present time machine

The Present of Work

Lori_Foster
Lori Foster
North Carolina State University & pymetrics 

Editor’s note: PeopleScience is supposed to host a conversation between academics and practitioners about the potential and challenges of applying behavioral science to solve real world problems now and in the future. Well, what should we do if one person is both a professor and a professional studying and practicing just that? We could ask her to grab some puppets and do a Robin Williams-esque manic dialogue between her selves … orrrrrrr … we could not force our fevered showbiz dreams onto her reality and just let her do her thing. As we do with Professor Foster below. Enjoy.

 

Have you ever been stuck in a pair of shoes that doesn’t fit? If so, chances are at least two things happened: (1) you were horribly uncomfortable; and (2) your performance suffered, whether you were standing, walking or running.

A job that doesn’t fit is like a shoe that doesn’t fit, but with more serious consequences. In 2005, Amy Kristof-Brown and her colleagues published a meta-analysis in the journal Personnel Psychology, which examined what happens when people are placed into jobs and organizations that don’t fit. Negative consequences for the worker included low job satisfaction and high amounts of strain. Negative consequences for the employer included having less committed employees with worse performance and greater intention to quit.

 

A job that doesn’t fit is like a shoe that doesn’t fit, but with more serious consequences.

These days, everyone’s talking about the future of work. The discussion usually centers on technological advances and often reflects dystopian views of automation, big data, algorithmic bias and worst-case-scenarios. (Editor’s note: Killer robots made of gluten!). Having just co-edited a book on the future of work with Fred Oswald and Tara Behrend, you might think that’s what I want to discuss here. Not exactly. Instead, I’d like to spend a bit of time reflecting on the present of work.

The truth is, the present of work is far from perfect. Finding the right fit is a major challenge for workers and employers alike. Is this because there aren’t enough workers to go around? Not really. Is it because there aren’t many job options to choose from? Wrong again. There is a diverse array of jobs that need to get done, some as old as time and others evolving or emerging from new technological advances. Collectively, these jobs require a wide variety of knowledge, skills, abilities, interests and personalities. This is good news, because worldwide, there is an equally diverse array of people who stand to gain from the economic and psychosocial benefits of working.

 

The bad news is that in the present of work, we don’t connect all this talent to all these opportunities.

The bad news is that in the present of work, we don’t connect all this talent to all these opportunities very well. Michael Lewis refers to this as “market inefficiencies” in his discussion of Moneyball, the book which describes how a baseball team with relatively few financial resources beat its wealthier competitors by abandoning the flawed, subjective hiring methods historically used to measure, recruit and retain talent.

Market inefficiencies don’t just hurt the athletes and teams. They hurt all of us, employers and workers alike. Right now, organizations struggle to keep pace with consumer demands and clients’ needs because important jobs are going unfilled or are occupied by underperforming employees. Precious resources are spent recruiting, hiring, onboarding and training people only to do it all over again when employees quickly leave because they weren’t a good fit.

Hiring Bias

Suboptimal hiring practices are the enemy of fit. Resumes and interviews are the most typical hiring methods used in the present of work. Unfortunately, both methods require recruiters and hiring managers to do something that humans are not exceptionally good at doing: Making judgments about other people based on limited information. Don’t get me wrong, plenty of people make judgments of others based on limited information. They just don’t do it well.

Grandma always said, “You can’t judge a book by its cover.” Grandma was right. (This is no surprise to those of you who read “Grandma Was a Behavior Designer.”) Behavioral science research has shown that we make faulty assumptions we don’t even realize when evaluating people for jobs. Perceptions of job candidates’ qualifications are influenced by a wide variety of factors such as what the candidate looks and smells like and whether the candidate reminds us of ourselves, is currently employed, demonstrates smooth social skills during an interview and is good (but not too good!) at self-promotion. This can lead to unequal opportunity on the basis of gender, ethnicity, academic pedigree, socioeconomic status, disability status and many other factors.

Biases during hiring decisions are not always conscious and intentional. In many cases, they are byproducts of the way we think and mentally organize information. Our brains take shortcuts particularly when engaged in what psychologist and Nobel Prize winner Daniel Kahneman calls System 1 thinking. Mental shortcuts often serve us well, enabling us to function in a world filled with more information than a human being can reasonably handle. However, they can also lead us astray, leading to flawed perceptions of others, including in the context of hiring where, according to one study, recruiters spend only six seconds per resume.

Biases during hiring decisions are not always conscious and intentional. In many cases, they are byproducts of the way we think and mentally organize information.

Nancy Volkers discusses hiring bias in this PeopleScience article, where she provides examples of the kinds of errors that can plague judgments during the hiring process, such as the association bias, the serial judgment bias, the halo/horns effect and the primacy effect. Various heuristics or mental shortcuts underpin such biased decision-making. For example, the representativeness heuristic refers to our tendency to judge someone or something as belonging to a particular group or category if it matches our mental representation of that group or category. While this shortcut helps us get through the day, it leads us astray during hiring. Imagine we are filling a banking position. The representativeness heuristic suggests that we will tend to judge applicants by comparing them to our mental representation of a successful banker. If most of the successful bankers we’ve been exposed to – and therefore “see” in our mind’s eye – are white men wearing a tie, we may subconsciously (and erroneously) assume that a candidate is less than qualified if her LinkedIn profile photo reveals a young woman with dark skin wearing a hijab.

