whether what’s left still predicts sentiment.

The short answer is: neither one of these critiques sticks. The model generalizes to the 1970s about as well as it fits the 2020s, and the price-perception signal stays strong even after the news-explained portion is stripped out. The most plausible mechanism for why consumer sentiment lags mainstream economic models today is that they don’t have data on excess prices, which voters care about in addition to the standard economic variables.

The natural place to start is with the overfitting objection, because it is the more fundamental one. If this is really just a post-COVID story dressed up as a general theory, then the model should fall apart the minute you ask it to explain a very different economy. I chose to back-test on 1972-1983, an era of persistently high inflation, weak real wage returns, and collapsing consumer confidence. If the excess-prices model still works there, it is probably capturing something real.

Subscribe now

Test 1: Does the model work in a different inflation regime?

One quick note on the variable, since I’ve changed it slightly from Tuesday’s version. My original “excess prices” measure was a look at inflation per the Federal Reserve’s preferred inflation measure, “Core PCE.” Unfortunately, the Fed (for its own particularly Fed-y reasons) excludes the price of food and energy from PCE — but people spend a lot of money on food and energy! Since we are trying to measure inflation in things the public cares about, we need to include those items, too.

So in this updated model, I use a composite of the things people actually buy a lot of — 25% food, 50% shelter, and 25% vehicles — and compare current prices to expected prices. The way this works is that for any month, m, I take that composite, fit a linear model on all the price data from 15 years to 5 years before m, and predict what prices for these items “should have been” on month m. Then, I take what they actually cost, and calculate the percentage residual between (1) what things cost and (2) what they should. Per Jared Bernstein’s terminology, I call this “excess prices.” I repeat this procedure for every month in the time series from January 1965 to March 2026.

Then, I divide the excess prices for a given year by the trailing average rate of headline inflation over the last ten years. This way, the excess prices variable measures roughly the same thing whether you’re in the late 1970s, when Americans had just had a decade of stubbornly high inflation, or the 2020s, after 30 years of inflation below 3%. The theory behind this adjustment is that a 15% excess in 1978, when inflation had been running 6 to 8% for years, doesn’t shock people the same way that a 10% excess in 2024 does, when inflation had been near 2% for a decade. A 15% excess on a 6% baseline is just 2.5x the prevailing inflation rate, whereas a 10% excess on a 2% baseline is 5x the prevailing rate.

This isn’t just a hunch, by the way. Recent political-science work by University of Texas at Austin political scientists Derek Epp and Christopher Wlezien, “Benchmarking in the Updating of Public Perceptions,” shows that people don’t update their beliefs about the economy based on new economic information evenly in different inflation regimes. They find a similar thing I do: that when the economy turns sour after a period of average (or good!) economic growth, people react more negatively than they do if you get a bout of inflation right after a big recession.

Table 2 from the Epp and Wlezien paper (reproduced below with permissions) shows this cleanly: the same change in objective economic indicators has wildly different effects depending on what people had recently come to expect for economic growth. A two-standard-deviation negative shock to the leading economic indicators moves sentiment by 13 points when it arrives after a period of good times, and just 8 points when it arrives after bad times. That’s a projected 50% larger impact.

If you apply this theory to inflation, this means that after years of low, stable inflation, the post-COVID jump in everyday prices registered as an extremely abnormal shift in the price level. A comparable shock in the late 1970s or around 1980 would still have hurt, but it would have landed on a public already conditioned to expect higher inflation, so they would have been less bummed. The deflator I use in the variable is, essentially, the empirical version of Epp and Wlezien’s theory: the same amount of excess price shock gets scaled up when the recent inflation environment has been quiet, and scaled down when it’s been noisy.

So you can better visualize what’s going on here, see the graph below. The panel in the bottom right shows excess inflation in prices of everyday household goods, adjusted for period-level inflation.

Tuesday’s piece showed that excess prices predict post-2020 sentiment well. But, fair point from the critics, that was a relatively clean test — both the training data and the test data come from a similar low-inflation regime, with similar information environments. The harder test is whether the same model also works when you ask it to predict a period so different from the 2020s that it might as well be a different economy entirely.

From roughly 1972 through 1983, inflation was running double-digit, unemployment spiked well above natural rates, sentiment collapsed, and real stock returns were negative. Pretty much everything was bad at once. The conventional macro story works fine here — high unemployment plus high inflation equals low sentiment, no further explanation required. Textbook stuff.

So if my excess-prices story were really just a post-COVID artifact, you’d expect a model trained on data outside the 1970s to badly miss 1970s sentiment, because the price-level shock isn’t doing the explanatory work — high unemployment and high inflation are. In this case, if you add another negative indicator on top of these ones — and especially one that is huge in nominal terms — you would get a predicted economic sentiment that was way too low.

To find out, I trained four nested models on data outside each test window and compared their out-of-sample fit. These four models are:

1. A baseline “conventional wisdom” economics model: Consumer sentiment is a function of unemployment (above the natural rate estimated by the CBO), inflation (in the PCE), and YoY stock returns.

2. The baseline model plus my inflation-adjusted measure of excess prices.

3. Model 2 plus a measure of “economic news sentiment” — how negatively or positively newspaper coverage of the economy is in any given month — taken from the San Francisco Fed.

4. Model 2, plus the “prices are higher” polling variable from the University of Michigan question I introduced above — the share of respondents who say their personal finances have gotten worse over the last year, specifically because prices are higher.

The line chart below shows the predictions each model would have made on economic sentiment after 2016, trained only on data up to 2016. The yellow line — the baseline model — does worst, while the other models — that add excess prices and attitudinal measures — do better.

Now, here is the same thing, but incorporating holdout predictions for the inflationary era in the 1970s and early 80s:

Note how the model does decently well in the 70s-80s, though sentiment is higher than we’d expect across the board. The monthly errors are mostly not larger than you’d expect based on the root-mean-squared errors of each model:

Note: the table above reports prediction error after 2021 for the forecast, but remember that the model stops training at 2016, so it’s not that it just “got lucky” because we stopped training right before COVID!

Two things stand out in that table. First, the baseline model — just unemployment, inflation, and the S&P 500 — fits the 1970s reasonably well, with an RMSE of 9.7 over 132 months, but misses badly in the post-2020 period, where its RMSE is 21.4. And there’s a 19-point average gap between what conventional macro predicted and what people actually told the Michigan survey. Second, Model 2, which is the same baseline plus the regime-adjusted excess-prices variable, performs almost identically across both eras: 8.7 in the 2020s and 9.6 in the 1970s. That’s the same model, with the same coefficients, trained on data from neither test period, fitting both regimes within a few index points of each other.

That, I think, is decent evidence that the relationship generalizes across inflation regimes — not just an artifact of post-COVID overfitting.

The regime adjustment is doing a lot of the work here, and you can see why if you imagine the alternative. Without dividing price excess by the trailing inflation mean, the model would see enormous price excess in the 1970s — prices were extraordinarily high in absolute terms — and would predict that sentiment should have cratered throughout the entire decade. The regime-adjusted version correctly recognizes that 1970s consumers were already used to high price growth, so a given-size excess simply wasn’t as shocking as it would be today.

This is basically the central finding of Epp and Wlezien’s paper, applied to the relationship between inflation and consumer sentiment specifically. The Epp and Wlezien paper suggests why a regime-adjusted measure of excess inflation ought to matter psychologically, and the historical backcast shows that the measure works empirically. The model behaves the way the behavioral theory (with completely separate data!) says it should. Bully!

Also worth pointing out is that model 4, which adds the Michigan attitudinal variable about price anxiety, tracks the full time series incredibly closely. This suggests that once you know what the unemployment rate and SPY returns are, then the only other question you need to know is how stressed out people are about prices, and you can predict the index of consumer sentiment. This suggests that consumer sentiment is largely a function of price anxiety, which is not the way economists typically treat the index!

Subscribe now

Test 2: Are perceptions actually downstream of news and social media?

The polling variable suffers from the big weakness that we don’t know whether people say prices are higher because prices are higher, or because they’re being “influenced” to believe that prices are higher. The consumer sentiment truthers responded to my first model, arguing that survey respondents who tell pollsters that “higher prices” are hurting them are possibly just repeating what they heard on cable news and social media.

Well, one way to test this is to look at the Model 3 (red line) results above. Does a model incorporating a measure of media sentiment do any better in predicting consumer sentiment than the baseline conventional model? No, in fact, it does about 1.6x as badly as our excess price and poll-response model in terms of expected error, and 3-4x as badly in terms of average bias. Ouch!

Here’s another test: If price anxiety were just recycled media tone, then the part of it that isn’t explained by news coverage should stop predicting sentiment. So I residualized price perceptions on news sentiment — extracted the part of price perceptions that news tone explains, threw it away, and reran the model with only the residual. Summary of the results below (the important column is the “model fit” column).

The raw correlation between news sentiment and price perceptions is r = -0.37, which is statistically significant but substantively pretty weak. News sentiment explains about 14% of the variation in price perceptions; the other 86% is doing something else. So when I throw away that 14% and use only the residual, the price-anxiety variable still predicts sentiment with a t-statistic of -18.6. That’s still a powerful predictor, and it still produces a higher R² (0.71) than news sentiment all by itself (0.58).

You can see this visually, too. The news-sentiment series (top of the chart below) bounces around but has more or less returned to neutral over the past couple of years. The price-perception series (middle) shows a sustained 40%+ plateau from 2022 through today. The two series, by any reasonable reading of the chart, are not moving together.

So most of the signal in “prices are higher” responses is independent of media tone. When people tell pollsters that higher prices are hurting them, the overwhelming majority of what they’re reporting is not coming from the news. The inference should be that this is coming from the grocery store, the rental market, and the insurance bill.

Subscribe now

2b: What about social media?

One problem with the chart above, however, is that it doesn’t test the role of social media on sentiment at all. The version above only includes news tone, as measured by the SF Fed Daily News Sentiment Index. But a careful reader would point out that Americans in 2026 experience prices through a much looser information environment than just the news. Many people get their impressions of the national economic picture from social media feeds and group chats, too. The SF Fed index captures one slice of the information environment, and a better analysis would try to capture all of it.

That objection is real, and I want to be upfront that the residualization test from earlier doesn’t fully dispose of it. News alone isn’t sufficient to explain the sentiment gap, but news isn’t the whole information ecosystem either.

The broader evidence, though, still points to something more layered than a purely mediated story.

First, the structural excess-prices model explains most of the 2022-2024 sentiment collapse on its own, without any information-environment variables in the mix at all. The Model 2 residuals shrink by 60% versus the baseline, and have little bias when compared to the trend (just 5 index points in 2025). So there’s evidently not a large independent role for the information environment to play, anyway.

Second, even when you do let media tone into the model, the part of price perceptions that isn’t explained by news coverage is still significant, and it still outperforms news sentiment all by itself in explaining consumer sentiment. That’s the residualization test: If the only thing happening in the survey responses were echo of the media tone, that residual should be statistical noise.

But third, and this is the part I think is most directly contradictory to the strongest version of the media-ecosystem theory: if exposure to political and media communication were the primary driver of price pessimism, we would expect the most-engaged Americans to be the most negative on prices. They are not:

The most-engaged Americans are less negative on prices, not more

If political coverage and social media really were driving sentiment downward, then the people most exposed to that content should be the ones most negative on prices. We can check that directly with our own polling data.

