Lies, damn lies and sexual stereotypes

New psychometric tests claim to eliminate gender bias, writes Hilde Syversen
Click to follow
The Independent Online
As many as one in four tests for personnel selection may have a strong bias against women, claims Oxford Publisher Press (OPP), a publisher of tests. Having analysed for gender bias 64 sets of scores for traditional tests, OPP found that 23 per cent of the sets showed a marked bias in favour of men, as well as a smaller but still significant racial bias. The solution? Its own new ability test, of course.

Like many such people in her field, Helen Baron, from SHL, a competing test publisher, is sceptical of such statistics. After all, she says, the fact that there are gender differences in scores does not necessarily prove that they are biased. "For instance, if you take the average height of men and women, then, you find that men are taller than women. We don't call the tape measure biased, because it is showing a real difference."

The bias tends to appear, she says, either when tests are irrelevant to the job in question (say, measuring lifting for an accountant's position) or when tests are designed to test a particular ability but end up measuring something else. Here, she provides a hypothetical example of a test that aims to measure numerical skills and that is based on cricket scoring. Yes, it would measure numerical skills to some degree, but mostly it would measure a candidate's knowledge of cricket. That would be biased because men tend to know more about cricket than women. Ms Baron concludes that OPP's research undoubtedly includes real and relevant differences in candidates, as well as the biased differences. As a result, OPP's statistics are inconclusive.

Some of the differences in traditional test results between men and women also derive from the fact that they can have very different experiences of education. The traditional ability examinations, for instance, measure skills such as numerical skills, verbal skills, spatial reasoning or clerical skills such as accuracy - each of which is tested separately. Men generally score higher in the numerical tests while women score higher in verbal and clerical tests. That doesn't mean that the questions or results of any test for numeracy are biased. It simply reflects how girls and boys are encouraged to take different subjects at school. "The case is made by the fact that more boys than girls take more numerical A-levels," explains Jayne Munkhouse of the Equal Opportunities Commission.

The fact that one gender comes out higher than another in certain tests is, then, not necessarily a problem, she concludes. Rather, the problem occurs when the tests are not applicable to the particular position. "The testing has got to be job-specific. That is the only way it can survive the sex discrimination and race discrimination legislation. Indirect discrimination, where you have a requirement to pass a test that is harder for people of one sex than the other to pass, must be job-related in order to be lawful."

Enter OPP's new Able test. Acute job-relevancy is, after all, one of OPP's major claims for its new ability test. Where traditional testing would have separate tests for separate skills, the Able test measures a mix of the skills in one test, trying to imitate the actual job as much as possible. In other words, a single question in the Able test might require both numerical and verbal reasoning in an attempt to mirror actual problems that you might encounter in a job. The candidate is also given a lot of additional information, often in the form of manuals and handbooks, which, if used correctly, will provide the answers to the test.

"Able has been designed to look at the way people learn job-related tasks," explains OPP's business manager, Russell Harper. "It's giving you an indication of someone's potential to do that job. For example, we have tests for call centres and different tests for accountants and different tests for lawyers, because each of the tests incorporates different tasks ... for that type of job."

Ms Baron believes there is a case for both the traditional approach of measuring different types of skills separately and the Able approach of measuring the skills together, holistically. "The advantage of the latter is that it reflects the job more. The disadvantage is that if you get a low score, you're not quite sure which bit of it is going wrong. If you measure the three things separately, you get a very diagnostic picture. For example; they're good at the checking and they're good at the verbal, but they're a bit poor at the numeracy. Then you can say that, `Actually, I don't mind that as much as if it was the other way around'."

Ms Munkhouse favours the idea behind Able because the better a test fits the job, the less likely you are to discriminate. But even if the test is good, employers should always make sure that they are buying the right test for the job. "You should have a proper job analysis done. Make sure that the test you use tests for the requirements for the job, that you administer the test properly, and that the outcome of the test is properly monitored and validated."

For further information, contact OPP on 01865 510203.