More and more companies are now using personality testing to predict job performance. And personality testing is now a huge industry.

But a new study, reported here, suggests that the link between personality testing and job performance is tenuous. One of the obvious problems is the potential for fake answers. These sorts of tests, unlike those for intelligence, can be messy and unpredictable, as they are based on self-reporting. Giving that sort of test is like asking someone if they are intelligent and the answer could change according to gender or self-confidence or other things.

In the past, I have written a column explaining that what you see with tests is not always what you get. Starting off with the Myers-Briggs test which was developed by a Pennsylvania housewife, Isabel Myers who started out in the 1940s at the age of 44 with no training in psychology, no knowledge of how to construct tests, no laboratory and no funding.

Still, personality are here to stay. What this means is they might need to be better refined. They neeed to be more sophisticated. The scope for fake answers should be tackled, and they should be linked to work samples, cognitive ability tests and structured interviews.