How Americans Rate Their Government: What the Polls Actually Show, 2005 to 2026

Two Numbers That Shouldn't Both Be True

Congress's approval rating sat at 10% in April 2026. Its disapproval rating was 86%, tying the worst ever recorded ^[4]. And yet most Americans still like their own representative. Roughly 40% to 50% approve of the specific member who represents them, and a slim majority say that person deserves to be re-elected ^[6]. Incumbents keep winning by wide margins, even in a Congress voters say they despise ^[20][21].

Both of those things are true at the same time. That's the puzzle sitting underneath every headline about American distrust of government. It suggests the story isn't simple collapse. It's something stranger — people can hate an institution and still like the person who sits inside it.

Part of the confusion starts with the words themselves. "Party approval," "party favorability," "party identification," and the "generic ballot" get used almost interchangeably in coverage, but they measure different things ^[1][2]. Favorability asks whether you like a party. Identification asks what you call yourself. The generic ballot asks who you'd vote for. Job approval asks whether Congress or the president is doing its work well. A headline that treats these as one number is already telling you less than it seems to.

Both Parties, Underwater, At Once

Right now, neither party is popular. Gallup measured Democratic favorability at 34% in 2025 — the lowest reading since that poll started tracking the party in 1992. Republicans stood at 40%, itself down from 43% the year before ^[1]. Pew found something similar in its own April 2026 survey: 59% of Americans viewed the Democratic Party unfavorably, and Republicans were nearly as underwater, at 58% ^[3].

About one in four Americans now say they dislike both parties. That figure has grown, from 21% in 2020 to 26% in 2026 ^[3]. It's a number that can support two very different stories — one where a single party is collapsing, and one where the whole party system is losing ground — and those competing readings get weighed later on.

A party's rating depends heavily on its own side. In Pew's 2026 numbers, 89% of Republicans liked their own party, and 84% of Democrats liked theirs. But independents who merely lean toward a party were far less enthusiastic — only about half of them approved ^[3]. Democratic favorability partly slid because Democrats themselves soured on their own party: CNN found just 63% of Democrats viewed their party favorably in March 2025, down from 81% at the start of Biden's presidency ^[9].

Meanwhile, a record share of Americans call themselves independents — 45% in 2025, the highest Gallup has ever measured ^[10]. That sounds like a mass exit from both parties. It mostly isn't. Most independents reliably lean toward one party and vote like partisans, so the actual two-party electorate hasn't shrunk nearly as much as the label suggests ^[11].

The Collapse That Never Quite Recovered

Congress's approval numbers tell a specific shape, and it isn't a straight line down. Gallup has tracked this since 1974, and the average across that whole run is about 32% ^[2]. In the mid-2000s, approval sat in the 20s and 30s — already low, but not catastrophic.

Then it fell hard. After a government shutdown, Congress approval hit a record 9% in November 2013, and the whole year averaged just 14%, the worst on record ^[5]. It has mostly stayed in the teens ever since. That's the real shape: not a slope, but a cliff followed by a plateau.

It still moves with events. When the new administration took office in February 2025, Congress approval jumped 12 points to 29%, its highest mark since 2021, as Republican approval of Congress rose to 53% ^[24]. By October 2025, during another government shutdown, it fell back to 15%, as Republican approval dropped 21 points ^[12]. By April 2026, it was back down to 10% ^[4]. None of that looks like a smooth decline. It looks like an institution whose approval swings sharply with who's in charge and what's going wrong.

Where the Steady-Decline Story Breaks

The presidency is where the tidy "everything has been falling for twenty years" idea runs into trouble. George W. Bush's approval didn't decline steadily — it spiked to nearly 90% after the September 11 attacks, then fell to 25% by 2008 ^[7]. That's an enormous swing, more than 60 points in one term.

Modern presidents don't move like that anymore. Biden's approval averaged 42.2% across his term, the second-lowest of any post-World War II president, but it never swung wildly in either direction ^[13]. Trump's approval in mid-2026 sat near 40%, according to aggregators tracking multiple pollsters ^[14][15]. The range has narrowed. Presidents no longer reach the highs Bush once did, but they're also cushioned from the deep lows.

Here's the mechanism behind that squeeze, and it matters for reading every approval number that follows: the gap between how a president's own party rates him and how the opposing party rates him has been growing for decades. The American Presidency Project puts that gap at 38.8 points under Eisenhower in the 1950s. It grew to 60.7 points under George W. Bush, 71.0 under Obama, and 83.4 under Trump's second term as of May 2026 ^[8]. One Gallup reading from August 2025 showed a 93% approval rating among Republicans and just 1% among Democrats — a 92-point gap that tied the largest Gallup had ever recorded ^[16].

What that means in practice: a president's approval number today is driven far more by which party you belong to than by how you think he's doing that particular week. Republicans approve of a Republican president almost automatically. Democrats disapprove almost automatically. The number still exists, but it's measuring identity more than performance — which is exactly why comparing a 2026 approval rating to a 1975 one, without accounting for this shift, can mislead.

