While Andrew’s off recharging his batteries, I thought I’d draw your attention to an article on ‘curbing the media hyperbole when reporting on statistics’ that was in last week’s New Scientist – I know it’s a little late, but as it represents a lot of what we strive for in our science communications, I thought it was still worth a mention.
The article uses some great examples of food stories that have hit the headlines over the years, including: ‘drinking hot cups of tea leads to an eightfold increase in the risk of developing oesophageal cancer’ and the even more unlikely sounding ‘a quarter of a grapefruit a day increases breast cancer risk by 30% in post-menopausal women’, but the story closest to our hearts, is the one that had Andrew hailed as the ‘saviour of the bacon butty’ when he countered the ridiculous sounding news that ‘a daily bacon sandwich raises the likelihood of bowel cancer by 20%’. Although this might be an accurate figure, it is misleading and doesn’t tell us anything about the absolute risk. David Spiegelhalter of the University of Cambridge sums it up nicely: ‘For an average person, the chance of getting bowel cancer at some point in their life is around 5%’, so based on a relative risk of 20% increase in bowel cancer: ‘this translates to an absolute increase in risk from 5% to 6%’ – this means in real terms, the increased risk for the population is just 1%.
This just goes to show how providing a simple statistic can distort the true situation, and as a reader there’s not a lot you can do to interpret the numbers in any other way if you’re not provided with the facts – the only safe thing to do is treat every headline-grabbing statistic with caution. But this is in itself is problematic – people are now numb to the numerous health scares that fill the papers.
This provides a dilemma for the Agency – we know that applying numbers to something helps people understand a risk and we also know that the media love them to sell stories. So if we want our advice and the results of research to reach a wider audience we have to play the game. But what we try to do is provide realistic statistic with the hope that journalists won’t jump to their own conclusions. To achieve this, we rely on our top team of statisticians in the Agency, they are involved throughout our research projects – from the planning stages, ensuring sample sizes are representative and the study design is appropriate, and in the interpretation of the results so that we can be confident that the results are statistically robust.
So, although we’re not completely guilt-free of using statistics to tell a story, we do endeavour to ensure they are honest and not misleading. I am a 100% certain of that.