Weekly Availability Calendar

Have we gone personality test crazy?


Have we gone personality test crazy?
February 11, 2015, 7:00:00 PM EST   By Mike Russiello

Have we gone Personality Test Crazy?

Image of a person in front of a personality profile

The world has gone personality test crazy. There are so many test vendors who talk about job-fit and data-driven processes. Those seem to be the buzzwords these days for the selection business - job-fit and data-driven personality testing.

Personality tests were conceptualized around the mid-1900’s. Slowly at first, their use for pre-employment selection has grown steadily since the 1960’s. Over the past 15 years, however, usage has really hit its stride. In 2014 the Wall Street Journal quoted analyst Josh Bersin of Bersin by Deloitte’s estimate that 60 percent of all US job applicants were asked to complete some kind of pre-employment assessment.1 Based on our experience in the industry, we estimate that more than 80 percent of these were personality tests.

This may be a bit perplexing for psychologists aware that multiple studies show only a weak relationship between personality traits and job performance. So what’s fueling the growth?

What’s behind the growth?

The driving force has been the Internet, which enables testing companies to test a large number of candidates, collect the results in a single database, and analyze that data - all quite cheaply.

How does it work?

Image of a cookie cutter carving out people-shaped cookies

When you have a lot of data to work with, you can use computers to find patterns within that data. So, let’s say you use a test that collects scores on 15 different personality characteristics (we know companies that collect more than twice that number). You test 6,000 applicants for a specific job, don’t look at the scores, hire 1,000 of them, and then find that 600 of these new hires stayed with the company for at least a year. You could then use the data you collected to figure out which personality characteristics predict that a person will be hired and will stay at least a year. This set of characteristics becomes your pattern.

Once you’ve defined your pattern, you then test everyone who applies for a job and only interview the people who fit the pattern. If you did everything right and nothing unusual has happened, you should see an increase in the number of new hires who stay with the company for at least a year.

This should sound pretty good to you. A responsible recruiter or HR manager who follows the process above should be able to quantify the reduced turnover in monetary terms, showing a strong return on the investment made in the testing system. In fact, many personality test vendor websites tout these figures openly on the site with statements like:. “XYZ Company saved over $1,000,000 a year in reduced employee turnover.”

It works. So is there a problem?

Yes, data-driven personality testing can work and save big companies a lot of money. But personality-based pattern-matching has limits, and can be used in ways that cost (rather than save) the company money.

There are several potential issues. These stem from the fact that personality tests measure how a person fits the job environment, not whether they can do the job itself. This means any given target personality pattern is tightly coupled with the particular work environment and the particular job.

The first problem is that jobs and work environments change. When either one of these evolves, sometimes even in a seemingly small way, the pattern may not work anymore.

Second, the bigger your hiring volume (in a specific job), the more data you have to work with and the more effective data-driven personality testing will be for you. Big data is only valuable if it really is “big.”

If your hiring volume isn’t high enough, you can’t create a useful pattern. Since patterns are highly job and work environment dependent, you can’t borrow someone else’s pattern, either.

What’s the right number of hires to have sufficient statistical power to create a useful pattern? The answer is “it depends,” but a good rule of thumb is at least 100 - for each type of job. If you don’t hire at that volume, the data-driven approach probably won’t be that useful.

Finally, since personality test questions are not directly job related, and they can sometimes be intrusive (aka “bad news makes me feel very anxious”), they annoy candidates, who may sometimes complain a test was unfair. Narrow patterns (also called ‘profiles’) can exacerbate the problem, since they often result in a higher amount of false negatives -- candidates who could do the job well but are screened out by the pattern, and who feel they were somehow cheated out of a job by the test.

What’s the solution?

The way to avoid the issues associated with data-driven personality testing is is to measure characteristics that are related to job performance, usually as a complement to the personality measures. Like two differently skilled team members, working together they almost always produce a significantly better result that any one member working alone.

Measures are related to job performance include knowledge, skills, and abilities.

As a result, many testing vendors have started to combine knowledge, skills, and ability measures with personality measures. It’s a trend we expect to continue until all pre-employment tests are a hybrid of both personality and job-related characteristics. So when you are choosing testing vendors, we recommend you select one who does so.


  1. http://online.wsj.com/articles/are-workplace-personality-tests-fair-1412044257

Follow HR Avatar
Follow Us Follow Us Follow Us

Follow Us
Follow Us LinkedIn LinkedIn
Follow Us Facebook Facebook
Follow Us Instagram
Subscribe to our Newsletter Tips & Updates
COPYRIGHT  ©  2024   HR Avatar, Inc.   |   Terms and Conditions   |   Privacy
Go to Home Page
This website uses cookies to ensure you get the best experience on our website. Learn More