Saturday, April 21, 2012

The Simpson paradox of mortality

I have just discovered a Simpson paradox in the mortality statistics for England & Wales. Well my annotations show clearly that I had noticed it at least five or six years ago, but that was before I had learned to call it by the name of Simpson.

I am looking at a copy of a table which I think comes from Population Trends – slap on the wrist for not making a note of that. Deaths by age & sex in England & Wales for various years since 1976 & some selected quarters ending with Q2 2006. Of course I have been defeated by the ONS website in my attempts to confirm this.

The overall mortality rate for males was higher than that for females right up to 1986 (118 compared with 114 per 10,000), though it had been narrowing.  By 1991 the rate was the same – 112 out of every ten  thousand persons died that year. After that the all-age mortality rate was consistently lower for males than it was for females.

Against a background of falling mortality, at all ages, the rates for men have fallen more quickly than have the rates for women. So in 1996 it was 107 for men, 110 for women and, by 2005, only 93 for men but 99 for women.

And yet the mortality rates at every age remained higher for the male of the species, most notably for those aged 20 to 24 who were more than twice as likely to die as a young woman of the same age. So,as my annotations show, I needed to check how population numbers & age structures by gender had changed. I never got round to that, whether because of better things to do or another defeat by the ONS website I am unable to say.

Links
Related posts