Death Metal Music Inspires Joy Not Violence, Study Finds (bbc.com)

An anonymous reader quotes a report from the BBC: I've had one desire since I was born; to see my body ripped and torn. The lyrics of death metal band Bloodbath's cannibalism-themed track, Eaten, do not leave much to the imagination. But neither this song -- nor the gruesome lyrics of others of the genre -- inspire violence. That is the conclusion of Macquarie University's music lab, which used the track in a psychological test. It revealed that death metal fans are not "desensitized" to violent imagery. The findings are published in the Royal Society journal Open Science. How do scientists test people's sensitivity to violence? With a classic psychological experiment that probes people's subconscious responses; and by recruiting death metal fans to take part. The test involved asking 32 fans and 48 non-fans listen to death metal or to pop whilst looking at some pretty unpleasant images.

Lead researcher Yanan Sun explained that the aim of the experiment was to measure how much participants' brains noticed violent scenes, and to compare how their sensitivity was affected by the musical accompaniment. To test the impact of different types of music, they also used a track they deemed to be the opposite of Eaten. "We used 'Happy' by Pharrell Williams as a [comparison]," said Dr Sun. Each participant was played Happy or Eaten through headphones, while they were shown a pair of images -- one to each eye. One image showed a violent scene, such as someone being attacked in a street. The other showed something innocuous -- a group of people walking down that same street, for example. "If fans of violent music were desensitized to violence, which is what a lot of parent groups, religious groups and censorship boards are worried about, then they wouldn't show this same bias. "But the fans showed the very same bias towards processing these violent images as those who were not fans of this music."

Comments

Re:"The test involved asking 32 fans and 48 non-fa

[hey!]

That is what statistics is for. Mathematics can tell you whether 80 is a sufficient sample size better than intuition.

Imagine instead of 80 subjects, you had a million. And suppose you managed to falsify the null hypothesis, and showed that death metal fans *do* have higher rates of desensitization than non-fans. One of two cases holds; either (a) the rate difference is too tiny to care about or (b) you could achieve a statistically significant positive result with a smaller sample size.

For this reason most well-designed social science experiments have moderate sample sizes. Experiments with a moderate number of subjects are affordable, practical, and are biased to false negatives; that means you are less likely to get statistically significant but practically insignificant results. Typical sample sizes (when they can be gotten) are in the 20-50 range. 80 is on the high end, but a *negative* result from a largish sample size is actually pretty robust. Either the differences between fans is non-existent, or it's very small, which is practically speaking the same thing.

Re:"The test involved asking 32 fans and 48 non-fa

[hey!]

You're a complete fraud if you think 80 models millions in a psychological human-reaction experimentation.

Minimum sample size doesn't increase linearly with population size; it asymptotically approaches a fixed value. So what you do is assume the population is arbitrarily large and size your sample accordingly. Yes, for very small populations, say hundreds, you could get away with smaller samples. But the sample size you need for a population of a million and a hundred billion aren't different at all.

The minimum sample size is *extremely* sensitive to effect size. So what you do is look up the minimum size in a table indexed by the smallest effect size you want to detect. Even if the population of the Earth has doubled since the time the table was published, the numbers are still good.