THE GOVERNMENT has put £1.5million up for another round of lie detector test pilots for social security helplines run by local authorities in the UK.

But it has drawn criticism for failing to prove that the last batch of pilots were a success. This comes on top of the criticism for having stuck lie detector machines on its phone lines in the first place.

The DWP said today in a statement that "initial results" from the seven pilots it has conducted across seven local authorities had been "successful". It provided no other details but said the results justified another round of pilot projects with another fifteen local authorities.

Danny Alexander, the Liberal Democrat work and pensions shadow, said the government should publish the full results of the first round of pilots before forking out any more money.

He also said more benefits funding was lost to government error last year than to the benefits cheats the government wants to catch with its lie detector software. According to the Libdems, £530 million of benefits money was lost to government error in 05/06, while £480 million was lost to fraud.

The figures did not specify how much of the fraud was derived internally and how much was attributed to "benefits cheats", but it nevertheless showed that the DWP should sort its own act out first.

The software, licensed through Capita by Digilog UK, which bought the UK licence from Nemesysco, an Israeli software developer, attaches risk scores to people after analysing their voices on the telephone. It will alert a call handler with a "pip" and an on-screen assessment if it thinks it has detected a "high-risk" person. According to Capita, half the people it assesses as high risk turn out to be no risk at all.

Harrow Council said today it had saved an estimated £420,000 in the year since it first installed the software. This was derived from a DWP estimate of the amount a person would typically claim on benefit.

The figure was calculated on the assumption that 132 people who refused to complete the voice-risk analysis assessment would otherwise have tried to cheat the system; and that 500 people who, though they had been flagged as low risk, had declared their personal circumstances had changed and no longer needed benefits would also have otherwise attempted to cheat the system.

Out of 1559 benefits claimants processed by the lie detection system, 118 were flagged as high risk. Just 24 of those, or 1.5 per cent, had their benefits decreased as a result of the intrusion.

The TUC protested last year that voice-risk analysis software like that being used in Harrow was more likely to finger "innocent but nervous" people as those who were genuinely trying to cheat the system.

"People with a particular regard for the truth can be the most stressed in these circumstances – lie detector tests have a tendency to pass people for whom deception is a way of life and fail those who are scrupulously honest," it said.

It is usual for such software to pick out innocent people for further questioning in over 95 per cent of cases. µ