According to a company which publishes the tests, professionals with only basic training in the procedures consistently out-perform their more highly trained colleagues when predicting future job performance from test results. Non-experts were up to 10 per cent better than experts, which in statistical terms was 'massively significant'.
Specialists in the so-called 'psychometric tests' - often used when employers are selecting graduates for employment - are guilty of 'over-interpreting' the data, according to Stevan Rolls of ASE, the test publishers, and John Harris of the Cranfield Institute of Technology.
'This raises the whole question of how much you can legitimately read into test results. As publishers of the tests we are going to act on the findings, which will have far-reaching implications for all providers of personality test training, including ASE,' Mr Rolls said.
He believes that interpretation of results might have to become more statistical and less intuitive. Experts had seemed to develop 'over-confidence' in their ability and went beyond the data in forming their opinions, he said.
The study conducted by Mr Rolls and Dr Harris involved 97 personnel professionals. The paper was delivered to the recent occupational psychology conference of the British Psychological Society in Brighton.