How to Select an Assessment That Will Predict Job Success Without Holding Bias
Tests measuring personality as a means of aiding in the hiring process remain controversial, and it’s not hard to see why. Although when done right they have been shown to predict a candidate’s success, there are many tests on the market, and it’s hard to know which ones are reliable. This confusion hasn’t kept professionals away from this useful tool though, with 22% of organizations using personality tests to evaluate candidates.
We should first deal with the obvious: do we have an accurate model of personality and reliably test it in someone?
For decades, researchers have been investigating exactly how many factors make up a reliable taxonomy of personality. Again and again, studies concluded that personality can be broken down into five distinguishable factors. Though the names given to these traits varied, researchers began to realize that they were all discussing the same basic factors, and eventually most of the scientific community settled on the same terms: Openness to Experience, Conscientiousness, Extraversion, Agreeableness, and Neuroticism, making up the “Big Five” model of personality.
But the question remains, even if we can reliably split the personality into five traits, can assessments accurately measure these traits in someone, and do these tests have any validity when predicting job performance? Yes, it turns out (well, if you’re using a reliable test).
Two researchers, Murray Barrick and Michael Mount performed a meta-analysis of 117 studies that showed a relationship between these personality factors and job performance. They found conscientiousness to be correlated most strongly with job performance, and extroversion and openness to experience to be correlated most strongly with training proficiency (such as the quality of post-training work samples and the length of time it took to complete training). While there were many subtle differences in how strongly a trait predicted a certain element of job success (most differences could explained by the different types of positions and the different measures of success investigated), overall the analysis revealed that measuring the Big Five helped to predict an employee’s performance.
In another meta-analysis, Robert Tett, Douglas Jackson, and Mitchell Rothstein came to similar results. They found agreeableness to be the strongest predictor of job performance, followed by the other four in varying degrees of correlation (all five were positively correlated). The differences in the predictive strengths of each personality trait between the two meta-analyses are probably due to the different kinds of jobs being evaluated, as clearly certain characteristics are of different value depending on one’s position.
So clearly these tests can help to evaluate a candidate’s potential success at a company, but how do you wade through all the unreliable ones to get to a version that will help shortlist your talent effectively?
For starters, you might want to be wary of tests that allow candidates to fill out their own responses. There are concerns that test takers will fill out their assessments based on how they think their potential employer will want them to respond rather than being honest. Even if candidates intend to answer completely honestly, there are likely to be questions that they are unsure how to answer and they may end up answering based on their current mood, which could change by the next day or even the next hour.
Additionally, a huge consideration when selecting a personality assessment is whether or not there is any unintentional bias built in. Does the test favor men over women? White applicants over those of color? Thankfully, there is evidence that the Big Five model itself does not discriminate based on race, ethnicity, nationality, or gender. However, there can still be issues of bias when there is a lack of diversity in the original sample used to create a test. For instance, the first version of the Minnesota Multiphasic Personality Inventory (MMPI), a test given mostly to law enforcement candidates, consisted almost entirely of white people from Minnesota, and thus it favored candidates who fit that mold. There is currently a second version of this test that expanded its sample, but there are still concerns that it under-represents Asian and Hispanic Americans.
You must also be conscious of how a test evaluates those with disabilities. Again, the Big Five model itself measures personality accurately for those with disabilities (reliabilities, means, variances, and correlational structure are all virtually the same as for the able-bodied population). Some tests, though, will start to veer into gathering information about one’s physical or mental health. Basing a hire on this kind of data is unethical and can get a company into legal trouble.
TalentSwot addresses all these issues in order to provide an effective way to shortlist your candidates. TalentSwot uses a chatbot to converse with applicants and uses artificial intelligence to compile a succinct, accurate personality profile based on applicants’ responses. As there are no questions a candidate needs to answer themselves, there is no risk of lying, and the results are consistent every time. The program pulls from huge amounts of data from diverse groups of people in order to analyze all text effectively, constructing a reliable description of personality. In addition, it utilizes machine learning to constantly integrate new data from new sources, decreasing the risk of bias even further. TalentSwot sticks to the Big Five traits without dipping into analysis of mental or behavioral illness, so there is no legal worry, and you get the most pertinent information for your hiring decisions. This is a way to cut through all the pseudoscience and discrimination that can fester in the world of personality testing and take advantage of its most useful aspects.