Lie Predictor — check how likely people are to lie about any survey question

Check any question to see how likely people are to lie about it.

Try:

Why this happens

Every question above is distorted before anyone answers it. Some of it is people managing what you think of them. Most of it is people managing what they think of themselves.

Early Studies is a consumer research company, and designs research around that. Most market research treats an answer as a fact. We treat it as behaviour — something produced under social pressure, by a person with limited access to their own motives. We build questions people can answer honestly without having to be brave, and that reveal what they will actually do rather than what they would like to believe about themselves. The result is a reading of what is true, not a record of what was said.

The five ways an answer goes wrong

Won't say
They know the truth and keep it from you. Social desirability, impression management, fear of disclosure. Anonymity helps this one.
Won't confront
They keep it from themselves. Admitting it would cost too much. Anonymity does nothing.
Can't know
They have no reliable access to the answer. Memory is reconstruction, not playback, and nobody counts their own behaviour.
Can't predict
The question is about a version of them that does not exist yet. Intention is a poor guide to conduct.
Won't bother
They did not do the work. Straightlining, acquiescence, an opinion invented on the spot to fill a box.

Questions

Do people lie in surveys?

Constantly, and mostly not on purpose. Deliberate concealment is the smaller half. The larger half is people answering honestly about a version of themselves they have never examined. Both produce the same thing: a number that does not describe behaviour.

What is social desirability bias?

The pull towards the answer that makes you look good. It is strongest when the true answer carries a social cost, and weakest when a lie would be exposed by a single follow-up question. People rarely lie about things that collapse immediately.

Does anonymity stop people lying in research?

Only half of it. Anonymity lowers what people will not say to you. It does nothing to what they will not admit to themselves. A promise of privacy does not make anyone more honest about their own spending, drinking or motives.

What is social circle questioning?

Asking what someone's friends, colleagues or peers do rather than what they do. It removes personal exposure, so concealment collapses to near zero. What remains is estimation error, which is honest and imprecise but not a lie. The Lie Predictor scores these questions below 10 percent.

Can AI replace consumer research?

No, but it can tell you which questions are worth asking. The Lie Predictor uses a large language model to score a question against five known distortion mechanisms before you field it. The fieldwork still has to happen with real people.

Who built the Lie Predictor?

Early Studies, a consumer research company. We design studies around how people actually answer, using behavioural science, consumer psychology and AI-enabled research tools to get a reading of what is true rather than a record of what was said.