A presidential approval rating is the percentage of people in a poll who say they approve of how the president is handling the job, after the sample has been weighted to match the population. Each pollster produces its own number from a sample of roughly 1,000 people. The approval figure you usually see in the news is an average of many of those polls, and different averages use different rules, which is why two sites can show slightly different numbers on the same day.
Here’s how each layer works, from the question to the average.
What question do pollsters ask? #
Most use a version of the question Gallup began asking in the 1930s:
“Do you approve or disapprove of the way [name] is handling his job as president?”
Some pollsters offer four choices (strongly approve, somewhat approve, somewhat disapprove, strongly disapprove), and the “strongly” and “somewhat” answers are added together for the headline. Anyone who doesn’t choose is counted as unsure. So the basic math is simple:
- Approval = share who approve
- Disapproval = share who disapprove
- Net approval = approval minus disapproval
Because approve and disapprove don’t always add up to 100%, net approval is often the better single number. A president at 44% approve and 50% disapprove has a net of −6.
Gallup, whose monthly rating was the best-known series for decades, announced in 2026 that it would stop publishing presidential approval ratings. Many other pollsters still ask the question every week or month.
Who gets surveyed? #
A pollster tries to reach a sample that stands in for a larger population. Approval polls usually survey one of three groups, and the choice can move the number a few points:
| Population | Who’s included | Common use |
|---|---|---|
| Adults (A) | Everyone 18 and older | Standard for approval between elections |
| Registered voters (RV) | Adults registered to vote | Common in election years |
| Likely voters (LV) | Registered voters judged likely to vote | Close to an election |
Pollsters reach people by phone, text, mail invitations and online panels, often in combination. A sample of about 1,000 random respondents gives a margin of error of roughly ±3 percentage points at 95% confidence, which is why most national polls land around that size. Our explainer on registered vs. likely voters covers why populations differ.
How does weighting change the number? #
Raw samples are never a perfect miniature of the country. Older, more educated and more politically engaged people tend to answer surveys more often. So pollsters weight the results: each respondent’s answer counts a little more or a little less until the sample matches known population totals for age, sex, race, education and region, and often party or past vote.
A simplified example: if 36% of adults have a bachelor’s degree but 50% of a poll’s respondents do, each college graduate’s answer is weighted down (0.72) and each non-graduate’s is weighted up (1.28). If graduates and non-graduates feel differently about the president, this can move the topline by a couple of points. That’s one reason polls taken the same week can disagree. We walk through the full method in how pollsters weight survey data.
How do approval averages work? #
Any single poll can be off by a few points from sampling noise alone, and each pollster has its own consistent lean, called a house effect. Averages smooth that out by combining many polls. The differences between averages come down to a few choices:
- Time window. A 7-day average reacts quickly to real shifts; a 30-day average is steadier but slower.
- Which polls count. Some averages include every published poll; others drop pollsters that don’t disclose their methods.
- Weighting. Some treat every poll equally. Others give more weight to newer polls, larger samples or pollsters with better track records.
- Adjustments. Some correct for each pollster’s house effect or convert all polls to the same population before averaging.
- Repeat polls. A pollster that publishes daily can dominate a simple average unless its polls are limited.
With FiveThirtyEight shut down by ABC News in March 2025, the approval averages people cite most now include Silver Bulletin, The New York Times, RealClearPolling, Decision Desk HQ and VoteHub, each with its own method. None is the “official” number. What matters is picking one and watching its trend, since switching between averages mixes different methods. Our guide on how polling averages work goes deeper.
How Election Tracker calculates approval #
Election Tracker shows approve, disapprove and the net number as a 7-day average of every published poll, drawn from VoteHub with the New York Times average as a fallback source. A week is long enough that one pollster’s unusual result doesn’t swing the headline, and short enough to show a real shift within days. The app labels the trend as rising, falling or stable, draws both lines on a 30- or 90-day chart, and lists the individual polls underneath with each one’s sample size, population and field date. It’s free on iPhone.
Why do approval ratings go up and down? #
- The start of a term. New presidents usually get a honeymoon, which fades within months.
- The economy. Prices, gas costs and jobs have long been among the strongest drivers of approval.
- Crises. Sudden national emergencies can produce a short “rally” effect. George W. Bush’s 90% in Gallup’s poll after September 11, 2001 is the highest Gallup ever recorded; Harry Truman’s 22% in February 1952 is the lowest.
- Partisanship. In the current era, most voters’ views of a president are locked in by party. Ratings move in a much narrower band than they did in the mid-20th century, so a shift of three or four points is meaningful.
For how approval differs from asking whether people like the president personally, see favorability vs. approval rating.
Frequently asked questions #
How many people are surveyed for an approval rating? #
A typical national poll interviews about 800 to 1,500 adults. Beyond roughly 1,000, adding more people improves precision only slightly, so most pollsters stop around there.
What’s the margin of error on approval polls? #
For a random sample of 1,000, about ±3 percentage points at 95% confidence. Weighting adds some extra uncertainty, and the margin on the gap between approve and disapprove is larger than on either number alone.
Why do different polls show different approval numbers? #
They survey different populations, use different methods and weight their samples differently, and each has a house effect. Averages combine them, but averages use different rules too, so small differences are normal.
Does Gallup still measure presidential approval? #
No. Gallup, which began asking the question in the 1930s, announced in 2026 that it would stop publishing presidential approval ratings. Many other pollsters and averages continue.
Is net approval better than the approval number? #
Often. Net approval accounts for both approval and disapproval, so it captures shifts that the approval figure alone can hide, such as people moving from undecided to disapprove.