It’s easy to blame the media for sensational
stories, but sometimes the science is wrongDon’t be fooled by sensational, headline-grabbing statistics
P
The Last Word
oliticians, we all know, bend statistics to
breaking point, but we expect better of
scientists. After all, they’re focused on getting
to the truth, rather than getting publicity. Some
research that makes headlines does make me wonder,
though. Take the recent claim that heat-related deaths
in the UK will soar by over 250 per cent by 2050
because of climate change.
That seemed like typical media hype, so I checked
out the actual research paper. It was published in a
serious research journal, and its authors were from
respected UK public health institutes. And they did
indeed conclude that heat related deaths are ‘expected to
rise by around 257 per cent by the 2050s’.
So, amazingly, the media reports were accurate.
Yet after reading it, I discovered it was the research
paper itself that was rather misleading. That headline-
grabbing figure came from estimates of how climate
and population changes will affect temperature-related
death-rates over time. As the UK is expected to get
hotter, it’s pretty obvious the risk of heat-related death
will rise. But the study found that the death-rate due
to cold weather will fall. Oddly, its authors didn’t make
much of this – which is puzzling, as the fall was so big
it actually led to an overall decline in all temperature-
related death-rates.
So where did that huge hike in heat-related deaths
come from? Simple: by multiplying the death-rate
by the estimated numbers of vulnerable people in
the population by 2050. And as these are predicted
to soar, the end-result is a hefty rise in raw numbers
of just heat-related deaths – one that has nothing to do with
global warming, and
everything to do with
an ageing population.
Cynics might also
think it had something
to do with getting
media coverage. But
not everyone was
taken in. The Science
Media Centre, which helps journalists assess new research findings,
took a dim view of that misleading ‘257 per cent’ figure. So did
one of Britain’s most distinguished statisticians, Professor David
Spiegelhalter of Cambridge University, who said: “This kind of
presentation gives ammunition to those who say that the effects of
climate change are being exaggerated.” As someone who routinely
uses statistical methods, I share his frustration with such apparent
proof that “you can prove anything with statistics”.
Used with care and sophistication, statistics can extract amazing
insights from data. Alan Turing and his colleagues used statisticalmethods to break the Nazi codes during World War II. In 2012,
Nate Silver used similar methods to correctly forecast the outcome
of the US presidential vote in all 50 states. Yet statistics also have the
power to bamboozle, which is why I believe everyone should be
taught how to make sense of stats in school.
Most of us have learned to be wary of research based on anecdote,
or small samples of people, or animal studies. We should also be
sceptical about research highlighting scarily large relative risks. A
ten-fold increase of a piffling risk – like being struck by lightning - is
still a piffling risk.
But there are more subtle statistical traps we should look out
for – like claims based on absolute numbers rather than rates. Did
you know that people over the age of 70 are dying in huge numbers
compared to the 1950s? Some might think that’s a scandal. The real
scandal is that the media and politicians can’t – or won’t – see such
statistical silliness for what it is.ILLUSTRATOR: JAN VAN DER VEKEN ROBERT MATTHEWS is Visiting Reader in Science at Aston University, Birmingham“As the UK is expected
to get hotter, it’s pretty
obvious that the risk
of heat-related death
will rise”
It’s easy to blame the media for sensational
stories, but sometimes the science is wrong