AP® United States Government and Politics review sheet from Aim for Five (aimforfive.com/gov/units/4/4-5)
Unit 4 · Topic 4.5
4.5 Measuring Public Opinion
Scientific polls measure public opinion by asking a carefully chosen sample of people and generalizing to the whole population. Different polls do different jobs: opinion polls on issues, benchmark polls at a campaign's start, tracking polls over time and exit polls of voters leaving the polls. A good poll uses a random, representative sample, reports its margin of error and asks neutral, clearly worded questions.
Key terms
- random sample
- margin of error
- benchmark poll
- tracking poll
- exit poll
- question wording
Types of polls
- Opinion poll: measures views on issues, candidates or officials at one point in time.
- Benchmark poll: taken at the start of a campaign to measure a candidate's name recognition and standing, setting a baseline to measure progress against.
- Tracking poll: repeated frequently, often daily, with new samples to follow changes in support over time, especially late in a campaign.
- Exit poll: surveys voters as they leave polling places (and now also early and mail voters) to learn how different groups voted and why. News organizations use them to analyze results.
- Focus group: a small group discussion led by a moderator to explore how people react to messages. It gives depth, but it isn't a scientific poll because the group is too small to represent the population.
Sampling
A population is the whole group you want to learn about, such as all likely voters in a state. A sample is the smaller group actually surveyed. In a random sample, every member of the population has an equal chance of being chosen. That's what makes a sample representative, meaning it mirrors the population's makeup.
Pollsters also weight results, adjusting them so that the sample matches the population on traits like age, education, race and region. If a sample has too few young people, each young respondent counts a bit more.
Bad samples include self-selected ones, like online polls anyone can answer, and samples drawn from a skewed list. Such polls can be huge and still wrong.
Sample size and margin of error
The margin of error tells you how much a poll's result might differ from the true value in the population just because of random sampling. A result of 52% with a margin of error of ±3 points means the true value is likely between 49% and 55%. Most polls report a 95% confidence level.
Larger samples have smaller margins of error, but with diminishing returns. A rough rule is that the margin of error is about 1 divided by the square root of the sample size.
| Sample size | Approximate margin of error (95% confidence) |
|---|---|
| 400 | ±5 points |
| 1,000 | ±3 points |
| 2,500 | ±2 points |
Question wording and format
Wording can change answers. A leading question pushes respondents toward an answer (“Do you support the reckless plan to…?”). Loaded terms, double-barreled questions that ask two things at once, and the order of questions can all bias results. Good polls use neutral, clear, single-issue questions.
Watch out for push polls, which aren't real polls at all. They're campaign tactics disguised as surveys, using questions to spread negative information about an opponent.
Worked examples
Try each one yourself first, then open the solution.
- Example 1
Reading a margin of error
A poll of 1,000 likely voters shows Candidate A at 48% and Candidate B at 45%, with a margin of error of ±3 points. A headline says “Candidate A leads.” Is that conclusion justified?
Show the solutionHide the solution
- Step 1: Find A's likely range: 48 ± 3, or 45% to 51%.
- Step 2: Find B's likely range: 45 ± 3, or 42% to 48%.
- Step 3: Compare: the ranges overlap (45% to 48% is in both), so B could actually be ahead.
- Step 4: Conclude: the 3-point gap is within the margin of error, so the race is a statistical tie.
Answer: No. The 3-point gap is within the margin of error and the ranges overlap, so the poll shows a statistical tie, not a clear lead.
Common mistakes
- Thinking a bigger sample always fixes a bad poll. A huge self-selected sample is still unrepresentative; random selection matters more than size.
- Calling a lead within the margin of error a real lead. If the ranges overlap, the race is too close to call.
- Confusing a benchmark poll (a starting baseline) with a tracking poll (repeated measurement over time).
On the exam
- Expect data questions that give a poll result and margin of error and ask what conclusion is supported.
- Questions may describe a poll's method and ask about its weakness. Look for sampling problems or biased wording.
Connected topics
Videos
Check yourself
4 questions on 4.5 Measuring Public Opinion. Pick an answer to see if you got it, and why.
A city council member wants to poll residents about a proposed stadium. Which question is worded most neutrally?
In the two months before a primary, a campaign surveys likely voters every few days to see whether its new television ads are changing support for its candidate. This is an example of a
On election day, interviewers stand outside randomly chosen polling places and ask voters who they voted for and why. News networks use the results to help explain and project the outcome. This is
A year before the election, a possible candidate for governor pays for a survey to find out how well known she is, how voters see her and which issues they care about most. This is best described as
0 of 4 answered