“Politicians twist science to suit policy” would be an unsurprising headline in the United States, given the rocky relationship between the Bush administration and the US science community. But the same message raised eyebrows on the other side of the Atlantic last week, when a UK parliament report suggested that politicians there also pick and choose scientific results that best suit their policies.

There is not the culture of using scientific evidence and research in the way the scientific community would understand it.

Britain's three successive Labour governments have made much of their desire to integrate research into policy-making; phrases such as “evidence-based policy” have become buzzwords for ministers. Science minister David Sainsbury, who left his position last week after eight years, also won the respect of researchers (see page 244). But the latest report from the House of Commons Select Committee on Science and Technology tarnishes that happy image.

The cross-party group of 11 MPs takes ministers to task for labelling policies “evidence-based” when no relevant research exists, and criticizes the civil service for its poor interpretation of research results. Perhaps most worrying, concludes Phil Willis, the Liberal Democrat member who chairs the committee, is the fact that government-commissioned studies regularly go unpublished when they conflict with a department's policy.

The report's most dramatic examples came from Tim Hope, a criminologist at the University of Keele, whom the Home Office commissioned to evaluate an initiative on reducing burglary. Hope told the committee that the Home Office ignored a result showing an increase in crime rates in one area, and focused solely on a second result that showed a drop in offences. The Home Office did not return Nature's request for comment.

Willis says that this and other examples — such as a 2005 anti-obesity drive developed in the absence of any evidence that it would work — show that the government lacks the academic ideal that all results must be aired, regardless of whether they fall as desired. “It's a level of scientific incompetence,” he says. “There is not the culture of using scientific evidence and research in the way the scientific community would understand it.”

The report recommends that the government should ensure that every department has its own chief science adviser, and should also establish a government scientific service charged with bringing scientists into government and securing proper career paths for them.

That is a laudable but old idea, notes Peter Cotgreave, director of the London-based lobby group the Campaign for Science and Engineering; a similar recommendation was made more than seven years ago, yet remains to be implemented. The government is likely to respond to the current proposals in the next few months.

Selective evidence: data on crime initiatives were ignored by the UK Home Office. Credit: J. FOX/ALAMY

The report also suggests that peer review by outside bodies, such as learned societies, could assess the degree to which the government's policies are based on evidence. The government may, however, baulk at the thought of having an outside body audit its performance. Some science-policy experts are also sceptical about the idea: “This would replace judgements currently being made by officials subject to democratic accountability with judgements made by those outside the process,” points out Roger Pielke Jr of the University of Colorado, Boulder.

Whatever the difficulties in Britain, political interference in science policy is far greater in the United States, says Willis. For instance, it is widely thought that David King, the UK chief science adviser, would be able to speak out if government science policy diverged sharply from the evidence, whereas few would expect such independence of US presidential science adviser John Marburger. But during King's six years in office, government policy and scientific thinking on high-profile issues such as climate change have not diverged substantially. An interesting test of his independence, and the government's commitment to evidence-based policy, will occur when they do.