AP® Psychology · Science Practices: Research Methods and Data
2.C Surveys, Likert Scales, and Self-Report Bias
A survey turns opinions into numbers using Likert scales, but the exact wording of a question and whether respondents feel anonymous can shift those numbers without any real change in what people actually think or do.
A Likert scale asks someone to pick a point on a labeled range, like 1 (strongly disagree) to 5 (strongly agree). That turns a fuzzy opinion into a number, and numbers can be averaged, compared across groups, and tracked over time. The catch is that the number only means what the words around it mean. Swap "should the government support" for "should the government be required to fund," and you are no longer measuring the same attitude, even though the scale looks identical. Small wording shifts change how strong or loaded a statement feels, which shifts where people land on the scale, which shifts the whole distribution of answers.
Self-report also runs into social desirability bias: people answer in ways that make them look good, especially when someone else can see their response. Ask face-to-face whether someone speeds, cheats, or drinks heavily, and answers drift toward the modest, responsible-sounding choice. Ask the identical question through an anonymous online form, and reported rates of the same behavior tend to rise, because the social cost of an honest answer has dropped. The behavior did not change between the two surveys; only the willingness to admit it did.
This matters for research methods because it is a threat to validity that lives entirely in how data is collected, not in the real world the data is supposed to describe. Researchers manage it by standardizing wording across groups, pilot-testing questions for leading language, and using anonymous or confidential formats when topics are sensitive. Connected ideas worth keeping straight: question wording effects distort an individual item, while social desirability bias distorts answers to a whole category of sensitive topics regardless of exact phrasing.
Key terms
- Likert scale
- A response format where people rate agreement or intensity on an ordered numeric scale, such as 1 to 5.
- social desirability bias
- The tendency to shift self-reported answers toward what seems socially acceptable when one's identity is known.
- self-report
- Any data collection method where participants state their own thoughts, feelings, or behaviors rather than having them observed.
- question wording effect
- A change in survey responses caused by altering the phrasing of a question rather than any real change in opinion.
- anonymity
- A survey condition where responses cannot be linked back to the individual who gave them.
Watch it explained
- How it worksHow does a Likert scale turn an opinion into a number that can be averaged?
- How it worksWalk me through social desirability bias shifting answers toward what sounds better.
- Worked exampleShow me the same question worded two ways and the two response distributions it produces.
- Going furtherWhat changes about the answers when the survey is anonymous instead of face to face?
Check yourself
End of topic.