The mathematics of rioting

BBC Radio 4’s More or Less programme, presented by Tim Harford, is often good at investigating statistical claims. I was therefore particularly interested in what it had to say about Britain recent rioting. According to the most recent episode:

• It confirmed that, even adjusting for (CPI) inflation, average monthly public spending has risen slightly over the past year.
• However, it said that over the past six months there has been a “distinct decline” in spending.
• It pointed out that it is necessary to divide public spending into different categories. For instance, there is departmental spending (defence, the National Health Service, police, etc) and annually managed expenditure (interest on the national debt, pensions, unemployment benefit etc). These are roughly the same size but departmental spending is being cut while annually managed expenditure is rising.
• As mentioned in my 11 August post Hariett Harman is suggesting that the riots  essentially happened in anticipation of future cuts.
• Youth services in Haringey, the London borough which includes Tottenham, have already been cut substantially.

The programme makes the always crucial statistical point that correlation does not equal causation. So, for example, if riots coincide with the implementation of spending cuts it does not prove that austerity was the cause of the rioting.

However, with that key caveat in mind I would have taken less seriously the recent academic study of budget cuts and social unrest in Europe from 1919-2009. The political context changed so much over those 90 years that it would be wrong to draw sweeping conclusions – at least beyond the banal observation that people dislike austerity.

For similar reasons I would be extremely wary of studies that link riots to a bulging youth population. For example, Jack Goldstone, an American academic, pointed to the rising young population in Tottenham. Even if there is some kind of link it does not show that demography was the cause of rioting.