Wednesday, 15 May 2013

Is no stats worse than bad stats?

There's nothing easier than attacking the media for misusing statistics - but I am puzzled at the moment by a major story in which the news media are avoiding statistics altogether. This may be an even worse reality, because the misuse of statistics is usually accidental, where this suppression may be deliberate.

Yesterday's news was full of the outcome of the Oxford trial where seven men were found guilty of grooming and abusing young girls in a terrible fashion. It is notable that the BBC report says nothing about whether the culture of the seven might have influenced this behaviour, not even in a piece headlined 'Who were the abusers?' Last night, though, Channel 4 News bit the bullet that most are dancing around and asked if race, religion or culture could have had an influence. Here Jon Snow asks the Deputy Children's Commissioner the straight question (and this is why I love Channel 4 News) 'Is it race?' Here's the interview:



She responds equally bluntly 'No.' She tells us this is taking place across all parts of our community and in all ethnic groups. The suggestion is that the reason we only see primarily muslim offenders, mostly with Pakistan as a place of origin, is down to the way the media reports the stories, and the way cases have been brought to trial. There was, for example, she tells us, the Derby group, which was primarily white (I wanted to put a link to that, but the only Derby trial I can find details of is clearly not the one she is referring to.) That's useful. But it isn't enough. The question that needs to be asked, but wasn't, is what the statistics are.

The UK population is currently around 62 million of whom maybe 20 million fit into the broad sex/age bracket giving them the potential to commit these crimes. The equivalent numbers for UK muslims is around 2.7 million, giving around 900,000 potential perpetrators, and for those of Pakistani origin 1.2 million in total with maybe 400,000 in the right bracket.

Given these figures, if there is no influence from these factors, we would expect around 4.5% of perpetrators to be muslims and around 2% to be of Pakistani ethnicity. If the actual percentages are significantly more than these,  and with the proviso that to do the stats properly we would have to look at other factors to make sure there is not another hidden dominant influence that needs to be controlled for, we can reasonably draw the conclusion that there is influence from race, religion or culture. If, on the other hand, these percentages are roughly comparable with the distribution of actual offenders, there should be an outcry because the media and the police/courts are grossly distorting the facts.

Let's be clear - I don't have these statistics, so I can't say which is the case. But either way this would be an important fact that needed acting upon. By not giving any statistics, we are being deprived of the key element of this news story. Statistics matter.

No comments:

Post a Comment