I have been reading some articles lately that say personality tests are valid for selection – depending of course on whether or not the personality trait is job-related. Well, a lay reader of that statement could say, “What good news! I get a few people together to agree on some factors and I can use my favorite personality test!” Wrong!
To understand my point, we need to first cover a little bit of the history of personality psychology. About a hundred years ago, psychology either involved experiments with animals or treating people who were nuts (yes, I know, “nuts” is not a politically sensitive word to describe the mentally disturbed, but it gets the point across, doesn’t it?). Anyway, hard-working psychologists at that time kept searching for a general theory of personality that could be used to both describe the human condition and treat mental illness (you might note that I did not mention “better hiring”). Eventually, a German psychologist suggested using a dictionary and listing all the terms people used to describe themselves – and a few years later someone did. The list initially included about 18,000 trait descriptions that over about 50 years were gradually reduced to 4,500, then 170, then 35, and finally 5 (computers are wonderful machines). The last set became known as the “Big Five” (yes, I know, psychologists are not very creative with words).
What’s wrong with this, you say? Well, these are all broad-based, non-specific, general-purpose personality theories that attempt to describe the entire universe of personality. They don’t specifically address selection issues. As such, it’s very hard, if not impossible, to statistically compare these models with anything that is highly job-specific.
For example, research using the Big Five consistently shows that three of the five factors are almost always associated with job “performance.” They include 1) caring about quality work, 2) not being crazy, and 3) being outgoing (I like to call them “Dr. Dubya’s Big Three” – no imagination, but to the point). Only a very, very dull recruiter would consider these factors to be rocket science. You can hear it now, “Say, Chuck, I have a recent job opening in my department. See if you can find me someone who believes in doing lousy work, is completely neurotic and a total introvert!” Sure.
Of course, “Dr. Dubya’s Big Three” are all much too generic to use in selection. Haven’t you noticed that a successful lineman, for example, has a slightly different personality from a salesperson, a manager, or engineer? Most people would say yes, but the huge bulk of personality research shows most tests have weak or zero relationship with any kind of job performance. In fact, a few years ago many eminent (any association with the rapper of a similar name is purely coincidental) researchers went so far as to agree that personality had nothing to do with job performance. Bummer! We see it, but we can’t seem to measure it!
Question: “Why did all the communication, clinical and personality type tests keep bombing out?”
Answer: “They measured the wrong things!”
Question: “Why do many test vendors today keep pushing their personality tests?”
Answer: “They either don’t know, don’t care or never did their homework!”
Look around you. Do all the highly successful people have identical personalities? Unless you work with androids, the answer is probably no. Are all the “dominants” successful? Again, no. I could go on, but it is only common sense that personality factors only predict performance when they are strongly associated with job performance. The same is true for leadership styles, communication styles and any other type of type of workshop-brand inventory. Nice for understanding people, but seldom correlated with job performance.
So how can you get the most out of personality tests? Start with examining the underlying test theory. Was the test originally developed for training or counseling purposes? Many times, people with impressive credentials unknowingly get outside their field of expertise and confuse communication style with job performance. If your test was developed by one of these people, give it back to the trainers and start looking for a personality test developed JUST for selection. It will have the predictive factors you need.
Next, save your self some time by demanding to see a study that shows the factors are backed by a study that associated them with job performance. Forget about examining high-producer averages – these are just empty exercises conducted by clueless people that tend to produce meaningless results. You see, averages hide individual differences among group members (if you averaged the gender of an equal number of men and women, you would have your basic hermaphrodite). Likewise, if you averaged the athletic ability of all the members of the top soccer team and compared that to the average of the bottom soccer team in the league, you would see very few differences. Both numbers may be fun intellectual exercises, but tell you nothing about test scores and performance.
So what about performance? In a production or sales job, performance may be easy to see, but in most jobs, performance exists only in the minds of people who say things like, “Find me a winner,” “She has commitment,” or “He is a team player.” I respond with, “Fine, but just how in the $%%# do you expect me to measure this kind of intangible factor? If you want to trust these poorly defined personal opinions, call the psychic hotline or phone Alan Greenspan. You have just confronted the most fundamental problem in selection: if you cannot accurately define it, then you cannot measure it. If you cannot measure it, then you cannot confirm it is associated with performance. End of statement.
Let’s say, for example, that you finally found both some measurable performance factors and a decent personality test. Now, the study begins. Begin with identifying at least two groups – a high performance group and a low performance group. You input test scores and performance ratings for each group into a computer and statistically analyze the results. If you don’t see a mean difference between scores on the test and performance scores, your test is bogus.
So where are we now? You think that is a lot of work? I agree. But consider your alternative. Use a personality test you think has value – but get nonsense results that are totally worthless job predictors and lawsuit bait, as well. I won’t comment publicly about some of the tests people have written me about, but I will tell you that every single personality test sent to me has looked the same – nice for the training arena, but worthless as a selection tool.
Personality tests can be a valuable part of selection, but you need the right tool, the right performance factors and a detailed study that confirms the validity of the test. If you are doing anything less, I have a bridge in Brooklyn for sale. <*SPONSORMESSAGE*>