A poll asks hundreds or thousands of people the same fixed questions so it can estimate how many people in a population hold a view. A focus group puts about eight to ten carefully chosen people in a room, or on a video call, for a guided conversation that shows why they feel the way they do and what words they use. Polls produce numbers you can generalize; focus groups produce insight you can’t.
Campaigns use both, usually in sequence. The mistake readers make is treating a focus group like a small poll.
What’s the core difference between a poll and a focus group? #
| Poll | Focus group | |
|---|---|---|
| Type of research | Quantitative | Qualitative |
| Typical size | 500 to 1,500+ respondents | About 6 to 12 people per group |
| How people are chosen | Random or statistically balanced sample | Screened to fit a specific profile |
| Questions | Fixed, mostly closed-ended, same order for everyone | Open-ended, guided by a moderator |
| Output | Percentages with a margin of error | Themes, quotes, reactions |
| Can you generalize? | Yes, within the margin of error | No |
| Best for | How many, how much, who’s ahead | Why, how it feels, which words land |
How does a political poll work? #
A pollster draws a sample meant to represent a larger group, such as likely voters in a state, and asks everyone the same questions. The answers are weighted so the sample matches known population totals, then reported as percentages. Because the sample is designed to represent the population, the pollster can estimate how far off the numbers might be, which is the margin of error.
Poll questions are mostly closed-ended: approve or disapprove, Candidate A or Candidate B, very likely or not likely to vote. That rigidity is the point. Everyone gets the same stimulus, so differences in answers reflect differences in people.
How does a focus group work? #
Recruitment #
A recruiting firm screens people with a questionnaire to find a precise profile, such as suburban women who voted for one party for president and the other for Senate, or voters who say they’re undecided. Participants are usually paid for their time. A campaign often runs several groups in different places to avoid reading too much into one room.
The moderator #
A trained moderator follows a discussion guide but chases whatever’s interesting. If someone says a candidate “seems fake,” the moderator asks what gave that impression, whether others agree and what would change their mind. Groups typically run 90 minutes to two hours, with the campaign’s team watching from behind a mirror or on a muted video feed.
Dial testing #
In some sessions, participants watch a debate clip or an ad while turning a dial or moving a slider to show how they feel second by second. The result is a trace line that shows exactly which sentences land and which fall flat. TV networks have used dial groups during presidential debates.
What do campaigns use each one for? #
Most campaigns run them in a loop:
- Focus groups first, to find the language. Before writing a poll, a campaign might run groups to learn which issues voters raise unprompted and how they describe them. A worry nobody on the staff expected, like a local utility rate hike, can come up this way.
- A poll to size it. The campaign then writes poll questions about that issue to measure how many voters share the concern, and among which groups.
- Message testing. Campaigns test competing arguments in polls, often by showing different versions to random halves of the sample, to see which one moves more people.
- Back to focus groups to understand why the winning message works and to refine ads before spending money on airtime.
Focus groups can also warn a campaign about something a poll would miss, such as a phrase that reads as condescending or an ad image that distracts from the message.
Why can’t you treat a focus group like a poll? #
Because a group of ten people isn’t a sample of anything larger. Three problems make that worse:
- Selection. Participants were recruited to fit a profile, and people who sign up for paid research may differ from others in that profile.
- Group dynamics. One confident talker can pull the room. People soften or hide views they think are unpopular.
- Small numbers. Saying “70% of the group” when the group has ten people means seven people. That’s a quote, not a statistic.
This applies to the focus groups news outlets publish, often of swing voters talking about the candidates. They’re genuinely useful for hearing how real voters talk, and they often surface themes that later show up in polling. They can’t tell you what share of voters feel that way. When you read one, look for polling on the same question before drawing a conclusion. Our guide to reading poll crosstabs shows how to find what specific groups think in real survey data.
Are there hybrids? #
Yes. Pollsters increasingly add open-ended questions to large online surveys and then code the answers into themes, which gives some of the “why” at scale. Online discussion boards let dozens of participants respond over several days. Some campaigns run large online ad tests where each viewer rates an ad, which combines focus-group-style stimuli with a big enough sample to count.
These are useful, but the basic trade-off holds: the more a method lets people explain themselves in their own words, the harder it is to turn into a trustworthy number.
Which should you pay attention to as a voter? #
For “who’s ahead” and “how many people think X,” look at polls, and preferably an average of polls rather than any single one. For understanding the mood behind the numbers, focus groups and interviews are worth reading, as long as you remember they’re a handful of people.
Election Tracker is built for the first job. It groups every published poll for Senate, House (the generic ballot), governor and 2028 primary races, shows each poll’s pollster, sample size, population and dates, and computes a 30-day average for each race. When a viral focus group quote makes a race sound like it’s shifting, you can check whether the polling actually moved. It’s free on iPhone.
Frequently asked questions #
Which is more accurate, a focus group or a poll? #
They measure different things. A well-run poll is accurate at estimating how many people hold a view, within its margin of error. A focus group is good at showing why people feel a certain way, but its results can’t be projected onto a population.
How many people are in a focus group? #
Usually about six to twelve. Researchers often run several groups on the same topic in different places to see whether the same themes come up.
Are focus group participants paid? #
Usually, yes. Participants typically receive a cash payment or gift card for their time, and the amount depends on the length of the session and how hard the profile is to recruit.
Can a focus group predict an election? #
No. It’s too small and isn’t randomly selected. It can reveal how some voters think about the candidates, but it can’t tell you how many voters think that way.
What is dial testing? #
Participants turn a dial or move a slider while watching a speech, debate or ad to show their reaction moment by moment. Researchers use the resulting trace line to see which lines work and which don’t.