 

She likens the way many employers select employees to the way people used to pick movies back when rental stores like Blockbuster Video were popular.

In the end, hiring bias can result in the selection of employees who do not fit, while those better suited to a role get overlooked. This is costly to employers, to the false positive they hired and to the false negatives who got overlooked. Frida Polli, co-founder of a company called pymetrics*, explains problematic hiring practices in the present of work as analogous to now-outdated modes of decision-making in other domains of life. She likens the way many employers select employees to the way people used to pick movies back when rental stores like Blockbuster Video were popular. You’d go to the store hoping to find a movie that’s just right for you. Chances are, you didn’t have time to walk up and down every aisle to scrutinize every movie. In fact, many people would flock straight to the “blockbuster” aisle – the latest hits with big budgets and famous actors. In the end, you might grab a movie or two, and it was kind of hit-or-miss as to whether you actually enjoyed what you rented.

Wouldn’t it be nice, instead, if data existed about the features of the movies you actually like, as well as the features of available movies to which you haven’t been exposed? What would be even better is if an algorithm could use such information to point you to the movies that are the best fit, including those you wouldn’t otherwise consider. Of course, this is exactly what systems such as Netflix do today. And while the consequences of error are far greater for hiring decisions than they are for movie selections, we continue to approach hiring in old ways that result in “market inefficiencies.” In many cases, employers go straight to the metaphorical blockbuster aisle (i.e. applicants with an Ivy League college degree and shiny internships), read the “back of the movie blurb” (i.e. a resume), and in the process, overlook the right person for the job.

 

While the consequences of error are far greater for hiring decisions than they are for movie selections, we continue to approach hiring in old ways.

Reducing Market Inefficiencies

Volkers recommends casting a wider net to help combat bias in hiring. Companies like Hyatt are doing just that by encouraging applications from candidates often overlooked by traditional hiring practices. They are focusing especially on young adults who are not in education, employment or training. Sometimes referred to as Opportunity Youth, these individuals have an unemployment rate nearly triple the national average.

Of course, if casting a wider net simply results in more resumes to review in a limited amount of time, there’s still a risk of the same old biases determining who rises to the top of the stack. Here, Volkers recommends, “Instead of resumes and cover letters, try anonymous, relevant data collection.” That’s why, with support from the Rockefeller Foundation, Hyatt is trying out new methods of selecting candidates who will be a good fit – methods that capitalize on advances in neuroscience and artificial intelligence (AI) to minimize market inefficiencies and maximize fit to the benefit of all involved.

Specifically, Hyatt has chosen a Netflix-like approach to selection by partnering with organizations like Innovate+Educate and pymetrics to hire Opportunity Youth for entry-level hotel positions like housekeepers, bell desk attendants, stewards, culinary jobs and servers. People playing pymetrics games are asked to go through a series of computer exercises, such as inflating animated balloons without letting them burst or trading money with an imaginary partner. Advances in AI facilitate detailed analyses of moment-to-moment decisions made during game play, enabling the assessment of 50+ cognitive, social and emotional characteristics such as planning, altruism, risk taking and willingness to trust others.

The pymetrics approach to matching talent to opportunity has three steps:

  1. a data-driven analysis of successful incumbents;
  2. a de-biasing process; and
  3. assessing applicant fit.

* (Editor’s note from Professor Foster: “Full disclosure: I currently work as Head of Behavioral Science at pymetrics and have contributed to the Rockefeller-funded project described above, which was spearheaded by my colleague Andrew Avrin.”)

The process begins by asking successful people in the job in question to play neuroscience games. A detailed analysis then reveals the attributes common to these high-performing positions. Once that model is built, a de-biasing process occurs. In other words, the model is tested in the population of data to see if it’s accidentally over-selecting any particular demographic group. If so, the model is tweaked and retested until a fair model is reached.

Next, job candidates play the games. The interesting thing about these games is that it’s not about not making mistakes nor about specific attributes being “good” or “bad.” For example, impulsivity can impede performance in some jobs but contribute to effectiveness in other jobs. Once applicants are finished playing, a comparison is made between their combination of attributes and the high-performers in the role.

Recruiters and hiring managers then use fit to help identify who to interview or take a closer look at. Meanwhile, applicants who are not selected are given a chance to assess their fit with other jobs. The idea is that whatever one’s unique combination of attributes, it’s likely a good fit for one job or another. The trick is putting AI to work in a manner that helps make these matches possible, thus minimizing market inefficiencies and opening doors.

Given what we know about flawed judgments during resume screens and interviews, tools like this can help surface people the recruiter may not have noticed otherwise. In the end, it’s up to employers to decide how they want to use such information about fit. Some may use this as an automatic screen to move high-fitting candidates to the next stage of hiring. Others may consider it a piece of information alongside other “data points” for deciding which candidates will make it to the next stage.

 

The trick is putting AI to work in a manner that helps make these matches possible, thus minimizing market inefficiencies and opening doors.

The Future of Work

As fear about the future of work grows, it’s worth taking a step back to reflect on the present of work. In doing so, we might remind ourselves of two things. First, one of the biggest challenges for workers and employers alike is finding the right fit. Second, the present of work is far from a level, efficient playing field worth preserving completely intact. Today’s most common hiring methods present biases and associated barriers to organizational and positional fit. There is room for improvement.

The future of work is not something that happens to us. It is ours to shape.
Lori_Foster
Lori Foster
North Carolina State University & pymetrics 

Subscribe

Get the latest behavioral
science insights.