The Mehrabian myth

In 1967 Albert Mehrabian, a psychologist at UCLA, published two experiments with colleagues. The first, with Morton Wiener, played listeners recordings of single words, nine of them, including honey, thanks, maybe, and brute, spoken in tones that matched or contradicted the word's meaning. Listeners judged how the speaker felt about the person being addressed. When word and tone disagreed, the tone carried the judgment.

The second, with Susan Ferris, used one word, maybe, spoken in three tones, set against photographs of three facial expressions. Participants again judged the speaker's feeling. The facial expression outweighed the tone by roughly three to two.

Both studies used small groups of female students, single words, and one question: how does this speaker feel? In his 1971 book Silent Messages, Mehrabian combined the two results into an equation for that specific situation, total liking equals 7 percent verbal plus 38 percent vocal plus 55 percent facial. He noted himself that the numbers were approximate.

Then the equation escaped. Stripped of its conditions, it became a universal claim about all communication, printed on training slides, quoted in management books, and taught in presentation courses as the scientific case that words barely matter. Mehrabian has spent much of the decades since correcting it. His own website states that the equations were derived from experiments dealing with communications of feelings and attitudes, and that unless a communicator is talking about their feelings or attitudes, they are not applicable. In correspondence he has described being uncomfortable with the misquotation from the very beginning.

The scale of the misreading has itself been measured. A communication researcher, David Lapakko, checked 79 websites citing the numbers and found one that described the actual finding correctly.

The test that exposes the myth needs one sentence. If words carried 7 percent of communication, you could follow a lecture in a language you do not speak and receive 93 percent of it.

What the studies do support is narrow and real: when someone's words and delivery disagree about how they feel, listeners believe the delivery. Fine, said in a flat voice, reads as not fine. That finding has held up in later work on emotional expression. Everything beyond it, the percentages applied to presentations, negotiations, meetings and speeches, is the myth. For another textbook staple that its own author watched escape, see the Maslow pyramid.

The place this myth costs money is preparation. Somebody planning a pitch or a difficult conversation hears that delivery is 93 percent of the outcome, and reallocates accordingly: less time on what to say, more on how to stand. The advice sounds scientific and the science underneath it is about single words spoken to students judging feelings.

The honest split is different. The content of a pitch, the argument, the numbers, the sequence of claims, is carried almost entirely by words, which is why it cannot be followed in an unfamiliar language. What the research actually licenses is a rule about consistency. When your delivery contradicts your words, the room believes the delivery. A confident claim in a hesitant voice reads as hesitation. Reassurance delivered while glancing at the door reads as the door. So the words earn the meaning, and the delivery has one job, which is not to argue with them.

There is also a quieter lesson in how the number survived. A precise statistic that fits on one slide travels further than the study underneath it, and almost nobody walks back to the source. The 79 websites result is the measurement of that. Any figure that arrives this clean, this memorable and this convenient deserves the walk back, which is roughly the founding habit of this library.

Source: Mehrabian and Wiener, Decoding of Inconsistent Communications, Journal of Personality and Social Psychology, volume 6, 1967, pages 109 to 114, DOI 10.1037/h0024532. Mehrabian and Ferris, Inference of Attitudes from Nonverbal Communication in Two Channels, Journal of Consulting Psychology, volume 31, 1967, pages 248 to 252, DOI 10.1037/h0024648. The combined equation appears in Mehrabian, Silent Messages, 1971. The audit of how the numbers are cited: Lapakko, Communication is 93 Percent Nonverbal: An Urban Legend Proliferates, Communication and Theater Association of Minnesota Journal, 2007.

Made to Stick

Chip and Dan Heath, 2007

On why some ideas spread and survive whether or not they are true. The Mehrabian numbers are a textbook case of what this book describes: simple, concrete, credible-sounding, and wrong.