In October 2025, I asked 1,567 respondents to our Strength In Numbers/Verasight poll how closely they follow political news, on a four-point scale from “never” to “all the time,” as well as two other questions we can consider proxies for news sentiment (1) “In your opinion, what is the most important problem facing the nation today?” and (2) “Do you approve or disapprove of how Donald Trump is handling prices/inflation?”

We are going to load up this data and see if people who follow the news more often say prices are more important to them, and rate Trump more negatively on the issue. We are also going to load up the response data for my yet-to-be-released April survey, which asked how often respondents post or comment about politics on social media, from “never” to “every day.” The results are shown in the graphic below:

Among the least-engaged respondents in our poll, 41% named prices or inflation as the single most important problem facing the country today, whereas among the most-engaged respondents, only 25% did. That’s a 16-point gap — in the opposite direction of what the echo-chamber theory would predict. The Americans who watch the most news and spend the most time on social media are the LEAST likely to say prices are important and the MOST likely to approve of Trump’s handling of them. People who aren’t online are the most focused on prices and most disapproving of Trump.

To be sure this result wasn’t noise, I ran weighted logistic regressions controlling for three-category party ID and a seven-point liberal-conservative scale. The effect persists: Each step up the engagement ladder cut the odds of naming prices as the top problem by about 13 to 15%, even among respondents with the same partisanship, ideology, race, and income.

I got the same effect on Trump’s approval for handling prices: Each step up the engagement ladder raises the odds of approving Trump’s handling of prices. High-engagement Americans, in other words, absorbed the more positive partisan messaging on the economy. Less-engaged Americans, who consume less of that messaging, are the ones putting prices first.

To be fair, MIP and Trump approval on prices are not direct measures of personal finances in either of these polls. But the basic finding is hard to square with the echo-chamber story, where you’d probably expect MIP to be correlated with negative price sentiment.

The people marinating in news and social media are not more focused on prices, they are less.

Subscribe now

People are telling us directly that they care about prices!

I just want to underscore something before concluding, and that is that we don’t really need to be doing all this fancy modeling to explain consumer sentiment. We can just ask people what they think, and believe them! Much of the discourse about the “sentiment gap” assumes that the economic statistical models are right and the people, in aggregate, are wrong in how they’re thinking about the economy.

But what if we flip that on its head? What if the national statistics are just.. missing something?

When you ask Americans who say that their personal finances are worse than a year ago, why they say that, a record high 54% in April 2026 say it’s because of prices. To put that in historical context, here is the full 65-year history of the series (only up to February, since UMich hasn’t updated this plot yet):

The current reading isn’t just elevated, it is at a level the series has only really matched a few times before, during the inflation shocks of the mid-1970s, during the Volcker years around 1980, and briefly in 2008-09 (when, fair enough, the “high prices” answer was probably overlapping with a massive panic over job losses).

Zooming in on recent history below, the current spike has held at roughly that 1970s level even as headline inflation has fallen from 9% back to something near 2%, and even as inflation expectations themselves have come down. Look at the same series since 2014, plotted alongside year-ahead inflation expectations:

For nearly three years, in other words, the share has been hanging out near the worst sustained stretch of stagflation in modern American history.

That alone would explain the historically bad consumer sentiment very well (as shown in my model 4)!

And yet, in each of these previous spikes, nobody was arguing that the price anxiety was somehow fake, or media-manufactured, or a vibes-based illusion. Inflation was genuinely terrible in 1974, 1979, and 1980, and a lot of Americans really were losing their jobs in 2008. We did not need to construct elaborate theories about why people were misperceiving the economy — people were, in fact, perceiving the economy correctly. This polling at least suggests the possibility they are doing so today, if you get out of the habit of viewing the economy only through the lens of unemployment and year-over-year inflation in the PCE.

And there is, I think, the broader point. To argue that we should ignore what people are telling us directly about prices, you need a pretty strong argument that the polling data is biased, AND an otherwise very convincing explanation for why consumer sentiment is now seemingly divorced from the usual economic statistics (I say “the usual” because my argument is that people just needed to be a bit more creative here).

But what they have offered instead is a series of much more elaborate explanations — that respondents are polluted by Twitter, that they are ignoring their otherwise healthy bank accounts because of news coverage, that some latent vibe variable is doing the psychological driving for fifty million Americans — that turn out, when you actually run the numbers, to be either weak, contradicted by the data, or both.

Here, we may just want to rely on our old friend Occam’s razor and accept the most parsimonious explanation. People are telling us, in numbers comparable to the worst stagflation stretch in living memory, that prices are wrecking their personal finances. They are telling us that consistently, across polls with different question wording produced by different firms, and on their own in social media posts, interviews, and conversations with their neighbors in day-to-day life.

People are telling us they’re upset about prices even as the rate of inflation has moderated and even as media coverage of inflation has cooled off. And if you look at the structural data on excess price levels, they really have a point.

Until somebody produces a good, empirical reason to ignore what millions of people are saying every day, I think the burden of proof sits firmly with the people arguing we should.

Subscribe now

Excess prices explain most of the gap in consumer sentiment

To recap what we have so far. The same model that explains post-2020 sentiment also fits the 1970s within a few RMSE points. Price perceptions still predict sentiment after stripping out the news-explained component. And the most media-engaged Americans are less likely to name prices as the top problem, not more. Excess price theory looks like a solid empirical explanation of consumer sentiment, while “social media and news contagion effects are driving dissatisfaction” does not.

Yet this model still has clear limits, and I want to be upfront about them.

The biggest one is this: Model 2 fits 2022 through 2024 quite well — residuals under 7 points across all three years, and 2024 is nearly perfect at +0.6 — but it misses 2025 and 2026 by about 11 and 12 points, respectively. That’s about the same size as the misprediction in the early 80s, so not unheard of, but still worth addressing. Over the last 6 years, excess prices explain the original collapse in sentiment from 2022 through 2024. But they do not explain the renewed softness over the past 18 months. If I’m putting on my speculation hat, that gap is presumably the result of cumulative Trump-related shocks to the global economy (tariffs have created a lot of price pressure and uncertainty that is only just now really impacting aggregate statistics), and it’s possible that news and social media play a role in this remainder, too!

So, a few things this analysis does not prove, and that I am not claiming:

• It does not show that prices are the only driver of sentiment.

• It does not show that media coverage has no effect — only that price perceptions contain a large signal that is independent of media coverage.

• It does not show that spending time on social media inundated with posts about prices has no effect on your psyche. But it does show that people who spend less time on social media are actually more negative about prices.

• It does, though, show that excess prices outperform the rival explanations I tested — news sentiment and conventional macro fundamentals — across two very different inflation regimes.

The strongest version of my claim isn’t “prices are the whole story.” It’s that prices beat the competing explanations, and they explain most of the gap.

But the research in this article does show something important that I think people should acknowledge: While the consumer-sentiment gap is not well explained by the standard macro story, a model built around the excess level of everyday prices — food, shelter, transportation, the things people actually buy on a weekly basis — explains much of it. And it explains far more than news sentiment or social media habits do. Plus, it survives a much harder historical test than I honestly expected when I started the project. The fit to the 1970s, with the same coefficients trained on different data, is the result that surprised me the most.

And a broader lesson of this exercise may be that economists and political commentators are very often tracking the wrong numbers, in large part because they aren’t looking at the data that is closest to the people, and that data that best proxies their current anxieties.

In my Tuesday article, I used the metaphor of the drunk man looking for his car keys under a lamppost:

Macroeconomists who insist that the unemployment rate is low, the stock market is up, and Americans should therefore be feeling fine about the economy are a little like the old joke about the drunk searching for his car keys under the streetlight. He’s not looking there because that’s where he dropped them. He’s looking there because that’s where the light is.

Here, the data on how people actually feel about the cost of living — aggregated from individuals’ lived experiences — is over in the dark, and analysts by and large are not all that interested in walking over to it.

Political commentators focus almost reflexively on aggregate statistics and on rates of change. Inflation is down! Wages are up! Unemployment is low! The stock market is near a record! Those are all genuinely true statements about the economy in 2026. They are also, it turns out, evidently not the things people actually care about when determining their views on the economy.

So when analysts keep insisting that the unemployment rate is low and Americans really ought to be feeling fine, and Americans keep responding to surveys saying that, no, actually, things are bad — the gap between those two stories is not measurement error. And it is not, broadly speaking, a vibes-based illusion. It is a category error. Economists are tracking the wrong variables. How people actually say they feel is the ground truth of sentiment, not what you can predict with a model.

If you want to understand why Americans still say the economy feels bad, even though the rate of inflation has come back down to something like normal, the first place to look is not cable news, or social media, or the irrationality of the median voter. It is the price level that people are still paying at the pump, at the grocery store, at the bus station, day in and day out. That number is bigger than it used to be.

Until someone shows otherwise, we can just believe Americans when they tell us the following: they’re stressed out because things are expensive. And if you look at the data correctly, that’s a valid explanation.

If this analysis was useful to you, consider supporting this work with a paid subscription. Paying members of Strength In Numbers get weekly premium posts, bonus podcast episodes, and the satisfaction of supporting independent data-driven political journalism — including our monthly poll (that subscribers can add questions to!).

Our session will take place on Thursday, April 23, at 11:15 AM at the America Votes Summit, which is being held at the Walter E. Washington Convention Center (the address and directions can be found here).

Elliott and David will dive deep into the hard data they’re relying on to make sense of the 2026 midterms, from special elections to rigorous polling. We’ll also be releasing an exclusive new poll explaining why voters are down on Democrats—and how to win them back. Plus, we’ll be taking your questions in an extended Q&A session!

The first 75 listeners who RSVP can attend free of charge. Just click below:

RSVP for the Strength In Numbers podcast

We really hope to see you there, and we’d love to chat with you after the show, too!

In this week’s live recording of the Strength in Numbers podcast, G. Elliott Morris and David Nir preview new results from the April Strength in Numbers/Verasight survey and dive deep into the big economic debate of the moment: Why do Americans feel so terrible about the economy when unemployment is near historic lows? Elliott walks through his new analysis of “excess prices” and what it means for the political environment heading into 2026.

Here are the big takeaways:

• Trump’s numbers hit new lows across the board. A sneak preview of our new Strength In Numbers/Verasight poll finds Trump hitting his worst approval overall, and on prices in particular. The generic ballot sits at Democrats plus 8, consistent with a D+7 average over the past year. That’s a margin large enough to flip the House and potentially put the Senate in play. More to come next week!

• Low economic sentiment is not just a social media problem — it’s an excess prices problem. Consumer sentiment just hit its lowest level ever recorded, and a popular theory holds that the news and social media have simply broken people’s brains about the economy. Elliott’s analysis points to a different culprit: Prices are roughly 10% above where they’d be if pre-COVID inflation trends had continued, and traditional economic models fail to account for this because they use year-over-year inflation rather than cumulative price shocks. When you swap in a measure of “excess prices” (adjusted for how accustomed people were to inflation beforehand), the models predict current consumer sentiment far more accurately, including in the high-inflation 1970s.

• There’s no quick fix for low sentiment, and that spells trouble for both parties. Actual deflation — bringing prices back down — is essentially impossible without triggering a recession, as Paul Krugman and others have argued. Even in the best-case scenario where inflation returns to 2%, Elliott estimates it would take six to seven years for excess prices to normalize and consumer sentiment to recover — and with tariffs, the Iran war, and mass deportations pushing inflation higher, the realistic timeline is closer to a decade. Until then, voters will likely keep punishing whichever party is in power, fueling what Elliott has called “an era of one-term presidencies.”

