The statistics you were taught

Four numbers dominate communication training. Three of them have no research behind them and the fourth means something other than what it is used for.

"Listening is 45 per cent of our communication time." This is Paul Rankin's 1926 figure. The method: 21 adults, including farmers and housewives, writing down what kind of communication they were doing every fifteen minutes for one day. His figure was 42 per cent. Across nine subsequent time studies the range runs from 15 per cent to 55 per cent, and the largest modern study, with 680 participants and a scaled instrument, found 24 per cent (Janusik & Wolvin, 2009).

"We remember only 25 per cent of what we hear." From a 1957 Harvard Business Review article by Nichols and Stevens. It reports no sample size, no test instrument, no control condition and no statistics of any kind. And the figure is misquoted: the article's 25 per cent is retention two months after listening to a talk. Its immediate figure is about 50 per cent. A quarter of a single unrehearsed talk retained after two months is ordinary forgetting, and says nothing about listening skill.

"93 per cent of communication is nonverbal." Two 1967 experiments. Participants judged a speaker's attitude from a single spoken word paired with a still black-and-white photograph, with 62 female participants, in a design where the verbal and vocal cues deliberately conflicted. Mehrabian's own published caveat: unless a communicator is talking about their feelings or attitudes, the equations do not apply. In the thin-slice meta-analysis, transcripts alone reached r = .29, against tone of voice at .26.

"Eleven million meetings a day cost $37 billion." The first figure comes from a 1976 trade book, p. 4, with no source given then or since. The second comes from a 1989 magazine article with no methodology. The modern replacements circulating as facts are a 2015 blog model whose own stated ranges are 36 to 56 million meetings and $70 to $283 billion wasted.

Two things are worth taking from this beyond the corrections themselves.

The first is a pattern. Every one of these numbers is more precise than its evidence, older than its user assumes, and repeated because it is useful rather than because it is true. When you meet a communication statistic, the questions that resolve it in about four minutes are: who was measured, how many, when, and does the number mean what it is being used to mean. The 25 per cent figure fails the fourth question rather than the first three, which is the harder failure to spot.

The second is what to do with the space they leave. The listening statistic was doing rhetorical work: it was there to establish that listening matters. Listening does matter, and the evidence for it is far better than a 1926 diary study. Being listened to reduces a speaker's anxiety, raises their self-awareness and increases their clarity about their own position, pooled d = 0.46 across five studies and 717 participants, one of them preregistered (Itzchakov, DeMarree, Kluger & Turjeman-Levi, 2018). That is Level 5, and it is a stronger argument than the number it replaces.

The same applies to the meetings figures. What survives peer review is duller and more useful: the average employee spends about six hours a week in scheduled meetings and senior managers about 23; meeting frequency correlates with effectiveness at r = −.24 and with satisfaction at −.30; and meeting duration shows no significant relationship with effectiveness at all.