Sunday, April 21, 2002

Health scare stories inspired today's rant on science reporting. Here are some booby prize winners:
I found most of these links through the Social Issues Research Centre site, and others through google's cool new news-search feature.

What these (generally brief) articles share is an extremely narrow look at an extremely broad topic, shallow references to experts and science, and numbers (when present) are included without context. Some specific examples:

  • "Working while pregnant increases the risk of the potentially deadly condition pre-eclampsia by nearly five times, research suggests. " Bad use of numbers! Five times from what? How common is pre-eclampsia? How often does it result in death? And extremely poor use of context -- although the article's title implies that overall risk is increased, they don't support that. It could be that the increased risk of pre-eclampsia is offset by other factors. Or have they compared whether working while *not* pregnant increases risk?
  • The pay phone article is supported by an interview with one doctor.
  • The doorbell article manages to put 3 statistics in under 200 words, but poor context makes the figures useless. "Almost 50% said they had experienced an event within the two-hour period before the stroke which the doctors considered to be a potential trigger." What's an event? How many two-hour periods occurred with an "event" where a stroke didn't occur? Only the study that occasioned the news article was mentioned, without any reference to other studies on the causes of strokes. No overall assessment was presented in the article.
It's not just the health scare stories that suffer from these problems but they're the ones that offend me most, because I know too many people who restrict their own lives due to the cumulative effect of these scares. So here's my proposed title for a news article: "Health scare stories reduce quality of life."

No comments:

Blog Archive

Creative Commons License
This work is licensed under a Creative Commons Attribution 3.0 Unported License.