I was lucky enough last week to go to a guest lecture called ‘Playing With Numbers’, about data manipulation in the press. Given by Dr.
I was lucky enough last week to go to a guest lecture called ‘Playing With Numbers’, about data manipulation in the press. Given by Dr. Sam Mugfold and Professor David Spiegelhalter, the lecture was both entertaining and interesting – I learnt a lot. I thought I’d share some of the salient points with you, so you can spot misinterpretation and manipulation in a piece of journalism in a heartbeat. Here are some things to consider while you’re reading…
1) Think about the journalist’s primary source (usually the person or organisation most quoted in the article), because it can skew the article’s bias. An article stating that ‘dried fruit is just as healthy as fresh’ citing the World Nut and Dried Fruit Congress a number of times…doesn’t take a genius does it?
2) Journalists are a negative bunch and will ‘flip’ statistics to the more pessimistic angle. For example, scientists found that ‘10% of people have a gene that decreases blood pressure’. ‘So hooray for those 10%!’ you cry. A journalist will flip that. ‘90% of people have gene that increases blood pressure’ – ‘OH NO!’ you sob. So root out the bright side.
3) ‘It’s easy to prove something is dangerous but almost impossible to prove something is safe’ – a great quote from Dr. Sam which can be applied to the GM crop debate. There’s always the unknown. Which means it’s more common to see reports of why stuff is 100% dangerous than why it’s 100% safe.
4) Scientists DON’T know all the answers. Their responses to direct questions are hardly ever as simple as a yes or no. Take this official stance on climate change: ‘the balance of evidence suggests that there is a discernible influence [of CO2 emissions] on global climate’. That’s as close as scientists can come to saying CO2 emissions affect the climate, because a control experiment with a second Earth is impossible.
5) Scientists are a balanced source. If they’ve done their experiment well, they’ve considered the alternatives and followed up their initial investigation. This means that pitting them against the mother of a child who had the MMR jab and is now severely autistic actually makes the report one-sided again, not balanced…
6) The MMR/autism case is a classic example of inflation and manipulation. The Wakefield study was leapt on because it linked the MMR jab with a higher incidence of autism. It only had 12 participants, no control group and no follow up investigation since has sided with Wakefield’s findings. The paper has now been retracted and yet there are STILL people wary of the MMR jab. Press Power.
7) A favourite amongst journalists is the ‘cats cause cancer’ story, or ‘the mix of the mundane and the dread’. The Daily Mail is always good for this- just look at its list of mundane items that have links to cancer. Here’s a second example in The Independent (http://www.independent.co.uk/life-style/health-and-families/health-news/fizzy-drinks-may-lead-to-teenage-violence-2375523.html). They take something we dread, teenage violence, and something mundane, fizzy drinks, and hey presto! A snappy headline, when the results are much less conclusive.
8) Even the most statistically significant results can have happened by chance, like Paul the Octopus. But this suggestion often gets omitted, and the line between correlation (there’s a correlation between the number of shark attacks and ice cream sales) and causality (ice cream sales cause shark attacks) becomes blurred. (FYI: both happen more in the summer months- because more people hit the beach…)
9) ‘Numbers do NOT speak for themselves’. Journalists should be weighing up the relative and the absolute risk in their reports, but often don’t quite get round to it. A ‘20% increase in people who get pancreatic cancer’ headline sounds a lot more scary than ‘pancreatic cancer up 0.2%’- pancreatic cancer sufferers make up around 1% of the British population. 1% of not very much remains not very much. So percentages are easily sensationalised- by using the relative, not absolute, figure.
10) I’ve left the most poignant quote of the lecture until last. Prof. Spiegelhalter called Fox news graphics ‘a complete fiddle’- and they are. Pie charts adding up to over 100, and the now famous ‘Fox Unemployment Figures’ chart (http://www.washingtonpost.com/blogs/erik-wemple/post/fox-newss-unemployment-chart-better-graphics/2011/12/12/gIQAUVgNqO_blog.html) are examples. This chart has been spread wide over the internet because it demonstrates dishonesty and inaccuracy- its trend line has been altered. In an attempt to imply that the unemployment rate remains high in November 2011, the scale of the chart suddenly changes to equate 9% (the October figure) with 8.6% (the November figure), in a thinly veiled attack on Obama’s employment policy. Outrageous.
I hope after reading this list you have learned and absorbed that you shouldn’t believe everything you read when it comes to statistics – sensationalism and manipulation are part of today’s media. But it’s great to be able to read something and pick out the holes. A weird feeling of secret power comes over you and you feel very intellectual.
You’re welcome.
