Bad Science
08/04/09 13:14
What skills should we be teaching students now to help them understand the impact - or lack of it - of statistics? There was an article in the Herald recently which showed that although murder rates have dropped (from 21 per million people per annum in 1985-92 to 12 per million per annum in 2004-2008), people still think that crime is getting worse. Perhaps in this case, the gap between reality and perception is partly due to an aging population; the older you get the more often you've heard of violent crime and horrible murders, so it feels like the murder rate has increased, even if the opposite is actually true.
But there's also the challenge of how to equip students to understand statistics, and be aware of when they can be misused. A classic example can be found in Ben Goldacre's book - 'Bad Science’ (have a look at his site - it’s a great source of stories on bad statistics and dodgy scientific claims). In the book, he sites an infamous trial in 1999 when solicitor Sally Clark was put on trial for murdering her two babies. The prosecution case rested on a statistic that claimed, the chance of two babies in the same family dieing of Sudden Infant Death Syndrome (SID) was 'one in seventy three million'. Sally Clark was convicted on that statistic. Unfortunately for her the defence team failed to point out that it was equally statistically rare for two babies in the same family to be murdered by their mother. And they failed to explain to the jury that the figure 'one in seventy three million' was dodgy anyway because it was just the statistic showing the chance of one baby dieing of SID - doubled; 8,543 x 8,543, which failed to take into account environmental or genetic factors that may have been shared by both babies.
Perhaps these test cases should be taught in schools.
|