'Computer says no': Are you unwittingly discriminating with personality testing? (Part 2)
By Dr Janet Miller and Tom Skinner
March 12 2012 - In 'Computer Says No' (Part 1) we identified a risk of discrimination against disabled participants in personality tests, which has potentially been overlooked. Whilst it is accepted that adjustments may be needed to the way such tests are conducted (for example, by providing test materials in alternative formats), we noted that there has been little or no acknowledgement that a participant's disability may affect the way they answer questions. This could have legal consequences for employers, not to mention potentially excluding a number of talented candidates. In this article, we look at how employers might seek to resolve this problem and the pitfalls associated with trying to make reasonable adjustments. We consider some of the options below.
1.Change the test questions
It may be hard for even the most sophisticated and experienced test providers to produce a test which is immune to all possible disability-related factors. However, making adjustments to the test questions themselves could be highly problematic. It is likely that only a very experienced psychologist could even begin to contemplate altering what are extremely complex instruments.
There are in existence tests for disabled groupings such as 'deaf people' or 'blind people' but there is a whole spectrum of hearing impairments and sight impairments with different degrees of impairment affecting people in different ways. Consequently, using a test based on such a generalised disabled norm group may be no better than using a test based on an 'able-bodied' norm group. By contrast, producing bespoke tests to take account of individual disabilities would, of course, be time-consuming, costly and, practically unrealistic.
2.Make adjustments to the test results
Whilst this would, theoretically, mean that existing tests could be used; adjusting test results is highly problematic. Firstly, only an experienced psychologist is likely to be qualified to consider how an alternative response to a particular question or statement might affect the overall result. Secondly, disability is a highly personal issue. Since most personality tests are self-reporting, adjustments could only be considered if it were possible to understand how a participant perceived their own disability would affect their responses. An assessment would then need to be made as to whether that perceived effect was reasonable or not. Even the most experienced psychologists aren't psychic!
There is one option which may, on the face of it, seem obvious - simply ask the candidate how they feel their disability affected their test responses. Unfortunately, this raises issues of legality. Any such line of questioning is likely to amount to pre-employment health-related questions, prohibited by the Equality Act 2010. An employer might argue that such questions are necessary to make reasonable adjustments to the selection process, but they may run up against a couple of problems here. Firstly, the employer would still need to demonstrate that they had sufficient knowledge and expertise to consider how any disability-related responses would affect the overall result. Even if the employer could do this, it may be difficult for them to argue that knowing more than simply that the candidate had a disability would put them in a better position to assess the extent of any adjustments. Secondly, an aggrieved candidate may raise a 'floodgates' argument; that is to say, if employers were able to ask (potentially intimate) questions about a person's disability simply because they had taken a personality test, this could open the floodgates for employers to use such tests as a means of circumventing the statutory prohibition on pre-employment health-related questions. The use of personality tests could then be used as an excuse to conduct a disability 'fishing exercise'. It seems unlikely that a Tribunal would be sympathetic to the employer in these circumstances. Overall, then, this option seems unlikely to work in practice, either.
3.Scrap the personality test
Given the difficulties of making adjustments to personality tests, the simplest solution may be to exclude such tests for disabled people altogether. Indeed, given that these types of test are often used in conjunction with other selection techniques (and many argue that they should never be relied upon in isolation anyway) it may be straightforward relying on or supplementing those other techniques. For example, at assessment centres, personality tests may be just one of a wide range of assessment techniques. Also, interviews could be used to explore the same types of issues addressed by personality testing.
Personality tests can, however, provide useful profiling data and, as noted earlier, by no means all such tests will contain questions or statements where responses will be affected by all disabilities. There may be tests which are unaffected by many disabilities. To dive straight into an alternative process every time you are dealing with a disabled person may therefore be somewhat of a knee-jerk reaction. Indeed, even if done with the best of intentions, such a step could, itself, be seen as discriminatory. After all, you could be treating someone differently simply because of their disability without any evidence, specific to that individual, to support your decision. So, as a solution, this too, may be far from ideal.
4.A possible solution?
Unfortunately, whilst we believe we have identified an important problem, we cannot offer an obvious easy solution. Organisations concerned that they may be exposed to this risk will no doubt need to speak to their test providers and possibly take legal advice. Indemnities given by test providers to their clients may already be wide enough to cover this risk, but as indicated above, there may be little that test providers can do in terms of the test content. It is really for the employer to consider the need for reasonable adjustments and test providers may, quite rightly in our opinion, be reluctant to bear the burden of their clients' errors. One solution that organisations could consider, though, is proposed below.
We have already indicated that, if there is a risk of discrimination in any given case, this may be hard to identify until after a personality test has been completed. We have also stated that personality tests can provide useful data and some may be unaffected by many disabilities. Where test responses could be affected by someone's disabilities it is highly unlikely that the employer will be able to make any considered judgment on that effect. One way of dealing with this problem, then, may be to continue with such tests as normal, but to include two questions at the end of the test. Test participants could be asked (i) if they have a disability and (ii) if they do (or prefer not to say), if they believe that any of their test responses were influenced by any disability affecting them. If participants indicate that they do not have a disability or that their disability has not affected their responses, employers may feel comfortable relying on the test data. If a participant subsequently claimed that the test did discriminate against them, it may be hard for them to take any action where the employer relied on their replies to these questions. Where, however, there is an indication that a participant's disability may have affected their test responses (or if they prefer not to say), this would put an employer on notice that they would need to treat the test results very carefully and consider alternative assessment techniques. This, then, would not solve the problem completely, but it would give employers a firm indication that they needed to give further thought to how they deal with the participant in question.
To date, the risk of discrimination in the interpretation of test results may have been somewhat of a legal elephant in a room full of psychologists. The problem is a real, but challenging, one with no easy solution. We have suggested some possible steps that could be taken to alert an employer to the risk of discrimination occurring. It will then be necessary to act with a considerable degree of caution to ensure that disabled participants are treated lawfully and, above all, fairly. However, there needs to be more input (or even just 'input') from those who produce and oversee personality tests. Guidance on testing disabled people cannot be limited to making a few changes here and there to the way a test is conducted; test providers and bodies such as the BPS and ICT need to get to grips with the broader issues in play here and how they affect test outcomes. This should include consulting with representative groups and bodies on the wider impact of disabilities on individuals. It would be practically impossible for test providers to produce bespoke instruments for every possible disability. Any guidance therefore needs to deal with the extent to which it would be appropriate to conduct personality tests (and, perhaps, other forms of psychological testing) with disabled people and, where tests are used, the risks associated with such use and how those risks can be minimised. Employers may also need to seek advice from employment law specialists on their level of exposure and the extent to which they can rely on contractual arrangements with test providers. Employers then need to reassess their existing use of personality tests and develop a consistent and transparent approach to their use with disabled people. Otherwise, the next time 'computer says no' they could find themselves in a spot of bother.
- More Recruitment Articles
(i) Dr Janet Miller is a Senior Lecturer in Human Resource Management at the University of the West of England and Panel Member at Bristol Employment Tribunal.
(ii) Tom Skinner is a Masters graduate (with distinction) in Human Resource Management from Bristol Business School and previously practised as a solicitor at a number of leading law firms. He currently provides HR consultancy advice on a variety of matters whilst actively seeking a more permanent role. If you have any queries please contact: firstname.lastname@example.org