Share

If you missed our video livestream, you can watch it by clicking play on the web version of this post at gelliottmorris.com. We record the podcast live every Thursday at 2:00 PM Eastern. We always discuss a few pre-planned topics and then answer questions submitted live by viewers.

You can also subscribe to us on your favorite podcast app to listen on your own time. And if you do listen via one of those apps, please drop us a five-star rating and review if you feel we’ve earned it — it really helps people discover the show!

A reminder that paid subscribers to Strength In Numbers get to participate in our live Q&A!

You can also read the transcript of our conversation by clicking the headline of this article to take you to the web version of the podcast, then clicking the button just below the byline that looks either like a piece of paper or is labeled “Transcript,” like so:

If you’re a reader of Strength in Numbers and haven’t yet subscribed to David’s newsletter, head to the-downballot.com.

Subscribe to The Downballot

And if you’re coming from David’s site, subscribe here to get the numbers behind the news.

Subscribe to Strength In Numbers

My Tuesday Deep Dive is a weekly premium article for paying subscribers of Strength In Numbers. Subscribers get exclusive posts at least once a week, plus early access to new data releases and bonus podcast episodes.

Subscribe now

EDIT: After you read this, read part 2 here.

The economist Jared Bernstein and I were not colluding on Sunday when we both wrote about the effects of long-run inflation on consumer sentiment. I guess the topic was on our minds because the new University of Michigan Index of Consumer Sentiment release last week was literally the worst release ever in the history of the survey:

Here’s Bernstein’s post:

Jared Bernstein

Weekly Wrap-Up: Inflation, Econ Vibes, and How the White House Decided to Go to War

I know I’m always saying this past week was especially consequential, but don’t blame me if events keep coming at us with such speed and fury…

Read more

10 days ago · 66 likes · 10 comments · Jared Bernstein

Bernstein and I are both writing about a popular puzzle of economic statistics right now: the broad disconnect between objective indicators of economic growth and consumer sentiment. Here is a chart from The Economist that really illustrates what we’re exploring in this discourse:

People have given various explanations for this disconnect, as I wrote on Sunday. One camp of analysts argues the economic statistics should be believed — that Americans are ackshually fine and the differential is entirely psychological, a product of negative news and social media sentiment.

The strongest version of this argument — call it the Stancil view, after Will Stancil, its loudest proponent on Bluesky — goes roughly like this. The unemployment rate is historically low. Real wages are rising. The stock market is near all-time highs. GDP is growing. By every standard macro yardstick we’ve ever used to measure whether Americans should feel good about the economy, Americans should feel good about the economy. If they don’t, the problem is not the economy. The problem is that they are being told — by cable news, by social media, by partisan operatives, by doom-curious podcasters — that the economy is bad. Under this view, if you strip away the bad-vibes megaphone, sentiment would snap back into line with the fundamentals. In the loudest form of the argument, the “sentiment gap” isn’t real, it’s measurement error produced by the information environment.

A version of this argument was very popular on Twitter in 2024, too, with people arguing the ICS was biased and economic data showed voters would reward Democrats for the post-COVID economic recovery. Welp...

I want to take Stancil’s argument seriously, because the people making it are not stupid, and because parts of it are obviously true. Real wages have outpaced inflation over the last 18 months. Unemployment is low. News coverage of the economy has been more negative than the underlying data would justify (although post-Iran, it’s not clear that’s true anymore).

But as I’ll show below, the Stancil view cannot account for the bulk of the sentiment gap — because even after you control for the news environment, the gap stubbornly refuses to close. Something else is doing the work.

Another explanation for the gap in actual and predicted consumer sentiment is that the vibes are off generally and that this vibes bias is “polluting” (so to speak) responses to surveys. One of my favorite theories is that the vibes are off largely because things are expensive, and you can see pictures of other people buying expensive things on social media, and that makes you feel big sad. People need to stop posting Lambos on Instagram!

But another explanation is that consumer sentiment is down because prices are stubbornly high. This is not something that would be accounted for in official inflation statistics, because those measure year-over-year changes in prices, when what we’re really talking about when we talk about high prices is prices now compared to what they were 3 or 4 or 5 years ago. When the rate of price change goes back down, it’s not like people forget about the original inflation shock.

I talked about this in my interview with Paul Krugman last September:

Back then, I hadn’t dug into the data or done the modeling to really prove (or disprove!) this hypothesis. But now, I think we have the data and analytical strategy to show pretty concretely that low consumer sentiment is being driven by high prices.

One of those variables is what Bernstein posted on Sunday — a new measure of medium-term excess inflation above expectations:

The other is what I shared: the percent of Americans who say high prices are hurting their personal finances is at an all-time high:

The preliminary April reading above is about 54% (that’s 54% of the 50% of people who say their personal finances are getting worse) while the record previously was <50.

So we’ve got two independent measurements of the same underlying dynamic, both flashing red while consumer sentiment plunges. Let’s get real nerdy and dig into the hard data on what actually explains consumer sentiment today.

Here’s the spoiler, so you know where we’re headed: the single best predictor of how Americans feel about the economy is how Americans feel about prices. The second best predictor is with actual structural data on prices. It’s not unemployment, not the S&P 500, not the year-over-year rate of inflation or the tone of economic news coverage. Prices — specifically, the share of Americans who tell the University of Michigan that higher prices are making their household finances worse than they were a year ago — beat every other variable I tried, in every specification I ran, by a comfortable margin. The rest of this piece is me showing my work.

So, why is the index of consumer sentiment so low when aggregate economic statistics are otherwise pretty okay-ish? Because people don’t like paying so much money for things. Yes, it really is that simple.

I’m not going to paywall this article because I want as many people as possible to digest this data and my arguments. But if you have the ability, please consider supporting my work with a paid subscription to Strength In Numbers.

Subscribe now

1. Prelude: gas prices and consumer sentiment

Before getting into all the fancy modeling and data processing for this piece, let me revisit what I wrote on Friday. I showed that there is a negative relationship between gas prices and presidential approval, as shown here:

And I teased the following:

Next week, I’ll take a look at how presidential approval ratings over the last decade have responded to media sentiment. Are voters responding to prices, or news coverage of prices?

Stick with me for a second, because this will be supportive evidence for the main argument in the next section. I built a statistical model to predict the month-over-month change in a president’s approval rating from the change in gas prices over the same period, plus the contemporaneous tone of economic news from the San Francisco Fed.

The theory for this model is simple: Expensive gas should drag approval down, whereas a more positive news environment should push it up. And that’s basically what we see. Here are the results by president, across four terms from Obama through Trump 2:

A one-standard-deviation rise in news sentiment — the gap between an average month and an 84th-percentile month — is worth about half a point of approval. A comparable rise in gas prices costs the president about 0.2 points. Some of these coefficients are statistically noisy, mostly because of sample size.

This news sentiment index turns out to be roughly as predictive of presidential approval as monthly change in CPI or the S&P 500. I also ran one model pooling all presidents and adding more macro-economic variables. The results are visualized below. News sentiment is positive and green, inflation and stock market crashes are negative and red.

The real lesson from this exercise is that gas prices matter, but they are not the only things that matter. And news sentiment, the other headline variable in this section and the Stancil boogeyman, turns out to be a pretty weak driver in its own right of how people feel about the president.

And if the sentiment of economic news articles is only weakly predictive of how people feel about the president, should we really expect it to fully explain the much bigger gap between hard economic data and the University of Michigan’s Index of Consumer Sentiment? My hunch is that the news-narrative theory simply can’t carry all that analytical weight.

Well, we don’t have to theorize! We can just do math!

Share

2. Anxiety about prices is driving down consumer sentiment

This is where Bernstein and I re-enter the chat. I argue that the disconnect between the official numbers and how Americans feel about the economy comes down mostly to one thing: nominal prices. It’s not the rate of price change that’s bothering people. It’s the level. When you fill the grocery cart, pay the rent, cover the utility bill, or renew the car insurance, you’re paying today’s sticker — not today’s sticker minus last year’s. When you compare what groceries cost today to what they cost in 2019, things still look terrible even though inflation has cooled off. As I said earlier, people don’t forget the original price shock just because the rate of change stopped rising.

But inflation is a popular data series and the default way people think about prices. And that’s why we’ve only seen charts that look like the following. The black line in this plot is the 2-month rolling average of the UMich Index of Consumer Sentiment (ICS from now on). The light yellow dotted line shows what you would predict the ICS would be based on three variables:

1. The unemployment gap — how far the current unemployment rate is above or below the CBO’s estimate of the natural rate for a given time period (this helps us account for the long-term secular decline in the unemployment rate)

2. Year-over-year change in the consumer price index for all urban consumers

3. Year-over-year change in the S&P 500 — self explanatory

These variables enter a Bayesian predictive model that is trained on all dates from January 1997 through December 2019. Then, based on the relationship between those variables and the ICS on those dates, we can predict what consumer sentiment “should be” for the period from January 2020 to April 2026. That prediction is shown as the thin, dotted yellow line on the chart:

Note the large gap between the thin yellow line and the black line. That’s the gap shown in the Economist chart, and what we’re trying to explain.

But what’s this? Another dashed yellow line below it — that looks like a much closer match to the post-COVID ICS data? That’s right! This model swaps out the change in headline CPI reading for year-over-year change in core PCE prices — the Fed’s preferred inflation measure that omits noisy changes in food and energy. This turns out to offer a much better forecast of prices post-COVID — and also, as we’ll see, in historical back-testing (so it isn’t a fluke).

Now we can start adding other variables into our model to predict the ICS. Instead of just hard national economic statistics, what if we include that news sentiment variable from earlier? Per the chart below, it seems like a pretty good match!

In total, I train six models, each layering in a new variable on top of the last, so we can see exactly how much of the post-2020 “sentiment gap” each piece of the data explains.

Here are the six models, from simplest to most complete:

1. Baseline using headline CPI: unemployment, headline CPI, and the S&P 500. This is the naive macro setup.

2. Baseline using core PCE: same thing, but swapping headline CPI for core PCE — as discussed above

3. Plus gas prices: Model 2 plus the year-over-year change in regular gas prices.

4. Plus news sentiment: Model 3 plus the San Francisco Fed’s Daily News Sentiment Index, which scores economic news coverage on a positive-to-negative scale.

5. Plus a measure of cumulative price inflation in the last 5 years: Model 4 plus the percentage gap between the current level of core PCE prices and where the pre-COVID trend said they should be. This is my version of Bernstein’s excess price variable. More on that below.

6. Plus felt prices: Model 4 plus the share of Americans who tell the University of Michigan that higher prices are making their household finances worse off than a year ago.

And a brief segue on the price trend mentioned in #5. Bernstein runs a linear model on the monthly CPI readings from 2014 to 2019 and then projects what the CPI “should be” for 2020-2026. That gives us a graph that looks like this:

Here is my version of the same calculation, but using Core PCE instead, since I found it’s a better predictor of consumer sentiment:

The gap between the black line and red dashed line in the plot above is a measure of how much extra Americans are paying for everyday goods and services than they would have if inflation had stayed steady at the normal 1990-2019 rate (around 2%). Even though inflation has cooled off, prices as measured by the PCE are still 10% above what they would be under the pre-COVID trend. You can see that clearly here:

So now we’re gonna run all our models: the standard fundamentals models, the fundamentals plus gas prices, plus news sentiment, plus excess price inflation, plus attitudes (the percent of people saying high prices are impacting their finances).

