News of the World shows us how not to use statistics

Thank god for churnalism. The News of the World yesterday published an article claiming to have found the most “workshy” neighbourhood in Britain, but of course that ended up locked behind its paywall. Luckily, The Daily Express and The Daily Mail have both churned out articles based on NotW‘s, so I don’t have to pay Murdoch to read it.

The community that they claim is most “workshy” is a small area called Cottsmeadow Estate in Birmingham. Before I even go into the statistics of this, I’d like to point out that having a lot of people on benefits does not mean an area is “workshy”. Perhaps it’s an area where a major employer recently went bust. Perhaps it’s an area with a lot of affordable, accessible housing perfect for disabled people. Or perhaps, as seems to be the case here, it’s an especially deprived area, which has been very hard hit by the recession. After all, the majority of people in the area receiving benefits are getting Jobseekers Allowance (and quite a few more people are receiving income support, which means they work part-time).

Anyway, the newspapers claim that 106 people of working age live on the estate, of whom 105 are on benefits. Population statistics for individual “census output areas” are only available by request, annoyingly, so I’ll have to take that on faith for now. However, the population data is just an estimate, not a robust census, and when you’re dealing with areas as small as 100 people (out of a population of 60 million), you’re bound to have quite a bit of error in there.

The newspapers claim to have tracked down the lone worker – the mind boggles over how they could possibly have gone about this (did they go from door to door asking people “do you have a job?”), especially since the data in question dates from June of last year – in employment terms, that’s rather stale. Now, if they had found the only person on the estate who wasn’t receiving some sort of benefit, they’d almost certainly have breached the Data Protection Act – giving the name of the only person who does not receive benefits is, in effect, exactly the same as revealing everyone else does. This is precisely why the DWP anonymises their data – they randomly round each figure up or down to a multiple of 5, so you can’t work out who is or isn’t on benefits by taking advantage of small numbers.

In this case, it looks like they’ve probably underestimated the population of the area. After all, according to the statistics, three months earlier there were 110 people receiving benefits in the area (code 00CNGP0059) – more than the estimated population! This seems to be the only reason to focus on such ridiculously small areas. The data is divided into “census output areas” – the smallest division that the Office for National Statistics uses, and therefore most error prone too. Looking at the ward Cottsmeadow Estate is in, Washwood Heath, there appears to be about 4,950 people receiving benefits out of a working age population of around 15,000. This is a sample almost 150 times larger than just Cottsmeadow Estate and a much fairer way to gauge the number of people receiving benefits in the area.

These articles, had they been written properly, could have carried an important message – some areas are more deprived than others, and we need to make sure that everyone has access to work. The way the Mail and the Express (and presumably NotW, but alas I don’t have a copy of the article) cover it however completely destroys any attempt at nuance, tarring whole neighbourhoods as being full of “workshy” “scroungers”, regardless of what the statistics and basic common sense say.

Advertisement