Time to Put the 7% Myth to Rest

You should have no trouble figuring out what he’s saying

One of the favorite statistics cited by communication “experts” is that only 7% of the meaning from spoken communications comes from the actual words spoken. As the story goes, 55% comes from facial expression, and 38% comes from body language, tone of voice, etc.

It has been around ever since Albert Mehrabian cited those statistics in a book entitled Silent Messages, published in 1971.

These experts use it to stress the importance of paying attention to non-verbal signals, whether you are the listener or the speaker. It’s a good statistic to cite because it’s appropriately surprising and it lends an air of science and precision.

The problem with the statistics cited is that it’s mostly false; in my own very unscientific estimate, it’s probably about, oh, let’s say 7% true.

If it were actually true, then when I was in Italy last week, I should have had no problem understanding 93% of what the taxi drivers told me (I didn’t). Plus, I could save a ton of money not buying headphones to watch airplane movies. If it were actually true, then listening to an educational podcast or talking on the phone garners you less than half of the message. And of course, you probably would not be able to understand this article unless I filled it with emoticons, which I refuse to do. It’s so patently untrue that when I read or hear that from someone, I automatically disqualify them as a credible source.[1]

But most people aren’t that simplistic. Some who cite the study come closer to the truth by qualifying it to the part of the message that contains feelings and attitudes. And that definitely makes sense in a lot of communications. If I ask someone how their meeting went, and they answer “great”, I can instantly tell whether they are sincere or sarcastic. In that situation, 0% of the message came from the actual meaning of the word; they could have answered me in Swahili and I would have understood.

But of course it gets more ambiguous as messages get longer, and it definitely does not apply when the speaker is deliberately trying not to show their true feelings. It also does not apply when someone is explaining factual or technical information. If I ask someone for directions, their facial expressions won’t make much difference in my understanding.

So, what did Mehrabian actually measure, and what did he say? Three female speakers were recorded saying one word, “maybe” in either a like, neutral or dislike tone of voice, and then 17 subjects listened to the recordings and were asked to infer what the attitude of the speaker was to a third party to whom they were presumably speaking. A follow-up study was then done with 30 subjects, using nine different words grouped according to the same three attitudes, and the results of both studies were combined to arrive at the statistics cited.

It’s fascinating to me that a study using 10 total words, 47 subjects, conducted in 1967, is still so influential today. As my friends on Sports Center would say, “C’Mon, Man!” (and you can imagine my tone of voice as I say it).