TAKE a look at the headlines, and it seems we are pretty certain about the state of the world. , , 鈥淪addam Hussein now has weapons of mass destruction鈥�.
Yet all these statements come with uncertainty attached. People often shy away from admitting this, be they politicians, experts or journalists expounding in the media, or doctors talking to patients. Perhaps they assume it will undermine people鈥檚 trust or make decisions harder. Yet making informed decisions also depends on knowing the unknowns.
Advertisement
In a paper in Royal Society Open Science, my colleagues and I have reviewed the evidence about how best to communicate uncertainty without putting off or wrong-footing an audience (). We suggest a checklist of questions communicators should ask to guide their approach.
First, are you dealing with an uncertain fact (summer Arctic ice cover has declined over the past decade), number (2226 tigers in India) or underlying hypothesis (eating bacon causes cancer)?
Second, where does the uncertainty come from: natural variation, measurement difficulty, limited knowledge or expert disagreement? (We set aside the future effects of randomness and chance.) The practical problems of counting India鈥檚 tigers, for example, may cast the precision of that number in a different light.
Third, is the uncertainty direct (specifically about the fact or number), indirect (about the quality of the underlying evidence) or a mixture of both? Conflating the two can sow confusion. Take the decision of the International Agency for Research on Cancer in 2015 to classify processed meat alongside cigarettes as 鈥渒nown carcinogens鈥�. This : the evidence says that both processed meat and cigarettes increase cancer risk.
It doesn鈥檛 mean that both increase cancer risk by the same amount, or that there is the same direct uncertainty surrounding that risk. Smoking increases cancer risk far more than eating processed meat 鈥� yet headlines such as were the result.
A final factor is who you are communicating uncertainty to and their relationship to what you are communicating. Are they experts or a lay audience? Is the topic potentially sensitive, as in medical situations? Is there an initial level of trust or distrust in the communicator?
Precision in uncertainty is key: words such as 鈥渓ikely鈥� or 鈥渦nlikely鈥� are interpreted very differently by different people. Numerical statements of uncertainty can add precision. But if we are talking about Arctic ice melt, say, uncertainty might be better presented by a over decades, or by those well versed in the issue .
Our initial studies suggest that, done well, communicating uncertainty needn鈥檛 undermine trust. A structured approach to communicating the 鈥渨hat鈥�, 鈥渨hy鈥� and 鈥渉ow big鈥� can allow us to express uncertainty confidently and unapologetically 鈥� and so help keep everyone in the know.
