Black people are now eight times more likely to be stopped and searched by the police than white people new Home Office figures show.

Unfortunately, the figures are useless and show nothing either way, whether this figure is bad or not. I'd like very much to know whether the police are racist, but: The proportion of black people being stopped is going to be larger than the proportion in the country as a whole simply if there are greater concentrations of black people where the stops tend to be (I've done some back-of-envelope calculations that show prove this, just to convince myself.)

Now this is utterly crude, but look at ethnic population proportion broken down by Local Authority, and see that urban LAs tend to have a higher proportion than rural ones. Then look at the absolute number of stop-and-search by Police Authority, and see that there are disproportionately more in urban than rural again (more than the population difference).

Those two differences themselves mean that the proportion of black people stopped-and-searched will be greater than the national proportion. That's no way to make any kind of judgement about institutional racism, especially because the intuitive way to do the comparison is the incorrect one. And shame on the Home Office for releasing figures like this which will naturally be misinterpreted.

This is horrendous, by the way, doing order-of-magnitude calculations like this. If I was at work I'd have population, census and Police/Local Authority stats on tap, and I could do the proper figures. But I want to know whether the police are racist, and that requires proper measurement. And given I can show the given figures are inadequate even with crude measures, it means proper measurement needs proper statisticians calculating a proper expected figure so we can crack down on the truly bigoted. At the moment we have no idea who they are.

We need a number of things:

  • The black population as a proportion of all
  • The number of stops and searches
  • The total population

...and all of those are needed down to postcode district (eg, W6) or preferably postcode sector (eg, E2 9) level. From that we can calculate the expected number of black people stopped and searched. Then comparing against the actual figure, we have a reasonable chance of finding abberations.

For evidence of institutional rather than individual racism - that is, racism in the way the organistion makes decisions rather than how individual police officers choose who to stop - we need to look at the levels of crime (committed and estimated prevented, and measures of seriousness and public concern) as correlated against black population proportion. We would hope that the police as a whole aren't concentrating their stop-and-searches in predominantly black areas rather than high crime areas.

I sincerely hope the Home Office are calculating these figures now, otherwise there's no way we can know anything about police racism, and we're being lulled into a false sense of security by incorrect figures.

These are all symptoms by the way. This has nothing to do with what causes crime, and whether and why the crime level and causes of crime (perhaps: poor schools, no jobs, poor housing) coincide with the denser black population areas, and even whether that may be because of general public racism moving wealth away from the non-white population. Perhaps, perhaps not. But the statistics don't even attempt to get near that, so I'm not so bothered.

I've been looking at local statistics for the past two years, by the way, and I know what sort of figures are available and the statistical structure of the UK. This gives me gut feelings about how these numbers work. I have to say, I doubt it's possible for the true figures to even be gathered, and so we'll have no idea whether these Home Office figures are bad or not until something fairly big happens at the Office of National Statistics.