In summary, for those developing personal risk calculators:
1) Are you trying to change behaviour or simply inform people? It changes your communication approach.
UK opinion last summer:
2) What are you trying to communicate? The risk of catching it and dying from it, or the risk of dying from it if you catch it? Whichever it is, be very clear
3) Who are your audience? Public or health professionals?
UK public keen for this information in our surveys
Then our advice on how to communicate the numbers:
Percentages seem clearest format, but also convey the smallest feeling of risk. For balance maybe include also ‘x out of 1000’ , which gives a larger feeling of risk. Do not describe as ‘your risk’ – maybe ‘risk result’.
People liked ‘positive framing’ (how many people likely survive), but the numbers were rated slightly harder to understand. Probably stick with the familiar ‘likelihood of dying’.
A visual scale: despite the difficulties of a linear scale, it was more trusted over a log scale.
Numbers need context. People don’t normally think in numbers. When asked to put a number on risks of people dying from COVID-19 if they caught it, they over-estimate (and the effect of the number format is evident again). But they do know the major risk factors.
Because people seem to think in terms of ‘personas’ the most helpful context seemed to be giving the risks for different, well-described ‘personas’ (e.g. 'the risk for a 70yr old with no health conditions) along the scale (and explaining the top of the scale)
It was not so helpful to give people ‘other risks they face’ as comparators, or numbers other than an absolute risk (such as a percentile) as they tended to mistake that for the absolute risk.
Total n in studies was 5,520 plus 28 in-depth interviews with public and 7 GPs.
All our data and surveys are freely available here: osf.io/auf8h/
Some musings from me on risk perception/communication on BMJ blog today.
We throw numbers around when talking about health risks - they’re a way of precisely defining a concept. But it’s like communicating colour through hexademical codes or wavelengths… bit.ly/32RzXSL
Firstly, the numbers only make sense to those already very familiar with the arbitrary mapping of number to concept. A designer or physicist might instinctively bring to mind a colour when you say 6d46c4 or 350nm, but the rest of us don’t.
Secondly, although the number defines precisely what colour we’re looking at, it doesn’t define what we perceive.
The circles below are both the same colour, but the context makes a difference to perception.
We’ve got a comment in Nature today on why science needs good ‘evidence communication’ and not the typical rules of rhetoric, designed to persuade rather than inform: nature.com/articles/d4158…
Alongside the comment piece, which ironically sets out to persuade you to consider not be persuasive, are our more detailed thoughts on how to do this: media.nature.com/original/magaz…
Consider your own motivations and ethics – you may feel passionately about your work, but how far is it ok to seek to persuade? Compare a doctor talking to a patient, an expert witness presenting evidence in court, a researcher talking in the media…