Does college GPA predict job performance?

A couple years ago, Lazlo Bock, SVP for People Operations commented that college GPA is essentially worthless as a predictor of job performance. He goes on to assert that the reason is that the work environment is nothing like the academic environment. He further states that test scores, interviews, and event the famous Google brain teaser interview questions don’t reliably predict future employee performance. (1)

So, is Mr.Bock correct, or not?

The answer is - well, partly. That’s because no prediction technique can foresee all the different ways a person can perform well or poorly in a given job. However, virtually all techniques are useful to some degree. The question is, how much.

May the hiring odds be ever in your favor

I was watching the most recent installment of The Hunger Games and one of the movie’s most compelling catch phrases caught my attention: “May the odds be ever in your favor.”

It occurred to me that this is exactly what we’re trying to do when we apply different selection and evaluation tools during the hiring process. We are trying to move the odds of a good decision in our favor. The real question is, how much can they really improve your odds?

That’s a tough question to answer. But, lucky for us and Mr.Bock, others have already tried to answer it.

What is the best predictor of job performance?

Between 1983 and 2004, John Hunter of Michigan State and Frank Schmidt, with the US Office of Personnel Management, published several studies that have helped evaluate the relative merits of various methods of evaluating job candidates, particularly at the entry level.

These studies use a technique called “Meta-Analysis” to combine the collected data and findings of previous studies to produce more generalized conclusions. For example, their 1983 paper combined over 500 independent studies (2). Their later papers incorporated even larger numbers of studies, as well as new techniques for fairly combining results.

The result was a useful quantification of the general validity of various methods of evaluating candidates.

The results are below for entry-level jobs. If you’re not a psychologist, you should understand the terms validity and percent of variance. In short, validity refers to the correlation of observed performance with the measurement. Percent of variance is actually equal to the validity squared. It’s useful because it shows the percentage of a s person’s actual performance that can be explained by the individual measurement.

Additionally, the authors included values for people who are already working in a job (shown below). Notice how much higher the numbers are in general. This suggests that a promote-from-within policy, combined with one or more of the evaluation techniques listed in the table can be quite effective at reducing hiring mistakes.

These tables above provide general guidelines for evaluating just how useful a particular type of evaluation might be in predicting future job performance.

What does all this mean?

Using the tables above, we can see that GPA accounts for about 4 percent of the variation in job performance between employees. It’s hard to quantify exactly what this means but here’s how I’d put it. Please note that the mathematicians and statisticians among you are going to have all sorts of issues. I’m just trying to keep things simple for the rest of us, so please forgive the broad brush.

Let’s say the chance of making a good hire if you just choose randomly from a pool of applicants randomly is 50% - one out of two. If you instead consider applicant GPA only, your chances of hiring a solid performer go up by about 4 percent, to 52%.

Woo - hoo. It’s no wonder Mr.Bock has lost faith.

So, by itself, GPA doesn’t do much to place the odds in your favor.

But what about a test of general mental ability (GMA). Using my somewhat less than scientific application of the table above, using GMA should raise our chances of making a good hire by about a quarter, to 63%. That’s better, but not great, either. On the other hand, if I’m hiring 100 people, that means I’m likely to bring in 13 more good hires (or 13 less bad hires) than I might have done otherwise. If the cost of a bad hire is about one times their annual salary, these kinds of results can add up fast.

Moving the odds in our favor

Playing off The Hunger Games may seem a little trite, but that is exactly what we’re trying to do here.

So, what should we do to make sure we hire great people? Unfortunately, there’s no magic bullet. However, use of all available information gathered either actively or passively, will usually result in the highest probability of making a good hire. To enhance our process, for instance, we might add a personality inventory and/or a biodata questionnaire, and we might find a way to give applicants a job tryout or work sample. What you can do usually depends on the job. The key is to collect as much information as practical.

If you do this, you are still likely to get some poor hires. However, you’ll get a lot less of them. Google probably would too!

Footnotes:
(1) http://dailycaller.com/2013/06/20/google-executive-gpa-test-scores-worthless-for-hiring/
(2) American Psychologist, April 1983, Quantifying the effects of Psychological Interventions on Employee Job Performance and Work-Force Productivity, John E. Hunter, Frank L Schmidt