Mail compares apples with oranges, comes up with bananas

Posted by Jamie Sport of Daily Mail Watch:

Journalism and statistics go together like Dog the Bounty Hunter on a dinner date with Tolstoy.

Usually, statistics in the Mail come from some press release sent out by a company with a vested interest (if the story’s science related), from a ‘report’ (by the TaxPayers’ Alliance), or from an NGO, quango, or think-tank (if the figures suggest a rise in crime, for example). The figures are often based on Government figures which have been analysed, edited, skewed, and reinterpreted. Such data are processed and packaged into an easy-to-understand, journalist friendly document by one of the third parties mentioned above, that tells the hack everything they need to know, like this: ‘X has gone up by Y, meaning Z’. The journo, who is thankful that they haven’t had to wade through boring old figures themselves, will then pad their stats based article out with quotes and additional information to establish context – or, in most cases, to completely mislead the reader.

Sometimes, though, an ambitious journalist will tire of rewriting pre-compiled reports and studies and decide to go and look at the statistics for themselves. This is a risky thing to do because the journo is well aware of their lack of training in stats and the potential for time-consuming redrafts if they make a mistake. On the plus side, it makes it look like they’re actually doing some research and deserve to get paid. Luckily, Mail hacks don’t have to worry too much about errors because, should they make an appalling mess of things, nobody will actually notice (or care).

Such is the case with Sue Reid’s ‘SPECIAL INVESTIGATION’ on migrant workers and unemployment in today’s Mail, headlined ‘Revealed: The areas where there are more migrants chasing jobs than locals‘. Sue seems quite proud of her data-mining, as there’s a little photo of her looking pleased with herself next to the words ‘SPECIAL INVESTIGATION’. No expense has been spared on art direction either; there’s a picture of a Romanian builder photoshopped into an image of Britain split up into different coloured areas to indicate the number of foreign people looking for jobs in each district. There’s a long column of statistics, and even a pie chart.
The piece begins proudly,

The true extent of the huge influx of foreign workers into Britain is revealed in an investigation by the Daily Mail.

In a line that wouldn’t be out of place in a BNP pamphlet, it adds,

The figure[s] expose as a sham the New Labour pledge of ‘British jobs for British workers.

Sue helpfully explains the methods behind her SPECIAL INVESTIGATION and where she got her numbers from:

[The article] is based on information from each local authority based on two sets of official figures.
The first is the total in each area of National Insurance Numbers given to adult overseas nationals entering the UK during 2008.
The second set of figures is the claimant count for each local authority area in July, compiled from Government statistics released last week.
A claimant is a person on job-seekers’ allowance who is actively trying to find employment. Newly arrived foreigners cannot get this payout.

Unfortunately for Sue, her methodology is catastrophically flawed. She has taken the cumulative total number of National Insurance number (NINo) registrations for the entire financial year 2007-08, and compared it to the number of people claiming Jobseekers’ Allowance (JSA) in the single month of July 2009. Unsurprisingly, this has thrown up figures such as Edinburgh where supposedly the 10,022 ‘local jobseekers’ are outnumbered by 12,450 ‘new migrant workers’.

Basically, Sue has found that there were more foreign people looking for jobs in a 12 month period than there were local people looking for jobs in one month, which is hardly surprising is it? And that’s ignoring the fact that using figures from two different financial years, which were experiencing vastly different economic climates, is somewhat questionable.

Her second failure is to compare a cumulative, stable figure with an average, changing figure. She tells us that in 2008 there were 733,090 new NINos given to migrants, the number she uses to compare against the number of JSA claimants in July 2009. This 733,090 includes everyone given an NI number between April 2007 and March 2008, many of whom, obviously, will already have found work and therefore will not be competing with the locals looking for work in July 09 – a whole two years later. While NINo registrants will have been entering into work during that period, thereby removing themselves from the fluctuating pool of people competing for jobs, many of the JSA claimants in any given month will the same people who were claiming the month before, and, chances are, the month afterwards. It is quite clear that comparing the two statistics is completely and utterly redundant; you might as well compare the number of motorbike accidents in 1972 with global temperature increases during 1990-1995. The relationship is meaningless.

Let’s be fair to Sue, because I can see what she was trying to do, and the Office for National Statistics website is a bit confusing. Let’s say the comparison between NINos and claimants is valid, and let’s assume that not a single new NINo registrant managed to find a job during Apr-Jul 07 (the first quarter of the 2008 financial year).

In that period, the national total number of NINos was 166,133. The average national number of JSA claimants over that same quarter was 887,757*, meaning that, actually, there are five times as many ‘local people’ looking for work than there are migrant workers. If we look at the latest period for which data is available (Oct-Dec 08), the ratio of local workers to migrants actually increases to 6:1, and the number of NINos granted to foreign workers decreases to 134,800. Is this the influx mentioned at the beginning?

How about the particular regions in which migrants supposedly outnumber local jobseekers? (I acknowledge that I’m taking a rather liberal attitude to statistics at this point, but, when you’re forced to compare apples with oranges, somethings got to give.) One of the ‘worst’ named areas in the article is Brent, where the Mail tells us that 19,240 migrant workers were given a NINo in 2008. In that same year, there were, on average, 6,647 JSA claimants each month. If we divide that 19,240 total NINo figure by 12 we find the average number of new migrant workers in any one month – 1,603. Looking at an average month in isolation and assuming that the previous months new NINo registrants and jobseekers all found jobs, that means there are actually four times as many local jobseekers than migrants.

Even using her own massively flawed methodology, it’s abundantly clear that there are not more migrant workers looking for jobs than British people doing the same. Sue Reid is the blacksmith of statistics, bashing blindly away at data until it transforms into something else, unrecognisable from the original materials.

The question is why, when migrant jobseeker numbers are actually falling, does The Mail want people to think they’re rising?

** Update from Liberal Conspiracy:

I’m not linking to it, but Sue Reid’s “investigation” is now a fully fledged press release on the frontpage of the BNP’s website, titled “More Immigrant Jobseekers than Indigenous Brits”, copied and pasted from the original in the Mail.

There couldn’t be a better argument for press responsibility.