2020-03-14_New_Scientist

(Grace) #1

30 | New Scientist | 14 March 2020


Podcast
Risky Talk
Libsyn

“I’VE got a lower than average
risk of developing coronary heart
disease and Alzheimer’s and a
higher risk of getting psoriasis
and coeliac disease. And I have
got absolutely no idea what to
do with this information,” says
David Spiegelhalter, in Risky Talk,
a monthly podcast about statistics
that started this month.
For Spiegelhalter, an academic
who has spent years trying to
improve the public understanding
and image of statistics, this kind of
information illustrates why he has
had his work cut out – and why
genetic risk is a good topic for
the first episode.
“We are all bombarded by
misleading anecdotes and biased
information, whether it’s people
trying to sell us stuff or persuade
us to vote for them. But to make
vital, everyday decisions, from the
personal to the political, we need
a clear view of the best evidence
about risks and benefits,” he says.
A few years ago, Spiegelhalter
sent saliva to a consumer DNA
testing company, which resulted in
those perplexing findings. We can
all pay for such tests, but they are
also being rolled out on a large
scale in medical settings. The US
National Institutes of Health plans
to sequence a million genomes
and the National Health Service
in the UK has a 10-year plan to use
genetic sequencing more often.
The biggest problem isn’t the
stats per se, but knowing what to
do with them. So far, tests flag a
few genetic conditions where you
can reduce your risk by adjusting
your lifestyle or through medical
intervention. But in most cases,
there is little you can do, in which

Protect us from dodgy stats


From genetics to immigration, our lives are ruled by statistics, so they’d better
be good. Timothy Revell tries a podcast that sifts the factual from the flaky

case stats aren’t just unhelpful,
they can also induce anxiety.
Saskia Sanderson, a behavioural
scientist at University College
London and a panellist in this
episode, says she has found an
answer to this. When people are
properly informed about the
possible outcomes of testing,

they tend to be good at opting out
of taking a test if they think they
won’t be able to handle the news.
However, few doctors are
trained to explain the implications
of genetic testing, and an NHS
source told New Scientist last year
that there were only 200 genetic
counsellors in England.
In episode 2, Risky Talk moves
on to other hot topics: crime,
immigration and opinion polls.

Views Culture


LAURENCE DUTTON/GETTY IMAGES

is a result of combining this tiny
data set with other figures, such
as the number of visas awarded.
The UK Statistics Authority
awards a quality mark for statistics,
and it has recently removed one of
its marks from the immigration
figures. Migration is a political hot
potato, so whatever your political
leanings, shouldn’t getting better
figures be a no-brainer?
If only. “Statistics aren’t
just numbers, they are numbers
with a social life,” Humpherson
argues. They are sought by people,
governments and corporations
with agendas who want to appear
to use cold facts to justify their
arguments. Small wonder people
feel they can’t trust statistics, or
the experts who cite them.
In that light, Spiegelhalter’s
attempts to sift the good from the
bad and generally improve the
dialogue can only be welcome. ❚

This time the problem is a shock. It
isn’t that there is lots of good data
and no one to interpret it, more
that some key stats are, well, flaky.
Take immigration. Spiegelhalter
finds a statistic showing that
13,000 fewer people migrated to
the UK, but when he downloads
a relevant table from the UK’s
Office for National Statistics (ONS),
he finds that the margin of error is
plus or minus 73,000. “This seems
rather large,” he says.
It gets worse. The way these
numbers are generated is even
flakier. Ed Humpherson at
numbers watchdog the UK
Statistics Authority says that the
immigration estimate primarily
comes from an “intention survey”.
This is a questionnaire the ONS
uses for people entering and
leaving the UK, asking about their
trip, whether they intend to stay
long and so on. No one has to reply.
Millions of people pass through
the country during the year, but
the ONS bases its number on just
5000 responses. Its final estimate

Deciphering statistical
data is rarely a
straightforward task

“ Migration is a political
hot potato, so whatever
your political leanings,
shouldn’t getting better
figures be a no-brainer?”
Free download pdf