April 3, 2014 12:52 am

How to cheat a psychometric test

The plight of the Co-operative Bank and the fall of Paul Flowers, its chairman, who faces allegations of buying illegal drugs, including crack cocaine, poses several questions: why didn’t anyone see it coming? Was there something fundamentally flawed about his personality? And, if so, could these traits have been identified and the risk contained?

Mr Flowers, a former Methodist minister, excelled in a psychometric test – which assesses personality, numeracy and verbal skills – when he applied for the position three years ago. But his subsequent failings raise questions. Did Mr Flowers distort the scores by faking the right answers? And are psychometric tests accurate indicators of performance?

Experts in psychometric tests say that although the questionnaires have a growing role in key appointments, they should never be used on their own. John Rust, director of the University of Cambridge’s Psychometrics Centre, says: “Openness to experience is one of the big five scales in personality testing, and clearly one that Mr Flowers may have scored highly on. But you would need to follow that with an interview that asked exactly what type of experiences he is open to.”

Nevertheless, Prof Rust agrees that people can game the tests. If you know what the job you’re applying for is, “of course, you’ll try and fit your answers to match”, he says, adding: “If you’re applying for a job in telesales, you’re unlikely to tick the box that says you’re vulnerable to rejection.

“A psychometric test is just one snapshot, and can in many cases be played in much the same way as a job interview.”

Banx cartoon. An elderly couple talking about psychometric test©Banx

Although some tests are poorly designed, so that applicants can choose to answer questions in a way that gives a positive impression of themselves, this is not nearly as straightforward or as common as may appear at first.

Tomas Chamorro-Premuzic, professor of business psychology at University College London, says the only way really to fake a test is to be able to understand how the scoring works. “That is, to be able to relate each possible answer to the actual score it will generate,” he says.

As an example, he points to the statement: I have a great sense of humour. “If you take this at face value you will have an inaccurate test. The reality – gleaned from data analysis – is that people who are genuinely funny, as rated by others, rather than themselves, tend to answer no, while people who are not funny tend to answer yes,” he explains.

Likewise, tests should be designed so that the connection between the question and the attribute being measured is opaque. For example, the statements “I’m very creative” and “I take a different route to work everyday” both assess creativity but the latter does so much more discreetly.

The reality, however, says Prof Rust, is that surprisingly few people attempt to manipulate the tests because people are often “proud of who they are”.

Most introverts are proud of being introverts and think extroverts “boring and shallow”, he says.

Likewise people who are conscientious and pay attention to detail are “usually proud of the fact that they can’t adjust to a changed schedule”, he says. “Conversely, people who don’t pay attention to detail may think others are overly obsessive and time-wasting.”

Ultimately, the real measure of a test is whether it can predict behaviour, says Prof Chamorro-Premuzic. But even here, the academic research is surprisingly positive.

“The best tests assume that people will lie and take that into account, because faking is a sign of competence,” he says. As an example he points to the questions “I love everybody I meet” or “I am an honest person”. Although a positive answer may in both cases be disingenuous, it acts as an indication of intent, so tends to predict future sociability and honesty.

The problem, however, is that the greatest strengths can also become the greatest flaws when, for example, being shrewd flips over into being mistrustful, or being charming turns into manipulative behaviour.

“Most tests will pick up on a candidate’s dark side”, says Prof Chamorro-Premuzic. “But if you hire a charismatic leader, be prepared to put up with a narcissist.

“If you look at the profiles of most chief executives or chairmen in the corporate world, many score highly on the dark triad of narcissism, machiavellianism (being socially skilled but enjoying the manipulation of other people), and psychopathy (lacking empathy).”

The problem is not that people are faking the tests, he says, but that the tests are not being used and interpreted well enough to filter out the dark characters.

Despite the criticism, psychometrics is a booming business. The tests were pioneered by the War Office Selection board in the UK during the second world war, and their use then spread to the Civil Service and gradually into professions.

Since then a multimillion pound industry has emerged, with the advent of cheaper online tests over the past decade giving the market a fresh lease of life.

Employers’ enthusiasm for the practice can be explained by its use in cutting the cost of recruitment, a desire to have a greater depth of information about candidates and an increasing scepticism about qualifications. Making a poor hire is also an expensive and time-consuming process that businesses are keen to avoid.

Martin Reed, chief executive and chairman of Thomas International, which supplies tests in 60 countries, says you can cheat the tests by having someone else take them for you online. “But if the recruiter has called in someone who says they’re outgoing and communicative and you have this meek, mild, non-communicative person in front of you at interview, you’re not going to be convinced.”

Ultimately however, the tests are only as reliable as the people who take them. “Although people are fairly predictable, no test is 100 per cent accurate, because the tests are based on probabilities,” says Prof Chamorro-Premuzic.

“Even when they are well designed and people answer honestly, there will be some errors in the prediction. Humans are complex and have a much wider repertoire of potential behaviours than, say, a fish, squirrel or robot.” That was certainly the case for Mr Flowers.

Copyright The Financial Times Limited 2014. You may share using our article tools.
Please don't cut articles from FT.com and redistribute by email or post to the web.

SHARE THIS QUOTE