The children's charity the NSPCC publishes regularly-cited research on the extent of child abuse and neglect in the United Kingdom.
Its 2013 report, How safe are our children? claims that nearly 800,000 children were victims of maltreatment by a parent, guardian or other adult in 2011. For every one child subject to a child protection plan or on a child protection register, the NSPCC estimates a further eight have experienced maltreatment.
Knowledge of the nature and extent of child abuse and neglect is a necessary spur for sustained and concerted action.
Should the NSPCC iself be in the business of compiling and publishing such data?
The question was posed this morning by The Guardian's David Walker, at an event on crime statistics hosted by the UK Statistics Authority (UKSA). The NSPCC, he argued, had an interest in raised public awareness of child abuse and neglect, not least of all because of the public fundraising work it did. While its work was in many respects laudable, it was hardly an impartial statistical body.
Lisa Harker from the NSPCC did not disagree. In the absence of a public body doing this work, she said, the NSPCC was the least worst option.
The implication of this exchange is clear. Organisations with a institutional interest in the results of a statistical exercise are not best placed to compile those statistics in the first place. This is the case regardless of internal checks, the rigour of external inspection and audit, or the quality and integrity of individuals working in these institutions.
So what about police recorded crime data? The police collect the data forming the basis of police recorded crime statistics. They also have a vested interest in recorded crime falling, for reasons explained by Tim Hope in his Dishonest politician's guide.
Given this, some police officers, some of the time, almost certainly fiddle the figures, as the Chief Inspector of Constabulary, Tom Winsor, told MPs in late 2013. In early 2014 the UK Statistics Authority concluded that police crime data for England and Wales could not be trusted.
Perhaps the work of collecting recorded crime data should be done by someone else? It was a point I put to the UKSA event this morning. No, said Tom Winsor. Too expensive to do it differently. External audit (by the Inspectorate of course) and better training of the police was the answer.
I'm usually sceptical of objections based on cost. Money can generally be found for priority projects, even in austere times. Continuing with a discredited approach, which has proven deeply resistant to genuine reform, does not inspire confidence. Its cost-effectiveness is also questionable.
One implication of continuing with the status quo is that the police can be trusted more than the NSPCC when it comes to the collation of statistics. This doesn't strike me as very plausible. But lack of plausibility is not normally a bar to the pursuit of dubious policies.
But with the Commissioner of City of London police Adrian Leppard telling the UKSA event that police recorded crime was 'dead in the water', change might just be possible.