Audit advanced Works with: Any AI

Sampling and Representativeness Check

Decision tree Has example

Judge whether respondents represent the population.

The Open buttons send the prompt text as currently shown below to that site only when you click. Fill in the placeholders first.

Help me judge whether my survey respondents represent the people I want to learn about. Target population: [TARGET_POPULATION]. How respondents were recruited: [RECRUITMENT_METHOD]. Number of respondents and a demographic or segment breakdown: [RESPONDENT_BREAKDOWN]. Known population figures to compare against: [POPULATION_FIGURES].

Run this decision tree and report your path:
1. Is the respondent list drawn at random from the population, or did people choose to respond? If self-selected, note the likely bias (the most engaged, the angriest, the most reachable).
2. What was the response rate? If it is low, ask what is known about non-responders.
3. Compare respondent profile with population figures in a table (group, share in sample, share in population, gap). Which groups are over- or under-represented?
4. Could the topic itself affect who answers (for example a satisfaction survey sent only to active users)?
5. Is weighting possible and sensible, and is the sample large enough in the small groups for it?
6. Which conclusions are still safe, which should be softened, and which should be dropped?

End with a plain-language limitations statement of about 80 words for the report and three practical ways to improve the next round. Do not claim statistical precision the sample cannot support.

Fill in before use

Optional: type your details and the prompt above updates. Nothing is saved or sent.

Replace each bracketed placeholder with your own details, then paste into Any AI or another assistant.

Path: self-selected survey link sent by email newsletter; response rate 6% (240 of 4,000).
Comparison with customer base
Group | In sample | In customer base | Gap
Over 60 years | 4% | 22% | under-represented
Enterprise accounts | 38% | 12% | over-represented
Conclusions still safe: opinions of engaged newsletter readers. Soften: "customers prefer..." statements. Drop: anything about older customers.
Limitations statement: "Results reflect 240 newsletter subscribers and may not represent all customers, especially older users. Differences below about 6 percentage points should be treated as uncertain."
Improve: follow up with a random sample of 100 non-responders by phone.

• sampling
• representativeness