How to Read a Presidential Approval Poll

What the question measures

A presidential approval poll asks one deceptively plain thing: do you approve or disapprove of the way the president is handling his job. It does not ask whether you like him, whether you voted for him, or whether you will next time. It asks about the job, right now, and the answer moves with the news.

That is the first thing to hold onto. Approval is a mood reading, not a scorecard. It rises on a good jobs report and a foreign policy win and falls on a war, a scandal, or a price spike. It tells you how a month felt to the country. It does not tell you what a presidency will have accomplished when the record is closed, and the two are frequently far apart.

The number also has a floor and a ceiling built from party. In a polarized era, a president's own party rarely dips far below the high 70s in approval of him and the opposition rarely climbs far above the low teens. The action, the part that actually moves, is in the middle: independents and the softer partisans who respond to events. When an average drops several points in a couple of weeks, that middle is what shifted.

One poll is noise; the average is signal

Every survey carries a margin of error, usually around 3 to 4 points on a national sample of a thousand or so adults. A poll reporting 37 percent approval with a 3.5 point margin is really telling you the true figure is probably somewhere between 33.5 and 40.5. Two honest polls taken the same week can differ by 6 or 7 points and both be correct within their stated error.

The fix is not to pick the poll you like. It is to read the average of many. When several pollsters using different methods converge, the shared signal is far more trustworthy than any one reading. In late July 2026, for example, eight national polls of President Trump ranged from 32 percent approval at Quinnipiac to 40 percent at CNBC, a spread of eight points (Quinnipiac University, July 29, 2026; CNBC, July 17, 2026). The Silver Bulletin average that cut through the range sat at 38.2 percent approval and 58.8 percent disapproval (Silver Bulletin, July 30, 2026). The single polls argue. The average decides.

House effects and how to spot them

Pollsters make defensible choices that push their results a point or two in a consistent direction. That lean is called a house effect, and it is not the same as bias in the loaded sense. It comes from method: whether the sample is registered voters or all adults, whether it is a probability panel or online opt-in, how the pollster weights for education and past vote, and exactly how the question is worded.

Registered-voter samples often run a touch more favorable to a Republican president than all-adult samples. Probability-based panels, where respondents are recruited by random selection, are the gold standard and tend to be steadier. The practical rule for a reader is simple: never compare one pollster's new number to a different pollster's old number and call the difference a trend. Compare like to like, or compare each poll to the average. A three-point drop that is really just the switch from a friendly house to a tougher one is a mirage, and mirages win a lot of arguments online.

Strong approval, soft approval, and intensity

The topline hides the temperature underneath it. Most quality polls also report how strongly people feel, and that split often matters more than the headline. A president at 34 percent overall with only 15 percent strongly approving and 50 percent strongly disapproving is in a different position than one at 34 percent with intensity balanced (CNN, SSRS, July 29, 2026). Strong disapproval predicts turnout against, donations against, and primary challenges. When the intense end of the scale tilts hard one way, the headline number understates the political weight of it.

Read the crosstabs when you can. A shift driven by one issue, say a war, will show up as movement among the groups closest to that issue while other groups hold. That is how you tell a broad collapse from a narrow, event-driven dip that may reverse when the event fades.

What history says a low number is worth

Here is the context that keeps a bad month in proportion. Presidential approval is volatile, and the modern record is full of deep troughs that were later climbed out of, and high peaks that melted. Harry Truman fell to 22 percent in February 1952, the lowest Gallup has ever recorded (Gallup). Richard Nixon sat at 24 percent the month he resigned in 1974 (Gallup). George W. Bush reached 25 percent in October 2008 amid the financial crisis (Gallup). Jimmy Carter fell to 28 percent in the summer of 1979 and lost reelection the following year (Gallup). Ronald Reagan dropped to 35 percent in January 1983, in the teeth of a recession, then recovered to win 49 states the next year (Gallup).

Recent averages sit lower than the mid-century norm on both sides. Joe Biden's term averaged 42.2 percent in Gallup, the second-lowest full-term average on record, and his final reading was 40 percent (Gallup). Donald Trump's first-term average of roughly 41 percent was the lowest full-term average Gallup has recorded, and he was elected again anyway (Gallup). The lesson is not that numbers do not matter. It is that a single low reading is a snapshot, and snapshots of a volatile series predict the next snapshot poorly.

Why the rankings use record, not popularity

All of this is why US Political Rank does not grade presidents on approval. Popularity measures the mood of the governed in a given month. The record measures what actually happened across a full term, and the two answer different questions. When the site ranks presidents by economic record, it uses full-term averages of real GDP growth and inflation from federal statistical agencies, numbers that do not care whether a president was cheered or booed while he governed.

Use approval polls for what they are good at: a real-time read on how the country feels, best taken as an average, weighed for method, and checked for intensity. Use them with a memory of how far past presidents fell and how often they came back. And when a pollster's number surprises you, remember that pollsters themselves keep score against each other, with track records that reward the accurate and expose the rest (see pollster accuracy rankings). A poll is a tool. The reader who knows how it is built is the one it serves.