Saturday, 10 March 2012

Understanding 10 per cent

Who knows how to explain a drop in rates in laymen's terms?

Following Jill Pell's study, headlines (and Richard Simpson MSP) advised us that the rate of premature birth dropped 10 per cent because of the smoking ban.

How do the papers come up with a 10 per cent drop? On the face of it you could say that between 65 and 55 the difference is 10, but we are not talking percentages. The figures look at rates per thousand. If the rate had been measured in hundred thousands, the difference between 55 and 65 would mean something different again. With thousands, it appears to me as a lay person that the range is no larger than 1 per cent of the sample, and I can't understand what the fuss is about.


Junican said...

Damn it, Belinda!

To be clear, as regards premature births, one needs to know the number of births in total. I know that the the number of live births in Eng and Wales is some 700,000 per an. How many of those were so premature as to be 'dangerously' so? Without that knowledge, the figures mean nothing. It is only upon the idea of 'dangerously so' that there is any significance - a mere 'average' of 'birth dates after conception' is nonsense since the date of conception is, normally, uncertain, which makes mockery of her 'between 24 and 44 weeks gestation'. This is important because of the RR of 4.52%. That is, the uncertainty of conception date outweighs the certainty of prematurity.

The point also is that her figures do not matter one way or the other, 'pre' or 'post' maturity, because of the uncertainty of the date of conception, without knowing the total numbers involved.

Does that make sense?

Belinda said...