Here is every model’s out-of-sample prediction, plotted against the actual UMich reading:

And here’s a version that might be easier to read:

The headline result is that the models using felt prices and cumulative prices both close almost all of the post-2020 sentiment gap — while the standard macro models miss by a mile.

The baseline model using headline CPI misses the post-2020 monthly ICS readings by an average of 20 index points. Switching from headline CPI to core PCE cuts that gap roughly in half, while adding gas prices barely helps.

The first big result of this inquiry is that adding news sentiment decreases error, but the model still under-predicts sentiment’s collapse. If all of the “vibecession” was due to news media coverage, this error would be zero. But note that in 2025 and 2026, there is still a huge gap between press coverage and how people feel the economy is working for them.

It’s only when we add a measure of price levels — either the structural Bernstein-style cumulative price shock or the direct survey of how consumers feel about prices — that the models catch up to reality. Predictions from those models are shown as the red and teal lines of the main plot:

These are very good predictions! Mean absolute error post-COVID with these souped-up models is just 5.8 index points using Bernstein’s excess PCE metric, and 4.7 points using my direct measure of price anxiety. And these are twice as good as the third model, which takes the baseline predictions and adds the news sentiment readings on top.

The single best predictor of consumer sentiment in any model I ran is the share of Americans who tell the University of Michigan that higher prices are making their household finances worse. A one-standard-deviation move in that variable is associated with a 5.4-point drop in the sentiment index, on average, holding everything else constant. That’s larger in magnitude than unemployment, inflation, the stock market, news sentiment, and even Bernstein’s structural data on price shocks.

The thing that best predicts how Americans feel about the economy is how Americans feel about prices. (And if you don’t have that survey data, then the best predictor is the hard data on prices.)

There’s one more subtle piece of evidence worth pointing out here, because it bears directly on the “it’s all news coverage” theory. When I add the felt-prices variable to the model, the coefficient on news sentiment shrinks by about a third — from +4.80 per standard deviation in the news-sentiment-only spec, down to +3.17 per standard deviation once felt prices are in the model alongside it. News sentiment still has its own independent signal even after that shrinkage. But the shrinkage itself is telling: a meaningful chunk of what “news sentiment” was picking up in the simpler model was actually price anxiety bleeding through into the tone of coverage.

This is self-explanatory because most of the time, when reporters are writing economic stories, they are writing about prices. A story about how Americans are stretched thin at the grocery store is a story about prices. When cable news runs a segment on gas hitting $4 a gallon, they are writing about prices. Some of the “media narrative” channel that analysts want to blame for the sentiment gap is really just the same price story, dressed up in a different variable.

That doesn’t mean news coverage has no effect of its own — it clearly does, and the +3.17 coefficient that survives is real. But it does mean the gap between “the economy is fine and the vibes are bad” and “the economy is not fine because prices are high” is smaller than it looks. A lot of the “vibes” are prices.

Subscribe now

3. Stress-testing the model

Because models can be overfit to the era they’re trained on — picking up relationships that happen to hold in one window but fall apart in another — a good out-of-sample forecast in one direction isn’t, by itself, proof that you’ve found a real structural relationship. Maybe the felt-prices variable just happens to line up with sentiment in the post-2020 window because both series cratered at the same time for unrelated reasons. Maybe any two series that both fell off a cliff in 2022 would look like they explain each other. The only way to rule that out is to ask whether the same coefficients that work in one regime also work in a very different one.

If our models generalize well, they should work in both directions: at the beginning and end of our time series. If I train models on the recent era (2004-2026) and use them to predict the late 1990s — a totally different macro regime, with a dot-com boom and unusually high sentiment — predictions should still be in the right ballpark.

Here is how the six models do on that back-cast:

We see the same story. The three baseline models (headline CPI, core PCE, gas prices) all under-predict late-1990s sentiment by 11 to 15 points. Adding news sentiment barely helps. But adding either the Bernstein cumulative price shock or the Michigan attitudes variable closes most of the gap. This is reassuring: The models that work post-2020 are the same models that work in 1997.

Share

4. It’s the prices, stupid

So what does all this mean?

First, if you are trying to model consumer sentiment — or presidential approval, or the generic ballot, or anything else that depends on how Americans feel about the economy — and you are using headline CPI as your inflation variable, consider using Core PCE instead. Even better, calculate a cumulative-price-shock variable.

Second, the debate about whether the post-2020 sentiment gap is “real” or vibes- and media-driven is mostly a false choice. The vibes are real, and they are tracking something real: prices that are still roughly 10% above where the pre-COVID trend said they should be, and that have barely budged in three years. News coverage matters at the margin. The labor market matters at the margin. But the thing doing most of the work is the price level — and the lived experience of the price level.

Finally, the big takeaway from this is that sentiment metrics aren’t made up/manufactured/useless just because aggregate statistics on economic growth are “disconnected” from perceptions. This investigation suggests that people who believe this are simply looking in the wrong place, consulting data that don’t tap into the price anxiety of the average American.

This is a little like the old joke about the drunk searching for his car keys under the streetlight — not because that’s where he dropped them, but because that’s where the light is.

Just because unemployment is low doesn’t mean Americans shouldn’t be bothered by the fact that a gallon of milk, a tank of gas, and a month’s rent all cost meaningfully more in absolute dollars than they did five years ago. Yes, unemployment and CPI inflation are poor predictors of consumer sentiment in 2026 — but that’s because they tell us nothing about how everyday Americans feel about price levels.

Poor economic sentiment isn’t about the news media and politics, or social feeds and people driving Lambos on Instagram. Or at least, it’s not mainly about those things. When it comes to how Americans feel about the economy today, whether you are measuring using objective structural price data or the polls, it’s the prices, stupid.

Subscribe now

Share

If you’re a frequent reader of Strength In Numbers, I’m confident you will get a lot of value out of a paid subscription. Paid subscribers get access to Deep Dives like this one, plus the full archive of everything else I publish.

Subscribe now

Related Articles

Edit: I am not in the habit of linking criticism of my work here, but here is one post bashing me for having my head in the sand about economic anxiety and only just now “figuring out” that people were upset when this data came out. The post is snarky but raises the possibility this article was not framed clearly enough/that it could turn off people who are unfamiliar with my past work on this subject, so here’s a short note:

I have been talking about high nominal prices dragging economic sentiment measures down since mid-2024. I wrote about this at FiveThirtyEight before Disney shut us down, and here is an interview I did with Paul Krugman about this last September:

Krugman and I talked about this partly within the context that other smart people have been putting together various statistical models to predict economic sentiment with objective economic indicators, and have concluded that sentiment is lower than the data justifies. With a limited model, you cannot understand why sentiment is so much lower than “predicted,” and are left to either (a) accept the gap as an interesting data point in and of itself and thoughtfully explore possible explanations for it based on other economic data and interviews with real people, or (b) ascribe causality for the gap based on your own pet theories of how voter psychology works.

On (b), the most prominent theory — pushed mostly by really left-leaning loud online commentators, not economists — is that the news media and social media are sources of negative information for people, which drags down consumer sentiment. This camp argues that people are actually broadly doing fine right now, they’re just being misled about the state of the economy by TV news hosts and clips of people filling up their F-150s at the gas pump.

The data presented in this article, however, fills a gap in the predictive models scholars are running by asking people directly if they are currently anxious about prices. This is close to what I talked with Krugman about, which is accounting for cumulative medium-run inflation in things most people spend a lot of their income on (e.g., housing, food at home, and car payments). Government economic statistics — namely, annual inflation — do not account for the cumulative strain put on households by step increases in the price paid for a gallon of milk (not dissimilar these days to the price of a gallon of gas). But the polling data here does.

So, why is there a disconnect between official measures of (broadly speaking) how the economy is doing and how people say they are doing, personally? Maybe because everyone is being hoodwinked by the news media and their social media algorithms, and the economy is actually going gangbusters right now! Real wages are going up by 0.6% annually, hooray! Or maybe the surveys are right, and things really are stressful for most people in America right now, but analysts are looking at the wrong data to properly explain how they feel.

So that’s the context for this piece. Also, in case it’s not obvious, “mystery variable” in the title is supposed to sarcastic; I think it’s obvious that people care about prices!

Back to the article:

Leading off: Consumer sentiment just hit the lowest level ever recorded by the University of Michigan — lower than during the financial crisis, peak COVID, stagflation, and Biden’s presidency. I write about what that means for 2026, and about an under-covered variable in explanations of low consumer sentiment.

On deck this week: As I mentioned Friday, I’ll have a Tuesday Deep Dive on the relationship behind economic news sentiment, prices, and consumer sentiment, and what this means for voter psychology and the midterms. Friday’s Chart of the Week is TBD.

Thank you for reading Strength In Numbers for another week! On with the data.

Subscribe now

1. One measure of how consumers feel about the economy just hit its lowest level ever recorded

The University of Michigan has been measuring consumer sentiment since 1952. On Thursday, economist Justin Wolfers flagged that the April reading came in at 47.6 — the lowest in the survey’s 74-year history. That index value is lower than for the depths of the 2008 financial crisis, the worst of the post-COVID inflation surge, the 2022-23 inflation spike, and any point during the stagflation of the early 1980s.

Charles Franklin at Polls & Votes put together a useful comparison of consumer sentiment in Trump’s first term vs his second. In the first 23 months of term one, the Michigan index averaged 97.5. In term two so far, it’s averaged 55.5 — and now sits at 47.6. That’s a 50-point swing from the first-term average to today’s reading.

Two things stood out to me about this new data release. First, here’s the thing about the first term that’s easy to forget: the economy was basically the only thing Trump had going for him politically. His approval ratings were otherwise underwater on pretty much every policy domain. His signature legislative achievement — the 2017 tax bill — was unpopular, as were his immigration policies. And his general vibe as head of state was divisive (less so than now but still a serious drawback).

Yet consumer confidence was sky-high from 2017 through early 2020, the stock market was booming, and unemployment was low. Republicans still lost 40 House seats in 2018.

So what’s going to happen when Trump’s policies are underwater across almost every issue, AND economic sentiment is literally worse than it’s ever been, according to this rating? We’ll have to wait to see exactly what, but signs are pointing to a solid rout.

In short, it matters not just that Trump is more unpopular than he was at this point in 2018, but how he’s more unpopular.

It’s both “vibes” and prices

There’s a theory popular in certain corners of the very online left that consumer sentiment is inexplicably low right now because of the way the news media is covering the economy. The theory points out that in most of 2024 and 2025 the labor market was doing fine by historical standards, but people still rated the economy poorly. The sentiment numbers, per this view, are a product of news and social media amplifying bad economic stories and data and dragging down the national mood.

I’ve got some data I’m going to dig into next week that speaks to this more directly. But for now, look at the following chart from the Michigan survey itself. It tracks the share of consumers who cite high prices as the reason they are personally struggling financially.

And here is the complete time series:

Before 2021, this number hovered near zero percent. Empirically speaking prices were a non-factor in how people viewed the state of the economy.

Then, everything changed. The share of adults citing high prices asa sources of anxiety went exponential during the 2021-22 inflation spike and never came back down. It’s now above 50%, likely because of the gas prices spike from the war in Iran.

This trend actually looks similar to a chart of cumulative change in food prices since 2014:

And the price of shelter:

While inflation dropping from 8% to 3% reflects a “cooling off” of the economy, but evidently, people still mostly just see high prices for things and get upset about that. And fair enough!

