Sunday, June 01, 2008

Leave the emos alone

I don't know anything of this story other than this article:
Around 100 teenagers have marched on the offices of a widely read British tabloid to protest at its suggestion that their favourite emo band, My Chemical Romance, encouraged suicide.

They say they object to Daily Mail's description of the US group as a "suicide cult band" in an article about a teenage student who hanged herself two weeks after she started listening to its music...

...The article that sparked the protest concerned a coroner's inquest into the death of Hannah Bond, 13, who hanged herself allegedly after starting to listen to the band's music and becoming obsessed with death.
I have to say I agree with the emos#. The implication is listening to music can be a causal agent for behaviour (in this case, suicide). This is an old argument, previously applied to rock and roll, punk, hip-hop, heavy metal, techno, grunge and now emo. There may be a correlation with behaviour and music, but correlation does not demonstrate causation.

I'm not worried about the actual facts of this sad story, or even if the Daily Mail made the claims as reported. Assuming the above is all correct we have a common example of the post hoc error (post hoc ergo propter hoc - after this therefore because of this).

The classic example is violent TV shows/movies/video games (entertainment) are correlated with violent behaviour. Therefore, some claim, violent entertainment causes violent behaviour. On it's own, correlation does not mean causation, it merely shows two variables may be linked. As such, there is good reason to investigate further.

Until further investigation is conducted, correlation does not show which direction the link (if there is one) goes. It also does not rule out the hypothesis that a third factor could be the causal agent of change for the two correlated variables.

For example, perhaps naturally violent people are attracted to violent entertainment. I.e., there is causation, but in the other direction. Or, perhaps there are no "naturally" violent people, but the social group they happen to be in promotes violence (this is the third factor). As such they are prone to acting more violently and being entertained by violence.

This is not to say violent entertainment isn't the cause (or rather, one of the causes) of violent behaviour, just that you have to show more than a correlation to substantiate this claim. Something along the lines of controlled experiments with half the randomised subjects watching Bambi, the other half watching Rambo, then getting them to give each other different strength electric shocks (measure who is the most prone to increase the voltage) or some other twisted experiment psychologists seem to enjoy creating.

Cause and effect with human behaviour is generally not unidirectional. For whatever it's worth, my bet would be, whilst violent entertainment or suicidal music is not the cause of correlated behaviour (but rather an effect of predispositions) it may reinforce and increase behaviour in the manner of a positive feedback loop.

________
#Just because I agree with them in this case doesn't mean I have any real time for people who sit around pretending to be depressed.