Monday, 14 March 2016

Correlation, causality and accusations of witchcraft

A story in the news today was a classic example of the confusion of correlation and causality. Scientists are always banging on about this, and some people wonder how this apparent statistical nicety matters - yet it's the reason that over the centuries people have been accused of witchcraft.

Correlation is when two things happen in proximity - of time, space or both. Causality is when one causes the other. Because we understand the world through patterns, when there is correlation, we tend to assume causality - but without evidence, this is a mistake. In the case of witches, an old person might curse a farmer for not giving them some milk. Two days later, one of the farmer's cattle dies. Burn the witch!

You might think that we are beyond such thinking in the UK, but unless there is data that wasn't presented in this news story, we clearly aren't (I read it in the i newspaper, but I'm sure it was elsewhere too.) There has been a large rise in the number of students accessing counselling services in universities over the last few years. Over the same period, tuition fees have pretty much trebled. There's the correlation, but as yet we've no evidence (let alone proof of causality). So what are we given?

  • A spokesman from mental health charity Mind said tuition fees and student loan debt were 'major contributors' to the rise in students seeking mental health help. Evidence? None given - appears to be a pure assumption. He went onto say it was 'unlikely' that the rise reflected greater openness around mental health. But even if this is the case it doesn't mean there is only one other cause - it's not either/or.
  • An NUS spokesperson said 'The evidence is clear - the marketisation of education is having a huge impact on students' mental health' - what evidence, except correlation. Pure political witch hunting statement.
  • A psychologist commented 'Research has shown that financial difficulties, such as being unable to pay bills, has an impact on mental health in students.' Quite probably true, but totally irrelevant. Student loan repayments don't kick in until the end of the course when the student is earning over £21,000.
Let's be clear, I'd prefer it if we didn't have tuition fees, and mental health issues are very important. But by leaping from correlation to causality in this way, the story totally undermines its argument. There may be evidence of causality - but it certainly isn't presented. If the psychologist is right, it seems far more likely that the cause (over and above more reporting) is financial difficulties while a student - which has nothing to do with tuition fees. But without data, to make this kind of suggestion is an irresponsible act.

No comments:

Post a Comment