My theory is that price levels account for much to most of the “puzzle” of why consumer sentiment is lower than you would predict based on the historical relationship between CPI, unemployment, the cost of money and etc. I will test the models and see if that’s true in reality too. The theory here is simple: prices didn’t used to dominate how people thought about their personal finances. But now they do, and that’s a source and level of anxiety that doesn’t show up in historical data on inflation, unemployment, etc.

Maybe that is unsatisfying as a grand political theory. But I think it’s workable descriptively and helps resolve some of the puzzle of voter psychology here. In terms of causality, there is still some work to be done.

Share

2. What Strength In Numbers published last week

This week’s first-ever monthly deep dive podcast episode on Monday featured focus group insights on what voters really want from their representatives right now:

The Deep Dive column on Tuesday covered new research that shows why ideological moderation is an incomplete and potentially suboptimal strategy for parties to win back defectors in non-competitive states:

Democrats posted some of their biggest election swings of the 2026 cycle so far in Wisconsin and Georgia on Tuesday night — more evidence of the shifting blue midterm environment:

Trump is underwater in every competitive House and Senate seat, according to my latest state and district level MRP estimates:

The normal weekly podcast explained how that MRP modeling works, and covers other newsy topics like the Iran ceasefire and Tuesday’s elections.

And Friday’s Chart of the Week looked at the relationship between gas prices and presidential approval — timely, given the Iran-driven spike:

If you’re a frequent reader of Strength In Numbers, I’m confident you will get a lot of value out of a paid subscription. You’ll get access to all of my Tuesday Deep Dives, monthly polling data, and more.

Subscribe now

3. Even more numbers!

• Nothing really aside from the consumer sentiment data above.

• I did think this piece on AI “polls” from Eli at Nate Silver’s blog was good, and it links to lots of interesting research (including mine!) on the capabilities of what industry folks are calling “synthetic sampling” or “digital twins.”

  • My point on this, starting back in 2024, has been that you shouldn’t think of these tools as polls, but as predictions. They use large language models and other systems to predict how “people” with certain demographic traits would respond to survey questions had they been asked them. This is similar in principle to how something like MRP works, except for the key factor that the information the model has about the beliefs held by certain types of people comes from the LLM, not underlying survey data.

  • That’s not polling, but maybe it could be a decent predictive model if you don’t have any other information about how people in a certain geography are feeling.

And that’s it for this week. I spent most of the weekend hiking instead of reading the internet, so section 3 is shorter than usual. Back to normal programming tomorrow. Here’s a picture of a gopher I saw waddling around in the woods:

Strength In Numbers will be back in your inbox Tuesday with a deeper look at the consumer price data.

Got more for next week? Email your links or add to the comments below.

Elliott

Leave a comment

Oil prices plunged this week after the U.S. and Iran agreed to a two-week ceasefire. Crude fell more than 15% on Tuesday alone — its biggest one-day drop since 2020 — settling around $94 a barrel. That’s still well above the ~$70 level before the war started, but it’s a dramatic reversal from the $110+ prices of recent weeks.

But gasoline prices haven’t declined yet. They usually do, with a lag of a few weeks — but gas is still up nearly 40% since the conflict began in late February, so it has a long way to fall back to “normal” levels. Here are two charts of oil and gas prices I found online:

One question subscribers have been asking me this week is: If gas prices do fall back to normal levels, will Trump’s approval recover to what it was before the war in Iran broke out? Chart of Trump’s approval, from FiftyPlusOne.news, below. There has been a small rebound in Trump’s net rating over the last week, but it’s inside the uncertainty interval and could be noise.

But this brings up the secondary question of how strong the historical relationship between gas prices and presidential approval is in the first place. So I pulled weekly gas price data from the St. Louis Fed and daily presidential approval estimates from our data at 50+1 going back to 2009 to find out. This week’s Chart of the Week is: a data-heavy look at the gas price–approval connection across four presidencies. Even if we don’t learn anything concrete, I think these are some fun graphs I haven’t seen elsewhere.

Subscribe now

The big picture

Here’s the full 2009–2026 view of both gas prices and presidential net approval rating. Gas prices are on top, with net presidential approval on the bottom. Shading represents which party is in control of the presidency.

The first thing that jumps out is that two series often move in the opposite directions, especially under Obama and Biden. When the national average price of a gallon of gas increases, presidential approval decreases. Yet the trend is not universal. During both Trump terms, the connection is much weaker.

Note, second, that there is a general secular trend downward in presidential approval for the first six months to a year a president is in office. That seems to persist regardless of what happens to gas prices. Then, there are occasions when gas prices are flat or falling, while presidential approval still moves down. That’s because voters care about things other than gas prices.

Share

Approval and gas prices by president

To get a cleaner look at the relationship between these two trends, I standardized both series within each presidency. This adjusts for inflation in gas prices and for each president’s baseline approval level (polarization appears to make presidents less popular, and approval less responsive). In the charts below, gas prices are inverted — so when the two lines move together, it means higher gas prices correspond to lower approval (or vice versa).

The correlations in the charts below reinforce the finding from above: Obama (correlation coefficient of 0.30) and Biden (r = 0.34) show a moderate relationship between gas prices and approval. Both Trump terms show almost none — r = 0.08 in his first term, r = 0.05 in his second (though it’s still early).

Under Obama, gas price spikes in 2011–12 lined up neatly with approval dips. Under Biden, the 2022 gas price surge tracked almost perfectly with his collapse in the polls. But Trump’s approval has not been as responsive to prices at the pump. That’s partly because gas was pretty cheap for most of his first term, so we don’t have a whole lot of examples of high gas prices under Trump. We are currently pushing the boundaries on that, however.

Here’s the same data from the four charts above, but as a scatter plot. Gas prices (z-scored within each presidency) are on the x-axis, while net approval is on the y-axis. Each dot is one day.

Obama and Biden both show clear negative slopes — higher gas prices, lower approval. But both Trump lines are basically flat. Gas prices explain almost none of the variation in his approval. Again, that’s try historically, but might not remain true in the future!

Share

Month-to-month changes do correlate across presidents, loosely

Finally, let’s take a look at monthly changes rather than levels. Each dot here is one month: the x-axis is the percent change in gas prices over the last month, while the y-axis is the change in net approval (in percentage points) over the same time period.

Now, we see a modest negative relationship across all four presidents — months where gas prices spike tend to coincide with approval dips. But it’s noisy, and the effect is small. A 10% monthly jump in gas prices is associated with roughly a 1-point drop in net approval. That’s small potatoes in the grand scheme of things.

Share

What this means for Trump’s approval

So will cheap gas “save” Trump? He might get a small bounce from lower gas prices. But the recent price shock is only one event that looks correlated with the president’s falling ratings. Fixing cheap gas doesn’t fix all the other components of his unpopularity.

(And “save” isn’t really the right framing, anyway — it’s just a convenient one that enables me to make a bunch of fun charts!)

The historical record shows that gas prices matter for presidential approval, but (a) less so for Trump and (b) less for other big events. Obama and Biden were both more sensitive to pump prices than Trump was in his first term. I could push this analysis further back in time, but then we risk overestimating change in approval today (which is lower because of polarization).

Trump’s approval rating in term two has mostly been driven by tariffs, backlash to his deportations policy, the Epstein controversy, the war in Iran, and a general sense that he’s not focused on the issues voters care about. Cheap gas would certainly not hurt his ability to win back voters, but it’s not going to reverse a 20-point net negative approval rating that has been falling for over a year straight.

Next week, I’ll take a look at how presidential approval ratings over the last decade have responded to media sentiment. Are voters responding to prices, or news coverage of prices?

This post was free for everyone. Paid subscribers get my Tuesday Deep Dive, exclusive polling data, and access to every chart and model I produce. If this kind of independent, data-driven analysis is useful to you, join a community of like-minded readers who think there is strength in numbers.

Join Strength In Numbers

Related Articles

Here are the big takeaways:

• Wisconsin was a wipeout, and the swing was especially large in rural counties. Liberal Judge Chris Taylor won the state Supreme Court race by 20 points, roughly double the margin of other liberal justices in recent elections and the biggest win since 1999. But the results ran far beyond the headline contest: Democrats won the mayor’s race in the conservative stronghold of Waukesha and flipped the county executive seat in Portage County by more than 30 points after losing it narrowly four years ago. Some of the sharpest swings came in Wisconsin’s most rural counties, where the GOP is usually dominant, raising the possibility of a bigger-than-expected Democratic wave in November.

• New modeling shows Trump is underwater in 135 GOP-held House and Senate seats. Using a technique called multilevel regression and post-stratification (aka “Mr. P”), Elliott estimated Trump’s approval rating in every congressional district and state. The results: Fully half of all Republicans in Congress sit in districts or states where Trump’s approval is negative. At least 30 GOP-held House seats show Trump more than 10 points underwater — more than enough to flip the chamber. And in every competitive Democratic-held seat Republicans hope to pick up, Trump is underwater there, too.

• Americans overwhelmingly said last month that they want a ceasefire with Iran — but that doesn’t mean they’ll support this one. Our March Strength In Numbers/Verasight poll found 60% of Americans favored pursuing a ceasefire and negotiations with Iran, versus just 29% who wanted to continue military operations. Only 60% of Republicans back the war — a significant defection from a president who normally commands 90%+ within his own party. Meanwhile, prominent MAGA figures including Tucker Carlson, Alex Jones, and Marjorie Taylor Greene, openly discussed invoking the 25th Amendment and impeachment after Trump’s social media posts over Easter weekend.

Subscribe now

If you missed our video livestream, you can watch it by clicking play on the web version of this post at gelliottmorris.com. We record the podcast live every Thursday at 2:00 PM Eastern. We always discuss a few pre-planned topics and then answer questions submitted live by viewers.

You can also subscribe to us on your favorite podcast app to listen on your own time. And if you do listen via one of those apps, please drop us a five-star rating and review if you feel we’ve earned it — it really helps people discover the show!

A reminder that paid subscribers to Strength In Numbers get to participate in our live Q&A!

Share

You can also read the transcript of our conversation by clicking the headline of this article to take you to the web version of the podcast, then clicking the button just below the byline that looks either like a piece of paper or is labeled “Transcript,” like so:

If you’re a reader of Strength in Numbers and haven’t yet subscribed to David’s newsletter, head to the-downballot.com.

Subscribe to The Downballot

And if you’re coming from David’s site, subscribe here to get the numbers behind the news.

Subscribe to Strength In Numbers

Subscribe now

Transcript

If there’s one thing Donald Trump knows how to do, it’s make a winning electoral coalition disappear.

A quick video here reacting to the special elections tonight. These are the first real elections since Donald Trump started his war in Iran on February 28th. There are two races: first, the Wisconsin Supreme Court race, and second, a special election to fill Georgia’s 14th Congressional District — the district left open by Marjorie Taylor Greene, who retired earlier this year.

So let’s take a look at the results.

Wisconsin

Read more

The president’s posts have caused a predictable backlash among Democrats, but even Republicans are calling him out now. On Tuesday, Marjorie Taylor Greene called for invoking the 25th Amendment: “Not a single bomb has dropped on America. We cannot kill an entire civilization. This is evil and madness.”

Tucker Carlson, arguably one of the most influential media figures in Trump’s rise, called Trump’s Easter post “vile on every level” and characterized the threatened destruction of Iran’s civilian infrastructure as a war crime. “It’s vile. It begins with a promise to use the U.S. military to commit a war crime,” Carlson said. “Those people who are in direct contact with the president need to say, ‘No. I’ll resign. I’ll do whatever I can do legally to stop this, because this is insane. And if given the order, I’m not carrying it out.’”

