Liberal Conspiracy briefing: Exposing the DWP’s lie detector testing


11:52 am - March 11th 2009

by Unity    


      Share on Tumblr

LC briefingHello LC readers I have a treat for you. Today we start publishing the first of our briefings – a document I’ve been working on for the past month or so. It’s not coming out all at once because there are some legal and other issues still to be resolved. But in coming days and weeks, more will be revealed.

Our focus is on the Department for Work and Pensions (DWP) and in particular its recent decision to use lie detection technology to catch out benefit claimants.

We think that this is not only unethical, but the technology itself is so prone to error as to be useless for the purpose for which is it supposedly intended. So why is the DWP spending over a million pounds promoting it across local authorities? Has it done research into its drawbacks and limitations? If yes, then why is it still using it?


The background

In December last year the Queen’s speech contained a proposal by the Dept of Work & Pensions (DWP) to make benefits claimants go through lie-detector tests. If found guilty they would lose benefits for a month.

Various media outlets have earlier been fed examples of local authorities using lie detection technology to tackle benefit fraud: this BBC article, this Times piece and in the Birmingham Post. But while some local authorities claim success in saving money – none of the coverage has questioned the scientific validity of the technology or its [lack of] reliability.

DWP plans

In December 2008 it was a revealed in the House of Commons that the ‘voice risk analysis’ system, that was piloted in Harrow, and in several other local authorities with large IT/business services contracts with Capita will be trialled in up to twenty-five local authorities over the next year. Funding for the expanded trial is being provided by the DWP and will, according to Kitty Ussher MP, Parliamentary Under-Secretary, include “independent social research to determine the impact on customers’ behaviour”.

The trial does not, however, appear to include any kind of independent scientific validation of the system itself.

According to academics, this ‘voice risk analysis’ system is sold in the United States as a package consisting of a $2,000 laptop, software and operator training for around $25,000 per system. Our own investigations suggest that the DWP are paying around £10,000 for each system it provides to a local authority and that the funding for this year will provide an additional 150 systems over an above those used in the first stage of the trial.

Our assessment of the costs of the system, to the taxpayer, have been confirmed by information obtained by MPs and Peers via Parliamentary questions, which shows that that the first stage of the DWP’s trial (2007/8) cost the taxpayer over £720,000 and that a further £1.5 million pounds has been allocated for the current phase of the trial. Nowhere in this has there been any mention of any funding for the ‘independent social research’ that the government claims will be part of the current trial.

We can also reveal that one of UK’s largest and, for some, most controversial players in the outsourced public sector business services sector, Capita, is behind these trials and that all the councils testing the system, to date, have outsourced IT and business services contracts with Capita.

Millions more will be spent if the trial is successful. There is a lot of public money at stake here and, based on our own investigations, a lots of ordinary people who could face serious hardship for no valid reason if this system is adopted by the DWP and rolled out across the entire benefits system. In the worst case, some people could lose their home as a result of being erroneous flagged as a ‘high risk’ claimant by a piece of technology that rightly belongs, with perpetual motion machines and cold fusion, in the realms of pseudoscience.

Does it work?

Harrow Council were quick enough to trumpet the sums of money they claim to have saved by piloting this system, which cost them a reported £63,000, but much less was made of the fact that of the 119 claimants identified by the system as presenting a risk of fraud, only 43 were found to have been incorrectly paid benefits, a ‘false positive’ rate of almost 64%.

Two out every three claimants investigated by Harrow Council as a result of their being flagged up by this supposedly ‘proven’ technology’ as a potential benefit cheat were found to be entirely legitimate claimants. Is that really an acceptable rate of ‘success’ for a system that claims to be carry out a sophistical mathematical analysis of an individual’s voice and accurately identify their mental and emotional state of mind?

We don’t think so…

Academics say it only works in the sense it tricks and/or intimidates benefit claimants into disclosing information they might otherwise have withheld, or into withdrawing their claims, outright.

Does the DWP even care whether the system works, or not, as long at it propels enough people out the benefits system?

Why are we doing this briefing?

A key issue is the huge sums involved in the trial, the software and, if successful, the cost of rolling it out across the country.

Using lie-detector tests has become a cornerstone to cabinet minister James Purnell’s attempts to reform and ‘modernise’ the DWP. But we are not convinced the department has adequately researched the technology that it is advocating.

