As we all know, 78.3 percent of statistics are made up, probably because exact, precise numbers seem very authoritative and neat. One such made up but commonly repeated statistic is this: 93 percent of communication is body language, or 93 percent of all human communication is nonverbal. The wording and exact percentage sometimes varies, but the essence remains the same: words mean almost nothing. Almost all communication is nonverbal.
It seems intuitively correct that nonverbal cues are important, but even if we know nothing about this claim or where it came from, we should immediately be very suspicious. Whenever someone tells you that a complex phenomenon with lots of variables is actually really simple, you should be suspicious. Human communication is insanely complicated. It seems prima facie impossible to design a valid experiment to quantify this exactly. And it’s easy to come up with mundane experiments anyone can do to disprove it: would you expect to understand 93 percent of a film in a foreign language?
Most people who repeat factoids never cite sources. This particular factoid can be traced back to science, but it’s not very good science. It’s based on two papers published by psychologist Albert Mehrabian in 1967. These two studies are flawed in two serious ways: the conclusions they draw, and the conclusions extrapolated from those conclusions, are much too ambitious for the evidence; and the evidence is dubious in itself, because of statistical errors and limitations in the experimental design.
The papers are: Decoding of Inconsistent Communications, by Mehrabian and Wiener (1967); and Inference of Attitudes from Nonverbal Communication in Two Channels, by Mehrabian and Ferris (1967). From these studies, Mehrabian infers his so-called 7-38-55 rule: that human communication is 7% verbal, 38% is tone of voice, and 55% is facial expessions. Trouble is, he does this by combining results from different experiments in a way that is not statistically sound. Further, the studies in question are very narrow: they involve subjects reacting to single words and facial expressions or tones of voice selected from very limited sets. And every single subject was a female psychology student. No men, no one outside of that limited background and age range. It’s like a textbook on how not to design an experiment if you want to make ambitious claims. Even if you assume the results obtained are valid, they are obtained from such a small sample in such a limited environment as to be almost completely useless in other contexts.
Given all this, even Mehrabian’s more modest equation, “Total Liking = 7% Verbal Liking + 38% Vocal Liking + 55% Facial Liking” seems highly suspect.
Obviously nonverbal communcation is important. But not that important. So why do we fall for this? Frankly, because people repeat information they hear uncritically, and because some people got the chain going by drawing invalid inferences from limited data. The reason this particular claim got so widespread probably has to do with the seductive qualities of numbers (they seem so authoritative, so precise) and the fact that they line up with certain people’s agendas.
Not that anyone reading this blog would fall for the “commonly repeated statistical factoid” fallacy. But just in case: use your heads, people.
dailymeh posted this on October 16, 2011
