How to Read Political Polls Without Getting Fooled by the Headline
Turn toplines into a better estimate by checking sample, field dates, wording and uncertainty.
Read field dates first
Opinion can move quickly during campaigns; a poll may describe a period that ended before the headline was published.
Check population and sample
Adults, registered voters and likely voters are different populations.
Treat margin of error as incomplete
Sampling error is only one source of uncertainty; weighting, nonresponse and modeling matter too.
Compare trends, not single polls
For day-to-day verification, the practical consequence is straightforward: A sequence of comparable surveys is more informative than one surprising result.
Read the exact question wording
For day-to-day verification, the practical consequence is straightforward: Order and phrasing can change responses, especially on unfamiliar policies.
Separate percentage points from percent change
A move from 40% to 44% is four points but a ten percent relative increase.
Poll sampling calculator
For issue questions, read the exact wording. “Support,” “approve,” “favor” and forced-choice questions can produce different answers. The wording belongs next to the percentage when the result is being used as evidence for a public-opinion claim.
One poll can move because of sampling noise. A trend is more persuasive when several surveys using reasonably similar populations point in the same direction. When comparing over time, prefer the same pollster and method where possible, and note methodology changes that break the series.
Compare trends with comparable surveys
Margin of error is only one source of uncertainty. Question order, wording, response mode, weighting and assumptions about likely voters can shift results. Small differences between two candidates may be less informative than the headline implies, especially when comparing separate polls with different methods.
A poll is a measurement of a defined population during a defined period. Check who was eligible for the sample, when interviews were conducted and whether the result refers to adults, registered voters or likely voters. A headline can remain technically accurate while readers assume a broader population than the poll actually measured.
Read the field dates before the topline
When a poll is used to claim movement, check whether the change exceeds ordinary sampling variation and whether the previous comparison point came from the same methodology. A dramatic headline built on a one-point shift deserves less confidence than a sustained change across repeated surveys.
Issue approval, candidate favorability, vote intention and expectations about who will win are different measurements. A person can dislike a candidate yet intend to vote for them, or expect a candidate to win while preferring the opponent. Headlines sometimes blend these measures because they move in the same political story, but the survey questions are not substitutes.