Imagine being sued by someone you didn't hire. That could happen if you use tests that have not been shown to be fair.
You may already have a set of questions you ask candidates. But can you prove that they do not indirectly discriminate against a legally-protected group? The Equality Act makes it illegal to unfairly discriminate based on:
- gender reassignment;
- religion or belief;
- marriage and civil partnership; or
- pregnancy and maternity.
Properly-developed psychometric tests have research to support their fairness, so you can relax knowing that you are not unfairly discriminating against protected characteristics.
With properly-developed psychometric tests you can show your compliance with the Equality Act.
Although tests and assessments may be used to show differences between candidates, it is problematic if the test discriminates unfairly. When a psychometric test discriminates unfairly, it may show differences between groups when no such difference exists, or fail to detect genuine differences. This kind of discrimination is often referred to as bias or “adverse impact”, and can occur in a variety of forms. Common examples of bias are significant differences in performance across gender, racial or cultural groups, simply due to the way the test has been written which may cause items to perform differently among specific groups.
Not only does bias unfairly disadvantage certain candidates, but it also invites additional noise and increases measurement error, meaning that the psychometric tool used is unlikely to be adequately valid. In the realm of high-stakes testing, this is a huge issue which can be avoided through rigorous psychometric processes when developing a test, to ensure that protected groups are not unfairly discriminated against.
The aim of psychometric test developers is to reduce, or if possible remove, bias to ensure the fairness of the test.
It is important to note that in the case of genuine performance differences across groups, this should not be considered to be adverse impact.
Ensuring psychometric tools do not discriminate against specific groups is both good practice and a legal requirement. It is vital for companies and organisations to select the right tools for use in selection, as there could be strong legal ramifications if they were to use a test that did not have sufficient psychometric properties. Although it is the responsibility of the test publisher to produce fair and valid psychometric instruments, those who implement psychometric testing are responsible for selecting a tool that is of high quality and to ensure that it is used appropriately.