Sample Newsletter February 2005

Beware of statistics

A cautionary word or two.

How often have we been persuaded by statistics to support an argument but how often have we actually questioned the basis on which the statistics have been collected or sourced?

We have probably all heard the old chestnut in which it is quoted that 90% of children suffering appendicitis have eaten ice cream therefore it must be ice cream that  causes appendicitis.  We all know that this is complete nonsense, however, it is not uncommon for unrelated data to be used to prove a point.  One might say that obesity or coronary heart disease is largely due to our sedentary lifestyle, and this can be supported by plotting the number of T.V. licences issued and the rise in deaths from coronary heart disease on a year by year basis.  The graph, as printed below, demonstrates this conclusively but is it true?  The dotted line plotting the number of T.V. licences issued and the solid line recording the number of deaths are clearly closely associated, so far as the statistics are concerned.

There was a case reported where a much respected member of the medical profession wished to show from studies he had made, that milk and dairy products caused an increased risk of heart disease.  He took data from studies made in 23 countries but only that from six of the countries supported the case he was trying to make, so he skilfully ignored the other 17.  The report which was published, bore his name, so because of his position within the profession, readers were persuaded to accept the recommendations he made.  They were actually statistically biased and obviously incorrect.

Can we trust statistics?

Reference

“Pure White and Deadly” by John Yudkin - Pub: Penguin Books Ltd 1986