When people want to convince others that a statistic is big or small, they tend to describe it in words instead of just stating it, according to a new study. Audiences find those word-based descriptions persuasive even when they are told the speaker has an agenda, MIT News reported.
The paper, “Numbers Tell, Words Sell,” is being published in the American Economic Review. Its authors are Victor de Chaisemartin, a doctoral candidate at the MIT Sloan School of Management; Michael Thaler of University College London; and Mattie Toma of the University of Warwick.
The researchers give a simple example. Suppose 57% of an office’s staff drives to work. A manager who wants the figure to look large might call it “a clear majority” or “approaching two-thirds.” One who wants to play it down might say “barely over half.”
The study found that the language people choose when trying to persuade is also more slanted than the numbers they report. The authors write that people appear to use the imprecision of language in ways they find acceptable. Reporting a number wrongly would be plainly false, but it may feel more comfortable to slant a message with words.
De Chaisemartin said that “language messages are particularly persuasive.” He added that telling an audience they are being spun does little to reduce the effect.
Statistics are often reported in words rather than figures, de Chaisemartin noted. One study of media practices found that 22% of news items include statistics, but 24% of those stories use only words to describe the numbers. Another study of academic research found that 37% of papers include numbers in their abstracts.
For readers, the takeaway is to look for the underlying figure when a headline or report calls a number “substantial,” “significant” or “a clear majority.” De Chaisemartin said the findings bear on media literacy and could help people analyze the statistical claims they read in the news.
Context also matters. Because most drivers wear seat belts, a policy that raises the usage rate by two percentage points might not look large in absolute terms, but could be a substantial gain within the remaining room for improvement. In other settings, a two-point change could be a small fluctuation. Words can help put numbers in context, the authors note, though the findings suggest caution when interpreting them.
De Chaisemartin said organizations could benefit from consistent standards for how they describe data. He pointed to climate scientists, who already do this for some forecasts, and said security officials, medical researchers, doctors and scholars may have reason to adopt similar standard characterizations of likelihoods and other quantities.
He is also interested in what happens when people send information to the same audience repeatedly. In that case, he said, senders might have added incentives to protect their reputations for accuracy, while audiences could become more discerning over time.
Around Prague, based on reporting by MIT News