Statistical lying once again

Life expectancy for the wealthiest in England has caught up with most comparable countries in the western world since 1990, but the health of people in the poorest regions is lagging behind, according to a major study.
In 1990, men’s life expectancy in England was lagging behind that in many major western countries, including Canada, France, Norway, the Netherlands and Spain. By 2013, England was doing better than all those countries, with men living for an average of 79.5 years, a gain of 6.4 years. The gains for women were more modest but still good: the average national life expectancy increased by 4.4 years to 83.2 years), equalling or surpassing most of the 18 western nations including Australia, Norway, Canada and the US, except for Finland, Germany, Ireland, Luxembourg and Portugal.

But the data shows huge regional variations. The gap in life expectancy is not about geography but income.


England average life expectancy is pretty good then. But there’s socio-economic divergence in it.

OK, but what’s the socio-economic divergence in those other countries then?

8 thoughts on “Statistical lying once again”

  1. Ill-health affects income as well as life-expecxtancy. So the cause of a majority of the lower life expectancy is the cause, not the result, of lower income.
    If you’ve got a couple and is chronically ill and the other looks after him/her (so working fewer paid hours and more unpaid ones), they will both be relatively poor and the strain of being a carer will reduce life expectancy.
    “The gap in life expectancy is not about geography but income.” is a clever lie because income can affect life expectancy but in terms of aggregate number of years deviation from the mean, the causation is the other way.

  2. Well, no, dearieme, technically it isn’t a prediction. It’s a sort of summary of all the actual death rates by age for the year in question. The number-crunchers run an imaginary cohort of a thousand newly-borns through all those death rates in turn to get an average.

    English Wikipedia is hard to understand on this point, but click on the French one and it’s much clearer, for once. One of my translation clients, INED, is cited in the notes with a little video to explain it – en français, évidemment !

  3. “Well, no, dearieme, technically it isn’t a prediction”: why bother saying that and then promptly contradicting yourself?

    “The number-crunchers run an imaginary cohort of a thousand newly-borns through all those death rates “: but those death rates are mere guesses, since they concern the future which is, you’ll recall, unknowable. So it is a summary of the consequences lots of individual year predictions, so it is indeed a prediction itself. QE fucking D. Even bloody Wikipedia explains it.

  4. One might then strive to become rich, as they would likely live longer.


    It’s the only fair way.

  5. I believe the way life expectancy is counted differs between American and the rest of Europe is how to treat stillborn. American count them into the equation, no one else does.

  6. Calculating life expectancy for those who survive to birth seems a more useful thing to do than calculating life expectancy for babies that died in the womb. What purpose can the latter serve?

  7. @ dearieme
    It only looks like a prediction.
    What the “life expectancy at birth” does is estimate the mortality rate for each one-year-cohort (based on the previous census and subsequent mortality data – so all the numbers from 2001 to 2011 were wrong) in the latest year for which data has been compiled and calculate from that the average age of death if a cohort of new-borns experienced thaose mortality rates.
    The predictions are the “projected mortality rates” that assume that new-borns will have mortality rates at age n equal to current mortality rates minus n times the average annual improvement observed in the recent past.
    That’s probably as clear as mud, but it’s the best I can do at this time of night.

Leave a Reply

Your email address will not be published. Required fields are marked *