Are You Using a Personality Test to Predict Job Performance?

We’re all pretty sure personality has something to do with job performance (at least until you become a senior manager). People graduating from seminars walk around the office wearing badges that proudly proclaim, “I’m a High D,” I’m an “INTJ,” “I’m an Analytical,” or “I’m a Concrete Thinker.” That’s nice, but advertising your personality type is not the same as wearing a badge that says, “I can solve complicated problems” or “I can effectively communicate my ideas to associates.” You see, personality is an important part of who we are, but it does not have much to do with “hard” skills such as mental intelligence or social intelligence. Nevertheless, people persist in efforts to find strong links between personality and job performance (see my earlier article on validation). Results of these studies have run from very strong to zilch (a technical term used by people in the “know”). On the average, though, most studies shown pretty sorry results. For one thing, experienced psychologists have never agreed on any general theory of personality. They have opinions all over the ballpark. Some say personality is hard-wired, some say it is learned and some say it is a part of both. I think some explanation for this disagreement can be found in personality history. (Incidentally, I am using the term “personality” to refer to any test where the applicant self-reports his or her traits, characteristics or preferences. Sometimes these are called “motivation” or “preference” tests.) About 100 years ago, a German psychologist proposed that a theory of personality could be developed by looking it up in the dictionary. No, not the definition – the words! He thought personality might become clear if someone listed all the words that described people. Forty years later, Gordon Allport made a list of about 18,000 descriptive English language terms (it was rumored he retired with a severe case of writer’s cramp). Cattell, another psychologist, used this list to come up with a system of 171 groups of synonyms clustered into what he described as 13 semantic- connected centroid factors. His work gave everyone a headache and no one was ever able to duplicate it. Some 20 years after Cattell’s work was published, researchers started using new toys – university computers and sophisticated statistical tools. Cattell was forced to use pencils and graduate students. Tupes and Christal published a paper for the Air Force where they stated all personality descriptors collapse into just five empirical factors. Yes…that’s right, 18,000 words can be put into 5 big buckets. Why am I telling you all of this? …I forget! Just kidding. I’m telling you this because most of the personality stuff you see on the market is either very old or is based on the broadest possible definition of personality. A majority of personality instruments were not, and I cannot possibly overstate this, developed to predict job productivity! The early psychologists were looking for the “normal” personality because many of them planned on using it in their clinical practice (this was before Medicare). Now we are in a late 20th century pickle! How can we use a test based on general personality theory to predict specific job performance? In many cases, we cannot. That is unless, by some stroke of luck, the aspect of personality you are measuring just so happens to be directly related to a specific aspect of job performance. For example, in one of my studies, the DISC successfully predicted job performance. I was so excited by the results, I tried using the DISC in a second study. Unfortunately, I got zilch. Both were sales jobs, but in one job, the DISC factors worked. In the other job, it bombed! The DISC was a nice basic model, but its factors were too general to predict performance even in similar job families. The results drove me to design two experiments: First, explore the relationship between personality and “hard” skill skills; and, second explore the relationship between personality and jobas seen through the eyes of the manager. In the first experiment, I studied about 260 mid managers. They completed a very good personality test (i.e., based on a broad “big-five” model widely used in business). The also completed some assessment center exercises. (Assessment center exercises can be thought of as “work samples”). Scores on these exercises provided “hard” data about problem solving ability and interpersonal skills. Guess what? Problem solving-personality scores had only a 2% relationship with problem solving work samples. People-skill scores had only an 8% relationship with people skill work samples. Pretty crummy odds when making a $100K employment decision. OK, I was convinced that personality was a poor predictor of “hard” skills. But I still did not know if personality could be used to predict subjective performance such as manager appraisal ratings. This required a slighty different approach than the assessment center study. I first had to develop a different kind of personality test. This test was also based on published research, but it only contained factors associated with either job fit or job performance– no broad-based or generalized theories. I asked about 100 people to take the new personality test. I asked managers to complete a detailed performance appraisal on each incumbent (based on a job analysis I had done). I analyzed the results using some pretty sophisticated software. Shazaam! Personality was a very good predictor of performance appraisal, up to 90% accuracy. But the results contained some very interesting surprises: factors emerged in “patterns” depending on the rated item and the position. Never were all ten important and never were there less than three associated with an item. But, that was not all. Not only did personality factors change with the item, they also changed with position. Wow! This study showed that personality was a good predictor of manager ratings but, 1) only in patterns, 2) in different combinations depending on the task, and 3) in different positions. So what conclusion can we draw from all this?

  1. A general personality test may produce totally unpredictable results.
  2. Article Continues Below
  3. A general personality test is a poor predictor of “hard” skills such as mental intelligence or social intelligence.
  4. A specially-designed personality test can be a good predictor of performance appraisal ratings.

So the next time someone suggests hiring a “high D,” an “INTJ,” an “Analytical,” or a “Concrete Thinker,” ask them, “For what skill area, in what position and combined with what other factors?” Then be sure to add, “By the way, how do you also plan to measure their ‘hard’ skills?”