How to accurately predict behavior (and identify top performers)


I want you to think about the best coworker you’ve ever had.

What were they like?

Did they work hard?

Were they organized?

Were they good at following through on everything?

How did they deal with stress? Did they freak out whenever things got urgent or high stakes? Or did they remain calm and collected?

Did they genuinely care about the company and the team? Or were they more into their own personal glory?

Were they good at following instructions? Or were they rebellious, always looking to do things “their way”?

Were they smart and able to learn quickly? Or did they have a hard time adapting to new situations and figuring out new tasks?

Chances are they were all these things: hard-working, organized, resilient, team-oriented, good at following instructions, and quick witted.

It turns out that decades of research has shown that these are some of the traits most predictive of job performance—no matter the role.

It also turns out that these are some of the traits most commonly asked for in job posts.

Employers have good instincts about what makes for a good employee, but they’re almost always *really* bad at assessing these qualities accurately.

Let’s look at some of the most common ways companies choose who to hire (in order of perceived effectiveness).

  • Resume review

  • Reference checks

  • Interviewing

  • Work-task tests

  • Job tryouts

Are any of these methods good at giving the hiring company a picture of the qualities listed above?

Can you really understand how hard-working someone is from a resume? Not really.

From reference checks? Better, but still not great. Applicants generally list their friends as references, and employers are very wary of speaking ill of former employees (for legal reasons).

How about during an interview? Not really. Interviewing will give you a general sense of how sharp (and how suave) someone is—but doesn’t tell one much about their work ethic.

What about a work-task test? Better. After all, if someone has an aversion to hard-work, they won’t spend the time doing any take home assignments or rigorous in-office tests. But almost anyone can force themselves to put in the extra effort to make a good first impression. So even work-task tests aren’t that informative.

What about job tryouts (where you hire someone for a trial period)? Even better, but these still fall prey to the same issue as the work-task test. You can think about job tryouts as longer term work-task tests. Yes, they give a more accurate picture of the work-ethic of the person in question, but they still are quite inaccurate indicators of long-term employee performance.

Why would this be?

One simple reason: a person’s behavior over a 1 or 2 week (or even 4 week) period is noisy. Let’s say that you’ve discovered someone who is *actually* the perfect person for the job. Someone who’s incredibly smart, hard-working, and organized. However, as fate would have it, their father has just passed away. They’re a bit depressed. Since you don’t know them well, you just think that this is their personality—low energy, kind of sad.

During your job tryout, they work well, but they don’t really go above-and-beyond. They seem more than happy to leave at 5:00 or 5:30 each day. They aren’t hanging out with the rest of the team, grinding away until 7:00PM or 8:00PM.

From this 2-week trial, you might decide to pass on this candidate. That, however, would be a mistake. It turns out that three weeks after you reject the candidate, they feel better and once again become their old, cheerful and hard-working self. They end up going to another company and becoming an all-star employee.

This seems like something that’s impossible to avoid. Lives are complicated. People are complicated. You’re never going to be able to get a true sense of what someone is *really* like…

…or can you?

It turns out that many of the best behavioral scientists of the past 100 years have spent their time figuring out how to do exactly this. And in a few dozen (or hundred) questions, you can build a very accurate picture of who some *truly* is.

That’s the art and science of personality testing.

While personality tests can seem sophisticated and mysterious, they’re conceptually quite simple: they’re nothing more than standardized, in-depth interviews.

Human beings are bad at doing the same thing the same way over and over. Computers, however, are built for this. If you tell them to always ask the same questions in the same way, they’ll do it perfectly—and they won’t forget or distort the responses in any way.

And when you’ve asked the same questions in the same way to a group of people, it makes it easy to compare them. This is the magic of standardization.

And computers can ask people questions that, if asked by a human being, would seem weird. For example, can you imagine asking someone “if they keep their room clean”? That would be a strange thing to ask someone, face to face, in an interview. It’s much less strange to be asked a question like this by a computer. Especially when you’re sitting in front of a screen by yourself. Human-to-human social norms make certain important types of questioning and information collection hard (or impossible). These norms, however, don’t exist when we’re talking about human-computer relationships. We feel much more comfortable revealing information to computers than friends or complete strangers. Can you imagine asking friends or family some of the questions you put into Google?

Standardization and social norms are what make digital interviews so powerful. And this is why personality tests are more powerful predictors of future behavior (and job performance) than even work-sample tests and job tryouts.

The question then becomes: which tests should you use?

Which is the topic we’ll discuss tomorrow.

PS: If you have any questions about behavioral science, habit formation, hiring, personality psychology, etc. you can book 15, 30, or 60 minutes with me here --