The DWP actually claims, on its website, that this is proven technology – we disagree and we are confident that we can demonstrate that, in the absence of independent scientific validation, the system the DWP is ‘testing’ should be regarded as being nothing more than a hoax and no more a reliable method of assessing benefits claimants than flipping a coin or employing a tarot reader to process claims.

There are other implications too. The same technology is being used in the insurance industry to purportedly detect people who lie about their claims. If the technology doesn’t work however, then there should be no reason to use it. On the other hand, if the government and corporations buy into the view that it does work well, then it may come into widespread use, with potentially devastating consequences for people.

So…

We’ll tell you exactly why it doesn’t work in the way the DWP claims, what it actually does and how it does it.

We’ll explain how easily the technology can be manipulated to produce results ‘to order’ and how this may have played a major part in convincing insurance companies and the DWP to buy into this system.

We’ll also explain how any council in the UK could generate the same results as Harrow with nothing more than a couple of press releases and some minor alterations to the scripts used by its staff when taking telephone calls about benefits claims.

We’ll also explain why something as simple and commonplace as the common cold, asthma or heavy smoking could get you branded a ‘high risk’ claimant by the Councils using this technology.

And…

This may be the biggest story we’ve run to date. It’s certainly the biggest investigation – over four weeks of detailed research, discussions with scientists and academics and hunting down the evidence which shows that the DWP’s much vaunted ‘lie detection’ system is a complete waste of public money that stigmatises benefits claimants on what amounts, in practice, to a coin flip.

    Share on Tumblr   submit to reddit  


About the author
'Unity' is a regular contributor to Liberal Conspiracy. He also blogs at Ministry of Truth.
· Other posts by


Story Filed Under: Blog ,Crime ,DWP lie detectors ,e) Briefings ,Labour party ,Local Government ,Westminster

Sorry, the comment form is closed at this time.


Reader comments


Is this the software i read an article on recently? Israeli company and the program is a couple of hundred lines of visual basic being sold as a state-of-the-art application

Yes, SL, it is that system…

Great stuff Unity, glad that this sort of thing is being pushed in a more official way.

Do Ivory Towers not have the internet, or has no one told them how to use

SL, they don’t care about evidence, they care about perception. Hence the best way to stop this isn’t necessarily to convince them it doesn’t work (they must know this) but to convince them that it won’t be hard to let the public lose confidence in the whole thing.

I accidentally left out the IT , a bit like this application then.

Sorry, couldn’t resist.

8. Shatterface

I agree this is a massive issue: best of luck with this campaign.

The automatic assumption of guilt and dodgy science not only reinvents the relationship between government and public, it re-writes the relationship between government and reality.

I’d almost tempted to suggest Paxman use this technology on MP’s on Newsnight or fitted to the mic during Question Time but voice-stress analysis doesn’t work at all on sociopaths.

“The automatic assumption of guilt and dodgy science not only reinvents the relationship between government and public, it re-writes the relationship between government and reality.”

Nice one

Interesting comment on Jeremy Vine Show by an official from the DWP dealing with fraud. The DWP closed down local offices/changed the system. Therefore a DWP official in Manchester can be dealing with a false claimant in London. Consequently, there is no local knowledge, so fraud increased. The official said the best way of reducing fraud was to ensure suspected fraud was dealt with by local officials undertaking regular checks, for example every 6 months. The problem is that when offficials are hundreds of miles away from the suspected fraudster, it is not cost effective to travel to their home .

8. Shatterface . Interesting point . Technology was found to be useless when dealing with people from Eastern Europe who has been brought up under communism as lying was a matter of survival . People said whatever the Communist Party wanted to hear.

Once again Labour is trying to introduce unproven technoloy, rather than using the the well proven mark one eyeball conected to a brain with common sense based on practical experience.

I hope someone is keeping all the national newspapers in the loop on this one…

12. Shatterface

Charlie (10): dead right on local knowledge. The DWP also screwed up when they changed Jobseeker Direct to the taxi ranking system so that customers could have jobsearches done by staff with no knowledge of the local labour market.

13. the a&e charge nurse

Is it envisaged that DWP officials will be adopting a similar format to the Jeremy Kyle show when announcing results from the polygraph ?

I look forward to reading more about this. Also agree that the way to stop it is to demonstrate how easily the public will lose confidence in it. As much as it might kill us to make this argument, showing how its unreliability will implicate and penalise the deserving poor is the way to go, I think.

Is this the same company that is threatened libel proceedings against two scientists who wrote a paper on how crap the technology is?

