I'm very happy every time I hear that the BBC
radio four programme ‘More Or Less ‘ is returning to the airwaves. The
programme fact checks statistics in the news, from whether cutting the winter fuel payment to pensioners causes excess winter deaths (answer: possibly, but
it’s impossible to put a number on it), to whether South-Eastern Railways clears 50 million leaves from their line each year (answer: unverifiable, file
it in the ‘silly drawer’).
There are also year-round weekly podcasts with longer
interviews, coverage of new books on numbers and statistics, and lately fact
checking claims made by Donald Trump, which seems rather an easy target.
What’s
it like?
The programme is non-polemical, light and often jokey
in tone, and made up of varied voices. However, this does not stop it paying
attention to the complexity of statistics and the questions being grappled
with, showing how click-bait headlines often mislead (and occasionally don’t),
how correlation does not mean causation and the results you get depend on the
questions you ask and the factors you consider. Sounds pretty basic research
practice, but as the programme shows, researchers don’t always follow best
practice, and even if they do, journalists often don’t when reporting the research.
But it goes beyond fact checking; if a statistic is
shown to be false, the researchers try to establish the real answer, and if not
ask whether a real answer is actually possible. Along the way they dip into
statistical principles such as different kinds of averages and sample sizes.
Three
items
Here are three items from the programme which have
stuck in my mind:
Have about one in three women in the UK had an
abortion? (Answer: probably yes; 21.53 minutes into the programme).
Are black mothers five times more likely than white
mothers to die during childbirth in the UK? (Answer: they are about four times
more likely to die during pregnancy or up to 6 weeks afterwards. This is due to
underlying health conditions such as heart disease or diabetes; also to behaviour,
such as whether they attend antenatal classes. There is no evidence that it’s
due to race per se, or mothers being treated differently because of race. The
risk of dying during pregnancy or up to 6 weeks afterwards is very low for all women in the UK – about 60 women per year.
Can you divide 1 by 0? (Answer: it can’t be done; 22.25
into the programme; and the answer isn’t infinity; 14.15 into the following
week’s programme after a storm of replies).
It’s
very BBC, or should be
The programme hits all three targets of BBC founder
Lord Reith’s vision for the corporation: ‘inform, educate and entertain’. In
fact, I would like to see BBC content in general more like this – focused on
fact-finding and checking, resulting in a valuable archive of programmes,
educational not only in content but in investigative method. Fewer shock
headlines and more measured findings. Less impact short-term, more long-term.
I
would like them to look at…
And what statistical question would I like the
programme to investigate? Well, Professor Brian Cox stated on Radio 3’s Private Passions that worldwide, more is spent on peanuts than on science; and, perhaps
more verifiably, that more is spent on mobile phone ring tones than on
meteorite defence systems. Sounds suspicious. Get onto it, More Or Less!