One of psychology’s strangest experiments found that spilling coffee made a highly competent person more likeable but did not rescue an average one, suggesting that imperfection is charming only after people have already been given a reason to trust the person underneath it

One of social psychology’s most memorable experiments begins like a job interview and ends with the sound of crockery.

Male university students listened to a recording of a stranger answering difficult quiz questions. Some heard a highly competent candidate. Others heard an ordinary one. At the end of half the recordings came a clatter, followed by the embarrassed admission that the candidate had spilled coffee over his new suit.

The mistake made the impressive man more likeable. It made the average man less likeable.

That 1966 result became known as the pratfall effect. It is often retold as proof that flaws are attractive. The experiment actually found something narrower and more useful: imperfection can humanise a person only after observers have been given a reason to admire or trust them.

Forty-eight students heard four versions of a candidate

Elliot Aronson, Ben Willerman and Joanne Floyd recruited 48 male sophomores from an introductory psychology course at the University of Minnesota. Each participant listened to one of four tape recordings presented as an audition for a quiz programme.

The design crossed two variables. The candidate was either highly competent or average. He either committed the coffee blunder or completed the interview without incident.

Competence was made difficult to miss. The superior candidate answered 92 per cent of the quiz questions correctly and described an excellent academic record, success in high school and membership of the college track team. The average candidate answered only 30 per cent correctly and presented a more ordinary biography.

In the pratfall versions, listeners then heard the experiment’s famous line: “Oh my goodness, I’ve spilled coffee all over my new suit.” The participants rated how attractive, in the interpersonal sense, they found the candidate.

The coffee did not make everyone charming

The most attractive candidate was the highly competent man who spilled the coffee, with a mean attraction score of 30.2. The least attractive was the average man who made the identical mistake, whose mean score was minus 2.5.

The statistical interaction between competence and the pratfall was significant. In other words, the meaning of the blunder depended on who committed it. The spill raised liking for the superior candidate and lowered liking for the average one.

Importantly, the pratfall had no significant main effect by itself. Mistakes were not generally attractive. A mistake changed the interpretation of competence already on display.

Aronson and his colleagues proposed that exceptional ability can make someone seem distant, “too good” or difficult to approach. A small clumsy error removes some of that distance. The observer retains evidence of capability while discovering evidence of ordinary humanity.

Trust appears to be the price of admission

The experiment helps explain why two people can make the same error and receive opposite reactions. When a respected colleague briefly loses their place in a presentation, the lapse may soften a polished image. When a new colleague who has not yet demonstrated command of the material does the same thing, observers may treat it as further evidence of uncertainty.

This is not necessarily fair. First impressions are noisy, confidence can be mistaken for ability and competence can be distributed unevenly across contexts. But the psychology is coherent: ambiguous information is interpreted through the story already forming about a person.

A blunder is therefore not an independent personality trait. It is a modifier. If the underlying signal says capable, the error can read as human. If the signal says unproven, the error may confirm doubt.

The principle has obvious relevance in technology workplaces, where expertise can create social distance. A brilliant engineer who admits forgetting a command can become easier for junior colleagues to approach. A founder who calmly corrects a mistaken forecast may appear more trustworthy than one who performs certainty. The prior evidence of competence does the protective work.

The classic study has very narrow shoulders

It would be a mistake to make this small experiment carry an entire theory of authenticity.

The participants were 48 male American sophomores in 1966. They evaluated a male voice on a scripted tape, not a colleague, leader or romantic partner encountered over time. The highly competent and average candidates were deliberately exaggerated. Attraction was measured immediately after a short artificial encounter.

The study also does not show that deliberate vulnerability works. Participants believed the spill was accidental. A rehearsed confession, strategically messy desk or manufactured typo may signal manipulation instead of humanity.

Later research found that responses can vary with an observer’s self-esteem, identification with the person, gender, similarity and the severity of the mistake. In a 1970 follow-up, Aronson and colleagues found that the observer’s self-esteem altered reactions to competence and a pratfall. Other studies have not produced one simple effect across every context.

The original result is a demonstration of an interaction, not a law that dropping things makes talented people popular.

Imperfection helps most when it is informative

There are at least two kinds of flaw. One threatens the quality people care about. The other reveals fallibility without erasing that quality.

A surgeon operating on the wrong site is not charmingly imperfect. A surgeon who laughs after fumbling with a presentation remote may become more approachable. A security engineer who misses a major vulnerability has created a competence problem. The same engineer admitting that they once locked themselves out of a test account has offered a human detail.

Scale matters, as does relevance. The classic coffee spill did not undermine the candidate’s intelligence. It introduced a harmless contrast to it.

This may also explain why polished self-presentation often feels brittle. When every sentence, photo and career story has been stripped of accident, audiences have no evidence of how the person handles ordinary friction. A minor, well-managed mistake can provide that evidence.

The lesson is not to become worse at your work

Popular advice sometimes converts the pratfall effect into a performance tactic: reveal a weakness so people will like you. That gets the order backwards.

The original experiment established competence first and fallibility second. Without the first signal, the same blunder reduced attraction. The useful sequence is not flaw, then trust. It is trust, then permission to be visibly human.

For leaders, this can mean admitting a small error without defensiveness after building a record of sound decisions. For experts, it can mean saying “I don’t know” when the boundary of knowledge is genuine. For anyone starting a new role, it may mean focusing less on looking perfectly relaxed and more on doing dependable work.

You do not need to spill the coffee. Eventually, something will spill by itself.

The question is whether people have already seen enough of the person underneath to interpret the mess generously.