Is this the same company that is threatened libel proceedings against two scientists who wrote a paper on how crap the technology is?

Looks like it.

Lets have lie detector tests for bankers and businessmen too.

After all , if you have got nothing to hide you have nothing to fear.

UKL:

Yes – same company behind the system, although when we get to the business deals and contracts piece then a much bigger corporate fish will enter the frame.

“We think that this is not only unethical, but the technology itself is so prone to error as to be useless for the purpose for which is it supposedly intended”

Huh? Why is it unethical to test if someone is lying when they make claims for financial assistance from the state? Obviously if the technology is unreliable then it shouldn’t even be anywhere near a benefits office but the ethics of lie-detection is a massive issue which you just skipped over on the assumption that you’re right without the need for a debate. How wonderfully open-minded.

20. the a&e charge nurse

‘Bankers and business men too’ – but isn’t that another worry, Sally ?

As Unity notes in the main item – today it may be a few so called benefit scroungers, a group held in low esteem by the public and hence easy to push around (at least that’s what the exponents of this approach are hoping for).

But once the PRINCIPLE has been accepted who will it be tomorrow strapped into the polygraph ?

Incidentally, i the technology is anything like the dreadful NHS computer system then nobody will be safe.

“But once the PRINCIPLE has been accepted who will it be tomorrow strapped into the polygraph ?”

I don’t think many private corporations would have any problem using this kind of thing on it’s staff. Just as long as senior management were exempt. Some private corporations are drug testing their employees now.

Don’t get me wrong I am not in favour of this sort of thing, but we only get an outcry when govt does it. If private firms loose data you never here Cameron making a fuss about it.

22. Shatterface

It’ll probably be fitted as standard on the iPod 3. Todays government technology is tomorrow’s toy. Once the technology is out there, you’ll have more and more uses for it whether it works or not. There should be laws against it’s widespread use as there are with unauthorised genetic testing.

Magnificent.

24. Shatterface

The down side of attacking lie detectors on the basis that they don’t work is that their supporters will simply look to improve them.

There is a principle here that fills me with more dread than omnipresent CCTV and the national database combined: that the contents of your head may become as open to surveillance as your body.

Once we allow the state to monitor what we think – even if only through our body’s responses – we are well and truly fucked.

Shatterface – I am not sure such a technology could even conceivably exist, as our understanding of bodily responses seems to suggest that their development are very fluid, contested and dependent on social and cultural construction. You can get plenty of ideas from body responses, but I don’t think you can really get at the content of what people are thinking. Not that there won’t be a constant market in people claiming they have such a technology.

“I don’t think many private corporations would have any problem using this kind of thing on it’s staff. Just as long as senior management were exempt. Some private corporations are drug testing their employees now.”

Some public organisations too, Sally. TfL for example.

Be careful with the maths on the coin-flip analogy. Even a 50% false positive would not be the same as a coin flip.

To use the coin flip analogy you’d need to put your numbers up against this, assuming heads on the coin means you’ve detected a cheat:

Chance of heads = 0.5
Chance of tails = 1 – 0.5 = 0.5

Chance of being a benefit cheat= a (don’t know what the real figure is – does anyone?)
Chance of not being a benefit cheat = 1-a

Correct positive rate = heads x cheat = 0.5 x a
False positive rate = heads x not a cheat = 0.5 x (1-a)

Correct negative rate = tails x not a cheat = 0.5 x (1-a)
False negative rate = tails x cheat = 0.5 x a

If the system correctly identifies a lot of negatives, and the positives are used to flag people who deserve further scrutiny rather than as a final answer, then the system might be useful for focusing resources.

28. Shatterface

Nick, I’m thinking about polygraphs that measure a while range of physiological responses rather than voice-stress analysis, and while they might only ever be efficient enough to distinguish between yes and no responses that does take us into a situation where the privacy of thought itself has been violated: you no longer hold a monopoly on the contents of your mind and once the state (or any other organisation) has joint ownership they then have a say in what thoughts you can have (through ‘re-education’ or whatever).

29. Shatterface

Just done a quick search of the apps available for my iPhone: there are 9 claiming to be lie detectors (or at least joke versions of such) so there’s definitely a market for this shit.

I should imagine they would be a great way to screw up a relationship so hopefully natural selection will weed the users out eventually.

Isn’t this the same technology used to ‘tell’ whether someone is in love with you? (Hmmm…sex and the military: the two great exploiters of new technology)

31. Shatterface

Phones already have a way of showing if someone loves you or not.