The 2026 electoral picture adds another wrinkle. Generic congressional ballot polling, which asks people which party's candidate they'd support, showed Democrats ahead by roughly 7 points in July 2026 — around 48% to 41%, up from a near-tie in January 2025 ^[15][17]. That's a different measure entirely from approval. A party can be broadly disliked and still be winning the head-to-head matchup, because the ballot asks people to choose between two options, not to grade either one.

Four Ways to Read the Same Numbers

Given all that, four different camps have looked at essentially the same data and drawn different lessons from it — and each has a real case.

One camp, including Pew's long-running trust research, calls this a genuine legitimacy crisis. Trust in the federal government to "do the right thing" sits near 17%, one of the lowest points in nearly 70 years of Pew's polling, and the mistrust reaches beyond Congress into the media and science ^[18]. In this reading, the low numbers aren't noise — they're the visible symptom of Americans losing faith that government can deliver at all. The gap in this argument is timing: trust jumped sharply for Republicans the moment their party took the White House in 2025 ^[18]. If it were really about lost faith in the system, it probably wouldn't swing that fast with who's in charge.

A second camp, made up mostly of political scientists studying negative partisanship, calls this a polarization artifact. Their argument: people rate the other party and Congress-in-the-abstract harshly, while still liking their own party's leaders and their own member of Congress. The growing partisan approval gap — from under 40 points in the 1950s to over 80 now — backs this up ^[8]. So does the fact that incumbents keep winning even as Congress polls in the teens ^[6][20]. The weak spot here is that it can't fully explain why a party's own base sometimes turns on it, the way Democrats did on their own party in 2025 ^[9], or why the share disliking both parties keeps climbing ^[3].

A third camp points to measurement and the changing media environment. Phone response rates have collapsed. Most polling shifted from live phone calls to online panels. The number of polls being published has exploded, and different pollsters now produce noticeably different results for the same question ^[14][15]. In this view, some of what looks like decline is really just noisier, less comparable data. The problem with leaning too hard on this explanation: the biggest swings — Congress hitting 9% in 2013, both parties being underwater at once — show up consistently across Gallup, Pew, CNN, and other outlets using different methods ^[4][3][9]. When that many different measuring sticks agree, methodology alone doesn't explain it away.

A fourth camp takes the long view: Congress has almost always been unpopular, and this is nothing new. Its approval has averaged only about 32% since 1974, and it spiked after events like 9/11 before falling right back down ^[2]. Political scientist Richard Fenno described members of Congress who "run for Congress by running against Congress" back in the 1970s — the my-member-is-fine, Congress-is-terrible split is decades old ^[6][20]. What this view can miss is that the shape of the RANGE has actually changed. The narrowing of presidential approval and the 80-plus-point partisan gap have no real precedent in the earlier data ^[8].

Underneath all four readings sit the same structural forces. Negative partisanship pushes people to rate the opposing party near zero regardless of performance ^[8][11]. Most House districts are safe for one party, so the real contest happens in primaries, which reward pleasing the base rather than the middle — that's part of why individual members stay popular even as the institution collapses ^[20][21]. Divided government gives each side a target to blame for gridlock, which drags down Congress's rating while leaving people's own member untouched ^[4][6]. Voters increasingly treat local House races as a referendum on the president, which ties the generic ballot closely to national mood ^[17][11]. A fragmented, partisan media landscape reinforces all of it by letting people live in separate information worlds ^[18]. And distrust isn't just a political story — the OECD finds American trust in national government below the average across 30 countries, part of a broader pattern across democracies rather than something uniquely American ^[22].

That last point shows up in how outlets frame the story, too. Gallup and Pew report the numbers with careful symmetry between the parties ^[1][3]. The American Presidency Project and data journalists like Nate Silver treat this as a structural story about polarization, not a verdict on any single leader ^[8][14]. Center-left outlets like CNN and further-left outlets like Rolling Stone have leaned harder into the Democratic Party's specific collapse as a leadership failure ^[9]. Fox News has emphasized Democratic weakness while giving less attention to Republican favorability also being underwater ^[1]. None of these framings changes the underlying numbers — they just choose what to put in the headline.

What Would Actually Settle This

Some of this is still genuinely unresolved, and it's worth being honest about what a single poll can't tell you. House effects — the tendency of different pollsters to land on different numbers for the same question — are large enough that a one-point move in a single poll is often just noise ^[14][15]. The shift from phone polling to online panels over the past twenty years makes clean comparisons between 2006 and 2026 shakier than they look ^[14].

There's also a data gap that's easy to miss: most long-running approval series, including Gallup's, ask about "Congress" as one institution. A clean, consistently polled split between House and Senate approval doesn't really exist ^[2]. And some of the most-cited numbers here are only weeks old — the July 2026 generic ballot figures and the April 2026 Congress reading are single snapshots that will likely move before the midterms ^[17][4].

The biggest open question may be the simplest one: is 2025's record-low party favorability a floor, or just a stop along the way? Democratic favorability had already ticked up somewhat from its August 2025 low by early 2026 ^[3][23]. Whether that's the start of a recovery or a temporary bounce isn't something the current data can answer. It's also the one thing that would tell us most about which of the four readings above is closest to right — and nobody knows it yet.