Thursday 22 November 2012

Correlation vs Causation

I recently came across this BBC article, which is a good example of correlation (most probably) without causation, but disappointingly, I feel the journalism lets it down slightly.

To quickly summarise the article, the BBC presents a graph of annual chocolate consumption and Novel price winners per capita, which demonstrates a very strong correlation, strong enough that there is only a one in ten thousand chance that the correlation doesn't exist. The BBC then presents several quotes from scientists saying that they believe chocolate helped them, before finally, right at the end, does it discuss the prospect that this is an example of correlation without causation, which is a common occurrence in science.

In a nutshell, just because a graph shows a correlation between two things, you can't conclude that one causes the other. Wikipedia offers a nice example: a correlation might exist between the number of deaths by drowning, and the number of ice creams sold. So one could conclude that ice cream consumption causes drowning. Obviously, this is ridiculous, a more likely explanation is that ice cream is sold more in the summer, the same time of year as more people are likely to go swimming in the sea. The more people swimming, the higher number of deaths by drowning. The rest of the wikipedia page deals quite nicely with the other types of correlation vs. causation, so is worth a read for anyone not familiar with the concept.

Back to the BBC article, by placing evocative quotes around the edge of the article, and saving the key scientific point until the end of the article, there's a risk that people who don't read to the end come away with completely the wrong message. Reading just a snippet of the article paints completely the wrong picture, which people could take the wrong way. Whilst there is some evidence to suggest chocolate can improve concentration, helping in study, taking it to extremes is obviously going to cause problems. Too much chocolate most likely has health implications that will far outweigh any gains in concentration.

Basically, this demonstrates the importance of effective communication in science, something I don't think the BBC have really succeeded with here. Leaving the main point to the end creates a significant chance of misleading people skim-reading the article, or not getting to the end. Now on a topic such as this, the risks are minimal, but take an important scientific issue, like nuclear fuel, for example. Say you want to tell people about new ways of dealing with spent nuclear fuel. Writing an article about the dangers of spent nuclear fuel, and the health issues it can cause, before finally concluding with "but actually, we've solved this now", runs the risk of giving people the wrong idea, and turning more people against nuclear power. Now the issue is not just people eating chocolate for the wrong reasons, you're impacting on public opinion of energy policy, which can have far reaching implications.

As always, comments, criticism and corrections are greatly appreciated. Thanks for reading.

No comments:

Post a Comment