If they don’t call back they don’t love you.

32. Laurie Penny

*applause*
I think I love you.

19: Polygraph tests aren’t admissible evidence in a court because of doubts over accuracy so why use them for benefit claims? I do suppose that if someone is hauled into court over claims of benefit fraud, the case will be thrown out because of this legality.

Lie detectors to catch benefit claimants??? They’re not serious are they???

Next, Jack Straw will be saying we should do away with juries and pass criminal sentences based on polygraph results.

Unbelievable. Well done for exposing these buffoons!

35. Shatterface

I’m no lawyer but isn’t evidence discovered as a result of procedural irregularities tainted and therefore inadmissable?

In other words, if someone is found to have been working in reciept of benefits as a result of an investigation following suspicions raised by lie detector tests, and those tests are later thrown out, would the later evidence be thrown out too?

btw: The reference that links ‘lie detector’ software to sex is via The Yorkshire Ranter or here

And here’s the ‘Love Detector’ website.

Plus! The story’s on the front page of Technology Guardian in the dead-tree edition.

Does anyone remember that daytime TV show “Lie Detector” from quite a few years back? Maybe we should make politicians go on it!

“James Purnell, are you in fact a heartless, inhuman piece of shit?”

“Very interesting question there, you see-”

“Answer the question, Purnell!”

“No, I am a human being with a conscience!’

*LIE DETECTED*

Bravo to #31!


Reactions: Twitter, blogs
  1. Liberal Conspiracy

    New post: Liberal Conspiracy briefing: Exposing the DWP’s lie detector testing http://tinyurl.com/au6zrn

  2. Purnell’s Lie Detector - The Scientific Evidence | Ministry of Truth

    [...] Liberal Conspiracy briefing: Exposing the DWP’s lie detector testing, and [...]

  3. Purnell’s plans for alcoholics « The Bleeding Heart Show

    [...] I suppose you could ask them whether they’re alcoholics, but how many people who are ever answer affirmitively? If you deny your addiction to your friends, your family and even yourself, you’re hardly going to open up to someone you only see once a fortnight and who only knows you as a name on a computer. And before we start going down the lie detector test route, we should note that this has already been torn to shreds. [...]

  4. sunny hundal

    Bizarre Margaret Hodge is criticising govt for using private contractors to get ppl off benefits http://bit.ly/9LoCkU

  5. Nicola Jones

    RT @sunny_hundal: Bizarre Margaret Hodge is criticising govt for using private contractors to get ppl off benefits http://bit.ly/9LoCkU

  6. yorkierosie

    RT @sunny_hundal: Bizarre Margaret Hodge is criticising govt for using private contractors to get ppl off benefits http://bit.ly/9LoCkU

  7. Jae Kay

    RT @sunny_hundal: Bizarre Margaret Hodge is criticising govt for using private contractors to get ppl off benefits http://bit.ly/9LoCkU

  8. Simon

    Liberal conspiracy report DWP attempt to force benefit claimants into unreliable & stigmatising lie detector tests http://bit.ly/9LoCkU

  9. mark wright

    a worrying sign that the DWP will be using lie detectors on claiments http://is.gd/f8coM

  10. Paul Sandars

    RT @markwrightuk88: a worrying sign that the DWP will be using lie detectors on claiments http://is.gd/f8coM

  11. jennifer roberts

    RT @sunny_hundal: Bizarre Margaret Hodge is criticising govt for using private contractors to get ppl off benefits http://bit.ly/9LoCkU

  12. Tony Bovaird

    Lie detector test for benefits claimants generated a ‘false positive’ rate of almost 64% in pilot council – http://bit.ly/9LoCkU

  13. Mrs Blogs

    RT @sunny_hundal: Bizarre Margaret Hodge is criticising govt for using private contractors to get ppl off benefits http://bit.ly/9LoCkU

  14. Shed Dweller

    RT @tonybovaird: Lie detector test for benefits claimants generated a ‘false positive’ rate of almost 64% in pilot council – http://bit.ly/9LoCkU

  15. Derek Bryant

    RT @tonybovaird Lie detector test 4 benefits claimants generated ‘false positive’ rate of almost 64% in pilot council – http://bit.ly/9LoCkU

  16. Boris Watch

    @cathynewman Could I humbly suggest a bit of research on the previous polygraph pushers, Cathy? http://t.co/JFH7JRsx #dwp #spiv





Sorry, the comment form is closed at this time.