Alex Jones — yes, that one — favors the 25th Amendment route to remove Trump.

But it is worth remembering that Congress does not have to take any of these improbable routes to stopping Trump’s crazed moves in Iran. Legislators could simply pass a war powers resolution that outlawed action in Iran. Democrats in Congress already tried this: On March 4, the Senate took up a motion to discharge S.J.Res. 104, a War Powers Resolution that would have required Trump to withdraw forces from Iran unless Congress authorized action. It failed, 47–53. The next day, the House voted 212–219 to stop Trump’s moves in Iran. Two Republicans — Thomas Massie and Warren Davidson — voted yes, while 4 Democrats crossed over in support.

Polling shows these senators and representatives who are voting with Trump are putting themselves in electoral danger. New estimates I’ve produced show that Trump’s approval rating is below 50% among registered voters in 135 Republican-held congressional seats — 104 in the House and 31 in the Senate. And his approval is below 45% in jurisdictions represented by 44 Republican members of Congress — 34 in the House and 10 in the Senate.

Almost every one of the members who opposed the effort to restrain Trump’s war authority in March represents voters who disapprove of his job performance in general and of the war in Iran in particular. The argument of this piece is simple: Trump is unpopular even in a lot of red seats, but Republicans vote with him because party and interpersonal incentives are currently stronger than district incentives. But the electorate is demanding change and reform, even from Republican lawmakers.

Put differently, this data suggests that House and Senate Republicans who stick with Trump are putting themselves in electoral jeopardy in November.

You won’t find district-level Trump approval data like this anywhere else — it’s original research I produce exclusively for Strength In Numbers subscribers. If you want me to be able to continue publishing data like this, become a paying member today.

Subscribe now

135 anti-Trump GOP-held House and Senate seats

These district-level approval estimates come from a multilevel regression and post-stratification model that I’ve been running on the Strength In Numbers/Verasight poll data collected over the last year. MRP is a statistical technique that combines survey data with official Census demographics and election results to produce estimates of Trump’s approval among registered voters at the hyperlocal level. These local estimates are then aggregated into every congressional district and state. (You can explore the full interactive and read the methodology here.)

Here are the full data on Trump’s estimated net approval in the 135 House and Senate seats that are (1) currently held by Republicans and (2) where voters, on net, disapprove of Trump’s handling of the presidency:

Some Republican incumbents stand out. Start with the front-line seats: In Arizona’s 6th district — which voted for Trump by just 1 point in 2024 and is held by GOP Rep. Juan Ciscomani — Trump sits at 38.1% approval. Brian Fitzpatrick’s PA-01 is at 38.5% Trump approval. New York’s 17th (Mike Lawler) is at 39.0%. In California’s 40th, represented by Young Kim, Trump’s approval is at 39.7%.

In total, there are 16 House seats and 2 Senate seats currently controlled by Republicans, where Trump’s approval rating among registered voters is below 42.5% (net rating below -15) and where Trump won by less than 10 points in 2024. (Sixteen seats is more than enough for Democrats to take back control of the House majority in November.)

But voters in even some pretty red seats now disapprove of Trump overall. Texas’s 21st District — a seat Trump won by 24 points and that is currently held by Chip Roy — is at 46.9% Trump approval. Dan Crenshaw in Texas’s 2nd, also Trump+24, is at 49.9%. Elise Stefanik in New York’s 21st, Trump+21, is at 46.7%. Lauren Boebert in Colorado’s 4th, Trump+18, is at 49.0%. These are deep-red seats where Trump is now underwater.

And Trump is currently underwater in nearly a dozen GOP-held Senate seats. Maine (Susan Collins) is at 41.2%. Pennsylvania, which voted in Dave McCormick in 2024, gives Trump 41.3%. Trump’s at 43% in North Carolina (Ted Budd and Thom Tillis). And both Texas senators — John Cornyn and Ted Cruz — represent anti-Trump electorates, giving him just a 44.1% approval rating.

Share

Do you want to vote for Trump or get re-elected?

Of course, not all anti-Trump voters are Democratic voters — and these estimates are based on the current congressional map, some of which have been redrawn for 2026.

But these are the strongest data I’ve seen on just how unpopular Trump is at the state and district level. These are the geographies that matter to Republican politicians.

And the historical parallels for the 2026 midterms are already brutal. In 2006, George W. Bush sat at roughly 38% approval as Iraq dragged on. Republicans lost 30 House seats and 6 Senate seats, and both chambers flipped. In 2018, Trump was at about 41% approval and lost 40 House seats. Presidents below 50% approval lose an average of 30 House seats in the midterms. Trump is 11 points below that threshold. And the out-party typically gains about 5 points on the generic ballot between spring and Election Day. Democrats are already at D+5 or D+6 — just a few points behind where they were at this point in 2018, before the 2018 blue wave.

So the question Republican politicians should be asking themselves in light of these data is: Do you want to win re-election, or do you want to keep backing your party leader who is threatening to “end civilization” in an unpopular, unsuccessful, unnecessary overseas war? Because these data suggest you can’t do both.

Our April Strength In Numbers/Verasight poll will test support for Trump’s impeachment. Other polls have shown voters split over the question, but those data were gathered before Trump started his war in Iran. Subscribe to get those results in your inbox. Here’s the full House and Senate approval data if you want more.

To support my work generating important data like this, consider becoming a free or paid subscriber.

Subscribe now

Related Articles

This number comes from Blue Rose Research, which recently replicated my earlier LLM voter segmentation work at larger scale. Blue Rose asked thousands of Americans to describe their ideal political party, then used a large language model to cluster the responses. Most self-described moderates landed in left, right, affordability-focused, or anti-system camps — not centrist ones. Compared to my work, Blue Rose finds comparatively more Americans in the left-leaning and anti-system buckets.

That finding arrived almost simultaneously with a second, complementary piece of evidence: a major new paper by David Broockman and Joshua Kalla showing that candidates who move to the “elite middle” on issue positions gain, on average, just one percentage points of vote share — and they often lose votes by moderating on popular left/right-leaning issues. But they use a maximally favorable experimental design where every voter has complete information about politics — in a real campaign, the effect is probably half that or smaller.

These two new research findings suggest a more nuanced view of political ideology in America is more helpful than one that reduces politics to one axis and focuses exclusively on the (minimal) role policy preferences play in driving voter behavior.

Today’s Deep Dive covers new data showing voters in “the middle” are not asking for ideological centrism, despite what centrist analysts have written. And trying to win swing voters through generic moderation produces effects so small and inconsistent that the advice is practically useless.

Subscribe now

1. The hollow center

Last fall, I used an LLM to classify open-ended survey responses about voters’ ideal political party and found that many self-described moderates were not describing especially moderate party goals. Then, in February, a reader named Steve replicated the result with ANES data and a different model. Now Blue Rose has done something similar again, at a much larger scale, with proprietary survey data and yet a different LLM.

These three analyses, on three datasets, with three different models, all come to the same basic conclusion.

Like my original survey, Blue Rose asked Americans to describe what their ideal political party would “argue for or believe in.” Then it placed each respondent into one of five buckets based on their responses: Left, Right, Affordability, Anti-System, and Moderate.

Here is the topline:

• Left: 30%

• Right: 25%

• Affordability: 21%

• Anti-System: 17%

• Moderate: 7%

That alone is striking. The Moderate cluster is the smallest segment in the entire typology, eclipsed by anti-system voters, affordability voters, right-leaning, and left-leaning voters.

But the most interesting finding is what happens when you zoom in on people who call themselves moderate. Here’s the crosstab by self-identified political ideology:

That’s right, only 8% of self-identified “moderates” fell into the Moderate cluster when asked what they want their party to fight for. The other 92% gave responses that clustered into left, right, affordability-focused, or anti-system camps. These voters describe their identity in ideological terms when forced to pick a label, but do not use centrist ideological language when they are allowed to think about politics in their own terms.

As I wrote in my original write-up, the right interpretation of this data is not “survey respondents are lying about their ideology” (although they might be, this data doesn’t land either way), it’s that the label moderate does not map cleanly onto what people actually say they want from a political party. A lot of voters probably use “moderate” to mean something like “I’m not an extremist,” or “I don’t like partisan conflict.”

But that is not the same thing as describing a coherent set of centrist party goals. When Blue Rose asked people to say what they actually wanted from politics, most self-identified moderates did not articulate anything that clustered into a recognizably moderate program. Some examples of what people said are listed in my original article.

That difference matters a lot for strategy. If “moderate” is mostly a weak or noisy label for what voters want from a party, then “win moderates” is not really a strategy; it’s a slogan.

Among swing voters — people who switched either from Biden to Trump or Trump to Harris from 2020 to 2024 — the biggest clusters were 25% left, 24% right, 23% anti-system, and 21% affordability. Just 8% were moderate. And among 2024 non-voters, even more were anti-system, and just 5% were labeled moderates.

The voters campaigns most want to persuade or mobilize are not sitting in some tidy centrist lane, waiting for a consultant to shave a few points off a laundry list platform document. They are much more likely to be angry about prices, alienated from institutions, or structured by concerns that do not fit cleanly on a left-right ideological spectrum. Or left-wing!

Together, the anti-system and affordability segments make up roughly 44% of swing voters and 50% of non-voters. The explicitly moderate segment is just 5% to 8%. The most parsimonious interpretation of this data is to spend more time campaigning on conditions and anti-system sentiment and less time on centrism. It is important for a party not to be viewed as extreme, but a lot of people have spent a lot of time and money campaigning on and covering that over the last year, to the detriment of other important factors.

A caveat here: the fact that self-identified moderates do not describe coherent centrist ideology in open-ended text does not mean they will never reward candidates who seem less extreme, more restrained, or less conflict-driven. They might. But the label “moderate” is too noisy to tell campaigns which positions to change or which voters they are actually trying to reach. That is the practical problem. “Win moderates” sounds like a strategy, but without knowing what those voters actually want — and the data says they want very different things — it is closer to a wish.

Subscribe now

2. Why “just be moderate” isn’t a strategy

That brings us to the second big piece of evidence published over the last month: a new working paper by political scientists David Broockman and Joshua Kalla testing how much voters respond to issue positions in a survey experiment.

Many people in the moderation camp have treated this paper as vindication. Matthew Yglesias cited it in a New York Times guest essay. The political strategist Liam Kerr, who helps run the centrist Super PAC “Welcome,” wrote it up on Substack in a sort-of communiqué to aligned funders and readers. And there was a rush on social media to claim that the empirical debate is now settled in favor of moderation.

To set the stage for the rest of this post, I do not think a fair reading of the paper actually justifies what the champions of moderation have been using it to claim. In fact, the paper is more of an indictment of the “just moderate” strategy than a vindication; it shows that moderation only works on a case-by-case basis and that the effect size is smaller than that for the national environment, fundraising, or candidate factors (you know, the types of things I was writing about a year ago) — and it does so with an experimental design that is maximally favorable to the Yglesias wing of the party.

But even if you take the results of the paper at face value, if the entire centrist operation dominant in the Democratic Party donor circles and among media elites today is only finding effects of one point for its core political strategy, that is actually a pretty big loss in two ways: First, because they’re ceding that other campaign tactics have higher efficacy; and second, because this coefficient is between one-half and one-third the size of the previous estimates people like Yglesias and Welcome PAC were pushing.

