Don’t think percentages, think real people

We see risk statistics in news articles every day. “One extra unit of alcohol a day increases a woman’s risk of breast cancer by 12%.” “Bacon increases your risk of colorectal cancer by 20%.” But what do those numbers really mean? (Beyond wide-eyed alarmism from the news media, I mean.) Michael Blastland (BBC) suggests converting these risk percentages to the number of additional (or fewer) cases amongst a static population. In his Risk-o-Meter, the statement “Bacon increases your risk of colorectal cancer by 20%” is converted to “about one extra case in every 100 people”. It makes sense: it’s really…

We see risk statistics in news articles every day. "One extra unit of alcohol a day increases a woman's risk of breast cancer by 12%." "Bacon increases your risk of colorectal cancer by 20%." But what do those numbers really mean? (Beyond wide-eyed alarmism from the news media, I mean.) Michael Blastland (BBC) suggests converting these risk percentages to the number of additional (or fewer) cases amongst a static population.  In his Risk-o-Meter, the statement "Bacon increases your risk of colorectal cancer by 20%" is converted to "about one extra case in every 100 people".  It makes sense: it's really difficult to make sense of these risk numbers without knowing the prevalence of the disease (…, event, whatever) in the first place.