by Karla Cloete
Many companies offer services, increasingly backed by AI, that claim to be able to match candidates’ personalities, IQs and general psychological profiles with the demands of a job opening. In some sectors this has become standard practice with young job seekers complaining of having to fill out hour long personality tests for each job they apply for.
But is there a method to this madness? Are psychometrics the secret to unlocking next level executives for your company? Here are some things to consider:
1) Bias
A common pitch for the use of psychometrics is that it can remove bias from the process by scientifically quantifying a candidate’s traits, interests, and abilities. Which in theory sounds pretty good right? Except it’s a little more complicated than that…
Psychometric tests are designed with very specific populations in mind. When being created a testing population is chosen of specific age ranges, genders, job functions, races, and cultures. These test populations are used to create norms. These norm tables are used to measure testers performance and abilities against that of the testing population and one another.
Some companies have been using AI to speed up the testing process, but this is a great concern for employers who strive to promote equity and fairness.
AI doesn’t have our human bias, but they do use our human data. AI models make decisions based on the data that’s made available to them. Research shows most AI models are trained on data based on straight white males, meaning that female candidates, gender non-conforming candidates, or candidates of colour are placed at a distinct disadvantage. In comparison these candidates may unfairly show poorer scores because they are being measure on a test and against norms that aren’t designed for them. Not to mention these tests usually do not account for or make accommodations for neurodiversity, learning challenges/disabilities or for their home language.
2) Accuracy
The accuracy of the data is a huge selling point. It’s hard to argue with cold hard facts, right? But how solid can this data be?
Firstly, the candidates completing these assessments haven’t completed any form of intake interview or filled in the necessary biographies. This means the tests haven’t been selected for each candidate, so the test may firstly not be appropriate for them to begin with.
Secondly, the candidate is taking the test at home where they cannot be assisted or monitored to ensure fair testing procedures. Many psychometric tests also require that they not be taken more than once in a 6 month or 2-year period. This is because taking the test multiple times leads to test-wiseness, where a test-taker can improve and alter their performance and answers based on previous experiences. If one considers that job seekers may be filling out many applications, they may also be filling out many psychometric evaluations.
3) Ethics
A lot of tricky questions come into play in any psychometric testing process. Questions about multi-cultural fairness and the suitability of a given test have already been discussed.
But what’s more are questions about data security. In true psychometric testing practice, there are strict rules on data security and confidentiality that are governed over by psychological boards and government bodies. The test-takers can be assured their data will be kept confidential and will be stored securely according to government regulation.
Mass testing companies like this do not provide test-takers with the same reassurance. A test-taker agrees to a general waiver but has no idea who will have access to their very personal data or how it will be kept secure. This is a huge concern given the increasing prevalence of data breaches.
Furthermore, we can ask ourselves if it’s really ethical to demand that candidates complete these sorts of evaluations in exchange for a chance to make a living. These tests are usually very time-consuming, and all psychometrics tests are considered risky as they can affect a test-takers well-being.
Any psychological act- whether that be therapy, or personality tests- must be agreed to with full free informed consent. Given that these test-takers don’t know the norms of the tests, who will have access their data, and they feel pressured as refusing to partake may jeopardize their standing in the selection process there are many questions to be asked about exactly how free and informed their consent truly is.
Psychometrics have been a part of the job selection world for many years and these tools have their time and place. When they are being administered in appropriate settings, by trained professionals- who can take into consideration the factors we’ve mentioned as well as countless others- they can be powerful tools.
There are many concerns for companies and startups that should be carefully thought over before choosing to use these tools, especially when they are applied at mass-scale and backed by AI models.