The average effects are small

Before getting into the gory political science details, some context. Yglesias’s NYT essay — “The Democratic Brand Is Toxic in Too Many States” — used the Broockman-Kalla paper to argue that Democrats should moderate strategically on specific issues where they are out of step with voters. That is a notable shift from the cruder version of the moderation thesis he and others were pushing a year ago, which was basically “Democrats should just be more moderate” as a generic brand strategy. Now, Yglesias argues that shedding certain issue positions would help the Democratic party establish a more “moderate” brand (whatever “brand” means).

That might be true, but it’s not actually what the paper says. One of the authors even said on social media that the effects they found in their experimental study almost certainly would not hold in the real world. Details time:

Read more

Here are the big takeaways:

• The focus groups reveal a level of economic desperation that polls can't fully capture. BlueLabs' detailed interviews surfaced gut-wrenching stories from voters. Some described working two jobs, skipping meals so their kids could eat, and spending hours managing coupon apps just to afford groceries. One participant in a Republican-leaning focus group, a woman in her mid-50s, put it simply: "We shouldn't have to work so hard to survive." This is the type of resolution you can’t get with quantitative polling data.

• Both parties are seen as out of touch, but in different ways. Trump voters in the focus groups called the Republican Party “embarrassing” and “too radical,” while Democrats were labeled “weak” and unable to follow through on promises — even by their own supporters. Corruption was seen as a universal weakness of federal lawmakers, something voters say is particularly bad with Donald Trump but not specific to him. The divide in the focus groups wasn’t left vs. right, it was working class vs. political elites.

• Voters don’t want policy platforms — they want someone who’s actually struggled. As one woman put it: “I would want somebody that’s for us and understands what it’s like to struggle.” And mobilization is a big concern for both parties. In these focus groups, Democrats are still seen as “for the working class” in theory, but convincing disengaged voters that showing up is worth their time remains the party’s biggest challenge.

Subscribe now

If you’re on email or the Substack app, you can read the transcript of our conversation by clicking the headline of this article to take you to the web version of the podcast, then clicking the button just below the byline that looks either like a piece of paper or is labeled “Transcript,” like so:

You can also subscribe to the podcast on your favorite podcast app to listen on your own time. And if you do listen via one of those apps, please drop us a five-star rating and review if you feel we’ve earned it — it really helps people discover the show!

Subscribe now

Leading off: Very liberal Americans, who have rated the Democratic Party poorly relative to other partisans since 2024, have swung sharply back toward congressional Democrats over the last few months. A new poll also finds voters who dislike both parties now prefer Democrats by 31 points. These gains should reassure a party that has faced internal strife since Trump’s second term began, but look less due to renewed faith in the Democrats and more like anti-Trump consolidation. That might not matter for the midterms — a vote won is a vote won — but it will matter for 2028 and beyond.

On deck this week: Tuesday’s Deep Dive will cover some new research on the level of ideological thinking in the electorate and the value (or not) of ideological moderation by the Democrats, and Friday’s Chart of the Week will respond to whatever’s in the news. I’m also finalizing questions for our April Strength In Numbers/Verasight poll this week — so subscribers, send in your question recommendations if you haven’t already! (Email or comments are fine.)

On with the data.

Subscribe now

1. Anti-Trump sentiment, not pro-Democratic enthusiasm, is uniting Democrats again

A new YouGov/The Economist poll, fielded from March 27 - 30, finds that Democratic voters have grown significantly warmer toward their members of Congress over the last few months. Earlier in 2026, Democrats said their party’s MOCs were favorable at a rate just 30 points higher than the rate they said their party was unfavorable. That gap has now grown to +55 — rivaling the favorability of Republican MOCs among Republican voters.

Aggregate Democratic views have increased because very liberal Americans have become sharply more favorable toward congressional Democrats since January. This group evaluated the party’s members of Congress favorably by a net +28 points margin — up from a -13 deficit in January. That’s a 41-point shift in two months:

Among Americans who are liberal but not very liberal, moderate, or conservative (basically everyone else), views of congressional Democrats barely budged.

Overall, U.S. adults give the Democrats a favorability rating of -21, 5 points higher than the rating they currently give Republicans.

That is a meaningful change. Last summer, Strength In Numbers documented that Democratic party favorability was unusually weak even as the party remained competitive on the generic ballot. We dug into the survey microdata and found out that this was because many left-leaning Americans were frustrated with their own side after the 2024 loss. Charles Franklin, who conducts the Marquette University Law School Poll, has been tracking the same dynamic in both national and Wisconsin polling. Among Democratic identifiers in Wisconsin, his data shows a net +56 favorability rating for their party, compared to +74 among Republican identifiers for the GOP.

Franklin finds that while Democrats still disagree about what they’re for, they are virtually unanimous in what they’re against: Donald Trump.

The simplest explanation for Democrats’ gains is that politically active party members on the left — who have had a lot of complaints about how the party is handling Trump 2.0 — are now responding to the same thing many other Americans are right now. That is, the president has moved public policy on many issue domains far to the right and up on the authoritarian axis (certainly far past the policy temperature “set point,” to use the language of the thermostatic model), and progressives are setting their differences with the Democrats aside for the moment as they focus on defeating an increasingly unpopular Republican president. This looks more like anti-Trump unity than pro-Democratic enthusiasm.

But it’s not just the base

The Democrats’ consolidation of left-wing liberalism is one piece of a broader backlash to Trumpism that shows up in the polling data right now. Another notable finding this week is from a new CNN/SSRS survey that found that about one-quarter of the public holds an unfavorable view of both parties. These are the so-called “double haters.” This group prefers Democrats on the 2025 generic ballot by 31 points.

This is a big deal for two reasons. First, that’s a massive shift; Double haters broke for Trump in 2016 and again in 2024. Now they’re swinging hard the other way.

Like Franklin’s polling, the CNN report also finds that Democrats’ gains are driven largely by opposition to the GOP, not enthusiasm for Democrats themselves. When asked what they dislike about Democrats, 22% of double haters called the party “do-nothing” and 11% said they aren’t standing up enough to Trump and the GOP, while 10% said they’re too liberal.

Will 2026 be a Democratic fake out?

So we’ve got two layers of anti-Trump consolidation happening at once. YouGov’s data shows the Democratic left is coming home, and the CNN poll shows voters who dislike both parties — a swing group that has been decisive in recent elections — are breaking heavily toward Democrats for the first time in years. Neither group is necessarily enthusiastic about Democrats. But both are currently heavily voting against Republicans. According to the CNN poll, 79% of voters who plan to support Democrats say their vote is a message of opposition to Trump. (Only 46% of Republican voters say they’ll vote to show support for the president.)

This could make for a big electoral win for Democrats in November, despite the division in the party and its overall nominally unpopular rating. According to CNN, Democratic-aligned voters are 17 points more likely than Republicans to call themselves “extremely motivated” to vote in 2026 — even though they’re 14 points less likely to view their own party favorably. Meanwhile, the Democrats have opened up a large lead in the U.S. House generic congressional ballot for 2026. They are up +6 in both the CNN and YouGov surveys, and closer to +5 on average.

This is the pattern I’d expect in a midterm environment that favors the out-party. But with many Americans (including the vaunted “double-haters”) still viewing the Democrats as weak and ineffectual, a big electoral victory will not completely solve their deeper problems of identity and division.

The trend in this data is good for the Democrats, in other words — but don’t misread a positive trend for a positive level.

Share

2. What Strength In Numbers published last week

Readers of Strength In Numbers got three articles last week — a lighter load, since I was out sick Monday and Tuesday.

This week’s Deep Dive asked a question I’ve been getting a lot lately: if Trump is 20+ points underwater, why aren’t Democrats leading the generic ballot by 20?

On Thursday, David and I recorded our weekly podcast about Trump’s record-low polling numbers on Iran and the economy:

And Friday’s Chart of the Week broke down the six big events that have dragged down Trump’s approval rating since he took office:

If you’re a frequent reader of Strength In Numbers, I’m confident you will get a lot of value out of a paid subscription. Paid subscribers get access to Tuesday’s Deep Dives — the long-form data analysis that really digs into the numbers — plus everything else I publish.

Subscribe now

3. Even more numbers!

• The party ID gap is at its widest in over a decade, per Gallup — Democrats and leaners at 49%, Republicans at 39%, the lowest Republican share since 2015. Fewer Americans calling themselves Republicans or Republican-leaning independents since 2015: Gallup data

• The CNN double haters poll: A new CNN poll reveals how people mad at both parties see the midterms

• The YouGov newsletter I drew from above has a lot more, including a great experiment on how the labels “hawk” and “dove” change how people describe their foreign policy views. Democrats are starting to like congressional Democrats again

• About six in ten Americans disapprove of Trump’s handling of the Iran conflict, and a similar share say striking Iran was the wrong decision, per a March Pew survey. Republican-leaning independents are genuinely divided on both questions. Americans Broadly Disapprove of U.S. Military Action in Iran

• A late-March CNN/SSRS poll found that just one-third of Americans think Trump has a clear plan for Iran — and only 29% say the war has been worth the toll, far below the 59% who said the same about Iraq in 2003. Donald Trump faces deep public skepticism about Iran war ahead of White House speech, CNN poll finds

• Kyle Saunders writes on the twin crises facing survey research — declining trust in institutions and declining confidence in measurement, both hitting at the same time. The Pincer: Bots, Trust, and a Quiet Crisis in Public Opinion Research

• The Guardian reports on how fraudulent church attendance data is a warning sign for the entire polling industry. ʼOur assumptions are brokenʼ: how fraudulent church data revealed AI’s threat to polling

  • A lot of this seems overhyped to me, and it’s worth noting that the leading authors of some new papers here do not provide estimates of the potential error or amount of fraud that may be impacting popular opt-in online panels right now. But “bogus respondents,” as Pew calls them, have been a problem for a long time; that’s why it’s important that if you’re doing polling online right now, you verify that you’re interviewing a real human person. (The Strength In Numbers poll partners with a mixed-mode online survey provider that does this.)

• A few people sent me this very booster-ish WaPo op-ed arguing that prediction markets are better than polls. Your mileage may vary, but the guy writing seems to be involved with some markets, and in the lead picture compares market odds with raw poll vote shares in calculating prediction error (big red flag).

• I did a webinar for the global party foundation associated with Germany’s SPD this week, discussing the outlook for the 2026 midterms. You can watch the replay for free here: Webseminar: Outlook on the 2026 US Midterm Elections

And that’s it for this week! Thanks for reading. Strength In Numbers will be back in your inbox on Tuesday!

Got more for next week? Email your links or add to the comments below!

Leave a comment

Related Articles

Donald Trump’s approval rating hit a new low this week. As of April 2, the FiftyPlusOne average puts his net approval at -21.4 — 37.2% approve, 58.6% disapprove. That’s the lowest mark of his second term.

How bad is -21.4? When compared to past presidents, Trump’s ratings are the lowest of any president at this point in their term, going back to FDR. The spaghetti chart below lines up every president’s net approval by time in office. Trump’s red line is all the way at the bottom. Fourteen months in, no one else was this far underwater. Only Joe Biden, at the beginning of the 2022-23 inflation crisis, was close — and Trump is outrunning him by about 10 points:

If you treat Trump as a second-term president, his rating today is higher only than Richard Nixon’s after Watergate and tied with confidence in George W Bush after Katrina a worsening war in Iraq:

What’s striking about Trump is that he’s reached basically the same place without one singular catastrophe doing all the damage. Instead, he has accumulated several negative shocks to his presidency, including fallout from his tariffs, mass deportations, the Oct. 2025 government shutdown, and an unpopular war. None of them alone ended his presidency, but together, they’ve dragged him down to the same territory that ended other presidents’.

