'Computer says no': Are you unwittingly discriminating with personality testing? (Part 1)

By Dr Janet Miller and Tom Skinner

January 25 2012 - The use of psychometric testing is now a well-established feature of HR practice. An increasing number of organisations now include assessments of personality or 'cultural fit', particularly as part of their recruitment and selection process or in connection with career development. There are strong, well-rehearsed arguments both for and against the use of personality testing, but we have identified a particular, and thus far potentially overlooked, risk of discrimination against disabled participants.

It is very important (not to mention a legal requirement under the Equality Act 2010) to ensure that candidates understand what is required of them and, where disabled people take these types of test, any reasonable adjustments are made. The provision of documents in alternative formats and extra time (where there are limits) are examples of adjustments entirely consistent with the practices recommended by many test providers, as well as the British Psychological Society (BPS) and the International Test Commission (ITC). However, there seems to be little or no acknowledgement that a participant's disability might influence the way in which they actually respond to a particular question or statement. There is, though, a genuine risk that this could happen, meaning that discrimination may occur, not simply as a result of the way that the test is conducted, but in how test data is then interpreted and used. To date, there is little or no evidence that anyone has identified this problem (and if they have, nobody seems to have done anything about it).

How might personality tests discriminate?

You might be viewing this suggestion with a healthy degree of scepticism and it may therefore help to consider a few examples. Below are examples of questions or statements from personality-related tests currently in use. Each is accompanied with comments on how somebody's disability/medical condition might affect their response.

Question/Statement "Just by looking at someone I can understand what he or she feels"	Effect of disability A visually impaired person may struggle to agree with this statement, simply because their disability restricts their ability to read body language and/or visual cues.
"I get stressed at situations that others find comfortable"	There are many disabilities which can make ordinary day-to-day tasks significantly harder, and potentially stressful. For instance, someone in a wheelchair may find the prospect of having to go up or down a flight of steps very stressful.
If I saw two neighbours' children fighting, I would: leave them to settle it uncertain reason with them	There are a number of reasons why a disability could affect the response to this statement. For instance, a profoundly deaf person may 'see' two children appearing to be fighting, but since they wouldn't be able to hear what was happening, they may be unsure that the children weren't just play fighting. If they intervened in what was not, in fact, a fight, they may be concerned about becoming the focus of (negative) attention.
If I had to choose, I would rather be: A forester Uncertain A secondary school teacher	Someone could respond to this statement having regard to particular constraints caused by their disability. This could lead to someone with, for instance, multiple sclerosis feeling that they had to discount 'forester' as a practical option.

These are just examples, where test providers have agreed to us quoting their material. However, we believe that this issue could arise in some personality questionnaires developed by leading test providers. The examples clearly demonstrate how someone's disability could affect their response. Some tests may contain few or even no such questions; others may contain a material number where responses, influenced or dictated by an individual's disability, could have a substantive effect on the test results. Moreover, effects will also vary immensely from one disability to another. Some questions could involve issues which could be affected by one or more disabilities, whereas others may be immune to any disability-related issues. All of this means that it may be very difficult for test providers to produce instruments that are completely immune from any disability-related factors. However, where an individual's disability does affect their responses to a number of questions or statements, it seems likely that this will (probably adversely) affect their overall test 'result'. In such circumstances, disabled participants may be materially disadvantaged, compared to 'able-bodied' persons. This will give rise to the statutory obligation on employers to make reasonable adjustments to the discriminatory practice or procedure.

It is worth remembering at this point that the impact of a disability on a test response may not simply concern physical or overt aspects of that disability; there may be recognised emotional or psychological effects of a disability which also affect the response. For instance, a report commissioned for Guide Dogs for the Blind in 2007 noted links between sight loss and emotional responses related to feelings of social exclusion. If it can be demonstrated that an emotional or psychological response is connected to the disability, that response must also be taken into account when considering the nature and extent of any adjustments. This is particularly important given that the Equality Act now extends to issues 'connected with' a person's disability.

Discrimination could occur at any stage when personality tests are used, but there is a particular risk when testing is used in the recruitment and selection process. Best practice suggests that personality tests should be used in conjunction with other selection techniques such as interviews. Many organisations incorporate personality and other psychometric tests into assessment centres which include a variety of other processes. However, some employers use personality or 'cultural-fit' questionnaires as a specific stand-alone stage of their selection process. In these situations, candidates need to 'pass' the questionnaire in order to progress to the next stage of the application process. Here, 'pass' means that the candidate's personality profile matches an 'acceptable' profile set by the employer organisation. This profile matching process is generally undertaken on an automated basis, with a computer generated report indicating a 'pass' or a 'fail'.

Quite literally, then, if the personality profile is not acceptable, 'computer says no.' But what if the candidate's responses to the questionnaire were adversely influenced by their disability? This could be highly problematic for employers using such selection processes. Candidates will often have replied to equal opportunities questions, including those relating to disability and the issue of adjustments to the selection process. However, with many employers seemingly oblivious to this particular risk of discrimination, it is unlikely that reasonable adjustments to the questionnaire stage (other than simple logistical adjustments) would be considered or agreed. The result is that the candidate could be discriminated against with an unfair profile being automatically generated. This leads to the candidate being rejected, without the employer organisation having any actual knowledge that disability was a factor in that rejection decision. The first time this might come to the employer's attention could be when they receive an ET1 which claims that the employer failed to make reasonable adjustments to the test process, despite prior notice of the candidate's disability.

As already indicated, whilst there may be a particular risk where personality tests are used in isolation, those risks will also exist when tests are used in conjunction with other assessment techniques. The issue, in those situations, will be the extent to which any test data is relied upon in making a final decision. Any reliance on flawed test data could be problematic.

Whilst, in principle, there appears to be a problem, there seems to be little evidence of any complaints or claims having been made relating to this kind of discrimination. Why is this? The honest answer is that we don't know. It may be that disabled participants have not considered this issue and how they might be affected. Or, particularly in recruitment situations, candidates have experienced discriminatory testing, but felt that there would be little to be gained from pursing the matter. They might be concerned about the effect (rightly or wrongly) of making a complaint on future job opportunities. Alternatively, they may feel that the benefits of making a complaint might be outweighed by the financial cost of doing so. After all, any assessment of damages would be based on the simple loss of a chance, the quantum of which is notoriously difficult to establish.

Although discrimination claims arising at the recruitment stage are, perhaps, less common, they do happen. Even if the practical risks of discrimination claims relating to personality testing are relatively small, there is one other factor to consider: If a category of applicants are disadvantaged because of a disability, this simply is not fair. With the BPS estimating that some 16% of the UK working-age population have a disability, there seems to be a real risk that a significant number of individuals may be unreasonably placed at a disadvantage; one which could mean that a number of talented candidates are being excluded when they shouldn't be. And, as already noted, this type of discrimination could occur at any time when personality testing is used. This could include assessment for career progression (or perhaps even redundancy) where the risk of complaints being made or proceedings initiated may be far greater.

How do we resolve this problem then? Well, making reasonable adjustments to the test may be easier said than done. Some of the options will be considered in a further article, shortly.

• 'Computer Says No' (Part 2)

• More Recruitment Articles