With no columns in the Guardian this week, there has been little Polly Toynbee output to factcheck. So I turn once again to Better or Worse? Has Labour Delivered?, the book Polly co-wrote with David Walker. The challenge is to open it to a random page, and try to find a factual inaccuracy.
On page 54 (my edition is a Bloomsbury paperback), Polly writes:
In 2003 the child of a father in the lowest social classes was twice as likely to die within a year of birth, five times more likely to die in a road traffic accident and fifteen times more likely to die in a house fire than those in the highest social class.
The sentence bears a truly striking resemblance to one from page 9 of a 2004 report from the Institute for Public Policy Research, written by Will Paxton and Mike Dixon (you can download a pdf version of the report from the Guardian's website here):
The list of disadvantages which are correlated with poverty is long, but to give just a few examples: in 2003, children of fathers in the lowest social class were twice as likely to die within one year of birth (ONS 2001), five times more likely to die in a traffic accident and 15 times more likely to die in a house fire than those from the highest social class (DoH 2003).
As I say, the similarities are startling, and for those interested in this sort of thing, the IPPR report is dated 2004; and Polly's book is dated 2005. I'm going to assume it is Polly who was handy with the copy 'n' paste.
But the differences are striking too -- for example, I'd always assumed the absence of sources from Polly's material was just laziness, as opposed to deliberate suppression.
And, as ever, going to the original sources is revealing.
The figure about children from social class V being twice as likely to die within one year of birth than those from social class I is taken from an ONS report called Childhood, infant and perinatal Mortality statistics - Review of the Registrar General on deaths in England and Wales, 2001 (pdf link here -- see Table 20) which was published in 2003, but actually refers to 2001.
By the time the 2003 figures had come out in 2005 (pdf link here -- again see Table 20), the ONS had moved from a five-class categorisation of social class to an nine-class one (counting 1.1 and 1.2 as two separate classes). As a result, the chances of dying before one in the top social class (1.1) was 0.32%, and the equivalent chance in the bottom social class (8) was 4%. A tenfold difference and not a twofold one.
As for the figure about dying in traffic accidents and in house fires, the report that Paxton and Dixon cite, Tackling health inequalities - 2002 cross-cutting review, was actually published on 19 November 2002, according to the page on the Department of Health website from which it can be downloaded (here). Therefore, the quality of its insights into causes of death in 2003 are likely to be limited. In fact, the closest it comes to sourcing its claim that "[r]esidential fire deaths for children are 15 times greater for children in social class V compared to those in social class I" is on page 48, where it makes clear that the data it is using cover 1989-92. Rather than 2003.