Looking closely at Trump’s approval rating over the last 14 months, six events stand out as primary inflection points for the public’s view of his presidency. One issue seems to be the dominant factor in his decline.

This week’s Chart of the Week takes a statistical approach to identifying “change points” in Trump’s approval trend, and triangulates one overriding cause of his bad numbers with fresh polling on voters’ economic anxieties.

Subscribe now

Six events that have dragged down Trump’s approval

Trump’s approval rating hasn’t fallen in one clean, steady line. The trend is a mix of a secular negative slope and political shocks that cause brief stages of accelerated decline.

To examine what events have caused these shocks, I put Trump’s approval rating in the 50+1 average into a statistical model called a “Bayesian changepoints detector.” This is a mathematical program that looks for events in a time series that, statistically, cause something about the trend to change. Using this model, I can break Trump’s approval trend since Inauguration Day into seven distinct segments.

The first drop was by far the sharpest. In the couple of weeks after Trump took office and started signing a barrage of executive orders, his net approval cratered at a pace of about 15 points per month. That was the fastest decline of his whole presidency — basically, whatever political “honeymoon” Trump enjoyed after his 2024 victory collapsed almost immediately.

From early February through early April 2025, that decline slowed to around 2 points per month. Then came his “Liberation Day” tariffs announcement, a year ago as of this Thursday, April 2, and the press coverage around the Kilmar Abrego Garcia deportation. That caused Trump’s approval to drop another 5 points in just a month.

After that, things eased off a bit more. The National Guard deployment to Los Angeles in June 2025 coincided with a slower rate of decline, closer to 2 points per month. And from June through October, Trump’s numbers were almost flat, falling by less than a point a month. If there was ever a moment when things looked like they might be leveling off, that was it.

But it didn’t last. The government shutdown and the No Kings protests in October 2025 pushed the trend downward again, roughly doubling the pace of decline. The president lost support after the 2025 statewide elections across the country, then recovered a couple of points of ground.

The trend snapped back to the negative with the first round of the Epstein files release in December. And now the war with Iran has sped things up again, with Trump’s net approval falling at about 2 points per month again — the fastest sustained slide since spring 2025.

Surprisingly, one set of events the model doesn’t pick up is the murder of Nicole Good and Alex Pretti in Minneapolis in January. There looks to be a slight decline in Trump’s rating after these events, but the model does not pick it up as a “change point” — maybe because of the noise in the trend in early Feb. This is good point to stop and remind ourselves that analyzing aggregate trends in observational polling data often leaves information on the table and cannot provide for causal analysis. For example, there was a big dip in Trump’s immigration issue approval after Good and Pretti were killed, suggesting it did break through to the average American when they thought about immigration in particular, even if there wasn’t an obvious hit to Trump’s overall rating.

Looking at the events that stand out for their impact on public opinion, I have two takeaways from this analysis.

Share

This post was sponsored by Incogni.

Every day, private companies called “data brokers” search the internet for your private data—phone number, DOB, SSN—and sell it to the highest bidder. This information can be used to target you with ads, scam you, or worse. I’m a pretty paranoid person, so I have personally been using Incogni for years to scrub my personal data from the web. Incogni helps remove sensitive information from corporate data brokers as well as shady people search websites.

You can try Incogni today and get 58% off annual plans with code GELLIOTT58.

Is economic anxiety a new albatross for all incumbents?

One common thread running through several of the events that hurt Trump most is economic anxiety. The “Liberation Day” tariffs, government shutdown over health care costs, and now the war in Iran — directly linked to a verging-on 40% increase in gas prices since Feb. 28 — all had consequences for the economy or Americans’ spending power that were negative and highly visible.

The public’s view of Trump’s handling of prices and the cost of living also hit a new low this week of -37.6 net:

And the broader mood on the economy looks awful for him. A new University of Massachusetts Amherst poll out this morning finds that 53% of Americans say Trump is “most responsible for the rise in the price of goods and services” lately.

Apart from this, the only events that seem to move Trump’s approval are high-profile occasions of executive overreach, such as on immigration and interfering in the Epstein investigation. The Abrego Garcia case and LA protests both look to have hurt him badly. Few or no other events have interrupted the trend the way events that stoke voters’ economic anxieties have.

The more data we get, the more it looks like voters are just in the mood to punish the party in power for high prices and impaired economic mobility. Trump won in 2024 because of this, but now that his party is in power, he’s on the wrong end of the stick.

Share

It’s hard to win back public support when your agenda is this unpopular

Finally, what really stands out about Trump’s trajectory is not just how low his approval is. It’s that it basically never bounces, either. Over 14 months, there hasn’t been a single sustained recovery in his overall approval rating. He got no rally from the Iran conflict, despite what history would suggest about the intervention, and no meaningful bounce from any of his legislative wins. Even the apparent “plateau” over the summer wasn’t really a rebound; it was just a slower decline.

Most presidents get at least some back-and-forth in their numbers. They lose support, then win some of it back after a successful policy fight or a crisis that pulls the country together. Trump hasn’t had that. His line only really moves in one direction.

This is to be expected from a president who has pushed the most unpopular policy agenda of all time. And if that pattern holds — and right now there’s not much evidence that it won’t — Trump will head into the 2026 midterms as the most unpopular president in modern polling history.

I wrote for my Deep Dive this week that Democrats aren’t doing as well as you’d expect based on Trump’s bad position. I would encourage you to read that investigation into the polling data if you want to understand why that is. But one important point is that Democrats don’t need to win the generic ballot by 20 points for Republicans to see the consequences of Trump’s most-unpopular-ever presidency. We’re already seeing them right now.

To sum up: Trump’s approval ratings slide doesn’t look like the result of one isolated crisis so much as the cumulative effect of a presidency that keeps making voters feel not listened to and less economically secure. Again and again, the biggest drops in his standing follow events that raised costs, heightened uncertainty, or reinforced a sense that the country is off track. And because the president has shown almost no ability — or interest — in winning back support, his unpopularity looks less like a purely political or economic liability and more like a defining feature of his second term.

If you’re a frequent reader of Strength In Numbers, I’m confident you will get a lot of value out of a paid subscription. You’ll get access to all of my analysis, including weekly Tuesday Deep Dives, and you’ll be supporting independent data journalism and original data collection.

Become a paying member today

Related Articles

Here are the big takeaways:

• Trump’s war in Iran is deeply unpopular — and getting more so. Strength In Numbers’ polling shows 72% of Americans oppose sending U.S. ground troops to Iran, including a majority of Republicans. On top of that, Trump’s approval rating on his handling of prices and inflation has cratered to a new low of minus 33, a slide that tracks almost perfectly with a 35% spike in gas prices since the war began. According to YouGov, Trump is now doing worse on prices than Biden was at the equivalent point in his presidency — a remarkably bad position for a president who campaigned on lowering the cost of groceries on day one.

• Trump gets record-low numbers on trade on the anniversary of his “Liberation Day” tariff announcement. One year after Trump’s sweeping tariff announcement, his approval on trade and tariffs has matched an all-time low at minus 20. The promised benefits — manufacturing job growth, lower income taxes, a revitalized industrial economy — never materialized. Instead, the U.S. has lost roughly 100,000 manufacturing jobs over the past year, consumers are seeing excess inflation in goods, and grocery prices have risen, due in part to Trump’s mass deportation of agricultural workers. When asked about a direct trade-off in polling, Americans chose lower prices over more manufacturing jobs by a 54-33 margin.

• Trump took a beating at the Supreme Court... The birthright citizenship oral arguments went badly for Trump’s solicitor general on Wednesday. Legal commentators are predicting a lopsided loss for the administration, and polling shows wide majorities of Americans in favor of birthright citizenship — which wasn’t the case just a couple of decades ago.

• …and in the shutdown fight. On the shutdown, Democrats emerged with a clear win: Republicans caved and announced a deal to pass virtually the same TSA-funding bill many had angrily rejected a week earlier, with no additional money for ICE or Customs and Border Patrol. Elliott’s flash Strength In Numbers/Verasight poll found 52% of Americans blame the Trump administration and congressional Republicans for airport chaos, versus just 25% who blamed Democrats.

If you missed our video livestream, you can watch it by clicking play on the web version of this post at gelliottmorris.com. We record the podcast live every Thursday at 2:00 PM Eastern. We always discuss a few pre-planned topics and then answer questions submitted live by viewers.

You can also subscribe to us on your favorite podcast app to listen on your own time. And if you do listen via one of those apps, please drop us a five-star rating and review if you feel we’ve earned it — it really helps people discover the show!

A reminder that paid subscribers to Strength In Numbers get to participate in our live Q&A!

Share

You can also read the transcript of our conversation by clicking the headline of this article to take you to the web version of the podcast, then clicking the button just below the byline that looks either like a piece of paper or is labeled “Transcript,” like so:

If you’re a reader of Strength in Numbers and haven’t yet subscribed to David’s newsletter, head to the-downballot.com.

Subscribe to The Downballot

And if you’re coming from David’s site, subscribe here to get the numbers behind the news.

Subscribe to Strength In Numbers

Yet Trump’s net approval rating (among all adults) in the same survey we published is a whopping -23 points. Just 37% of American adults approve of the job he is doing as president, while 60% disapprove.

If Trump is 23 points underwater, why are Democrats only ahead by 6?

Part of the answer is straightforward, though it accounts for a small part of the difference. The big issue here is that the two numbers measure attitudes among different populations: The generic ballot margin I report is among registered voters, and the one that eventually exists in reality will be among voters. The approval rating number we report out, however, is among all adults. Polling data suggests these groups have different political leanings.

Among registered voters specifically, Trump’s net approval is -21 — still terrible, but two points less terrible than the all-adults number. And among all adults, the Democratic generic ballot margin is D+7.5, not D+6 — about 2 points in the opposite direction. So some of the 15-point gap among RVs is just a population mismatch between the two metrics people tend to compare.

But even after you compare apples to apples, there’s still a 15-point gap between Trump’s net disapproval among registered voters (-21) and the Democratic ballot margin among registered voters (+6). Over the last 6 months, the generic ballot margin is running at roughly 40% of net disapproval.

If we take a deep dive into our polling microdata and historical exit polling, what can we learn about this gap? How big is reasonable? And what can Democrats do to bring their generic ballot number closer to Trump’s approval rating?

Subscribe now

It’s all about the “cross-pressured” voters

Among registered voters in our March poll, 50% strongly disapprove of Trump, and another 9% somewhat disapprove. These groups say they will vote Democratic in the House generic ballot this November at much different rates.

Read more

Actually, according to 50+1, Trump is 21 points underwater as of this morning — a new record low. A poll from UMass Amherst out Monday had just 33% of adults approving of his job as president.

Back in your inboxes tomorrow with what I think is a comprehensive answer to a very important question, drawing on our exclusive Strength In Numbers/Verasight polling partnership and external data.

Elliott

Here’s a quick bonus podcast mini-episode for paying subscribers answering two questions that didn’t make it into the main podcast this week.

1. On the generic ballot gap

Why are Democrats only up six to eight points on the generic ballot when Trump’s net approval is around -20? Two things are going on.

Read more