When the stellar line-up at Leveson this week was announced, it prompted a fair few jokes about these being good days to ‘bury bad news’. While far from being bad news, the announcements around Eric Pickles’ ‘troubled families’ scheme are likely to receive rather less scrutiny than they would in a week where the Westminster village isn’t fascinated by goings-on at the Leveson Inquiry.

The pretext for the round of interviews that Pickles did over the weekend and this morning was the news that every eligible council has signed up for this scheme, which will see local authorities paid £4,000 for every household  where lives are turned around.

However, there is one point about this policy that has as-yet gone unexplained. Who exactly are these families?

The communities secretary has repeatedly stated that there are 120,000 of them, and that they currently cost the country £9bn a year. Given that this is meant to be very targeted public expenditure with measurable results, the question of exactly how these families are selected is relevant to discussion of the merits of the policy.

According to government guidelines, a troubled family is one that meets five out of the following seven criteria: having a low income, no one in the family who is working, poor housing, parents who have no qualifications, where the mother has a mental health problem, one parent has a long-standing illness or disability, and where the family is unable to afford basics, including food and clothes.

All well and good, but how has the number of people meeting this criteria been arrived at? This morning on the Today programme, Pickles dismissed a question about this as “a rather silly, academic point”, and declined to provide an explanation of the statistic.

The clearest explanation of how this 120,000 figure came about that I’ve been able to find comes from Professor Ruth Levitas, speaking on an edition of the Radio 4 programme More or Less last month.

She explained: “The 120,000 is a rounding up of 117,000. It actually comes from a piece of secondary analysis of the 2004 Families and Children’s Study… In that study they found that two per cent of families had five of the seven characteristics…

"Normally speaking you would say that if this is a random sample, you would expect to say [the margin of error] is plus or minus three per cent – that’s the problem, there’s a range and the 200,000 is the mid-point of that range.”

So it’s a figure drawn from a piece of research that is eight years old, potentially with a margin of error larger than the number itself – doesn’t seem to me to be an academic point to question the figure, especially when taxpayers’ money is being spent on the people this eight-year-old research identifies. As Levitas emphasised, this is a very difficult area in which to collect data, but to dismiss attempts to understand it as “silly” isn’t helpful. When asked by More or Less for a statement, DCLG said that the 120,000 figures is an “indicative estimate based on available data” that comes “closest to what they wanted to capture”.

When pressed, Pickles revealed that practical data-gathering is now underway:

“For the first time we’re actually bearing down and getting their names and addresses, and that’s what the first part of the programme has been about… The substantive point is that the local authorities are bearing down and getting the names and addresses, identifying them and we’re agreeing a programme of action over three years, and we’re moving beyond that rather petty point of how many they are.”

While it’s good that research into the precise number of ‘troubled families’ is now going on, it does still feel a little like we’ve come at this problem back-to-front, designing the solution before we had up-to-date information on exactly what the problem was.

UPDATE 11/06/2012 16.39

The Department of Work and Pensions has been in touch to contribute the following:

"The 120,000 figure is a widely used and respected figure based on government research on families with multiple problems and is a number that is recognised by all 152 local authorities who have signed up to work with that number of families in their area."

A list of families broken down by council area they've sent me (not yet published online) puts the total number at 118,085 - slightly higher than the estimate in the 2004 research and a couple of thousand under the rounded figure of 120,000.

Tags: Eric Pickles, Troubled families