Randomised Response, explained.
Randomised response mixes truthful reporting with a known random mechanism to obscure individual answers while permitting population estimation under assumptions.
Why it happens
Half of respondents answer a private fair coin and half answer the supplied truth. Expected yes fraction is 0.25 plus half the prevalence; inversion gives the raw estimate 2 times yes fraction minus 0.5.
Randomised response mixes truthful reporting with a known random mechanism to obscure individual answers while permitting population estimation under assumptions.
Read the result
The raw prevalence estimate can exceed the parameter’s range through sampling noise. Keeping it unclipped exposes that limitation rather than hiding it.
A worked example
Recover a population signal from deliberate noise
Keep the fictional trait prevalence forty percent and collect one hundred reports.
Compare the realised yes fraction with expected yes fraction 0.45 under the declared coin protocol.
Apply twice the yes fraction minus one-half and distinguish estimation error from an incorrect reporting formula.
OPTIONAL DEEPER DETAILGo deeper: inside the model
Inside this model
Half of respondents answer a private fair coin and half answer the supplied truth. Expected yes fraction is 0.25 plus half the prevalence; inversion gives the raw estimate 2 times yes fraction minus 0.5.
Where this idea is useful
A practical use
Distinguish individual ambiguity from recoverable aggregate information in a carefully specified survey protocol.
A common misconception
“Deliberately random answers make every aggregate estimate impossible.”
A known reporting mechanism can preserve recoverable population information while obscuring individual answers, subject to compliance and sampling assumptions.
What this explanation leaves out
- The teaching population is synthetic and compliance is perfect. No production privacy guarantee is made; noisy estimates can lie outside zero to one and remain unclipped.
Is this a production anonymity guarantee?
No. The teaching protocol assumes perfect compliance and synthetic respondents; implementation, disclosure and adversarial issues require additional analysis.
Which random mechanism and compliance assumption make the aggregate inversion valid?
Associated thinkers
Associations marked provisional are awaiting source review.
Further reading
Explore the original research or the teaching reference behind this experiment.