What the Pollster's Mistakes Can Teach Us About Predicting Work Behavior

I often wonder if I’m getting it right. As a psychologist, this is my responsibility when evaluating the health of an organization. Are we posing the right questions? Are the responses representative of the population? Are we obtaining a clear picture of what is really happening within an organization? Based upon the available data, will valuable employees remain engaged? Will they walk away?

What am I missing — and what has been held back?

Each time I examine diagnostic results, I obsess over these questions.

When we consider how wrong the vast majority of pollsters were in predicting the outcome of the Presidential election, I quake in my boots. The Atlantic, skillfully takes us through why things went woefully wrong — and poses an unnerving question we must all contemplate when making data-based predictions concerning human behavior:

Did we all believe Clinton would win because of bad data, or did we ignore bad data because we believed Clinton would win?

Yes, confirmation bias may have played a role here. When we become too sure of any future outcome, we essentially stop considering the other potential end points. We lose perspective and become myopic — and that can spell danger for decision formation.

We must also consider technique. One polling organization, the USC Dornslife/LA Times Election Poll seemed to have the ability to capture what was really happening. Interestingly, their methods were a departure from other polls, with a stable panel of 3200 from which daily polls were pulled. Moreover, they considered the likelihood of an individual actually voting. So, in essence, the poll attempted to measure both sentiment and intent to behave in a certain way (voting). This is how they explain it:

…we calculate a ratio of a person’s likelihood of voting for a specific candidate to his or her estimated chance of voting.

So, let’s jump to the business of predicting how employees feel and behave in the workplace. What we can learn from the inability of the polls (and the candidates) to predict voting behavior?

• Bias abounds. There I've said it. As human beings we are indeed flawed as decision-makers and we often see what we want to see. If you think your organization, or team, or employee is in a good place — do not think for even a moment that this comes with a long-term, “forever” guarantee. Try to build “bias” protection into your decision-making processes.
• Explore the small shifts. I’ve learned that where there is smoke there is fire. If your organization is growing rapidly or is undergoing a significant change effort, pay particular attention to trending sentiments. You may not want to acknowledge the spark, but that spark could still ignite a firestorm.
• Consider who might be silent, but resolute. There are always individuals who have formed strong opinions and have already planned their future steps (and they do not feel the need to share their opinion or plans). This could be your star employee, who has observed over time that their path with you is “less than”.
• Time can erode your core base. Consider how time and events might impact your core. Elements such as stress and burnout can influence just as many departures as a lack of engagement. Consider how history might affect even your most dependable people. (Consider the resulting impact of Wisconsin. Or Michigan, for that matter.)

Have you ever been dead wrong when predicting behavior? Share your observations here.

Dr. Marla Gottschalk is an Industrial & Organizational Psychologist. She is the Director of Organizational Development at Allied Talent. A charter member of the LinkedIn Influencer Program, her posts on workplace topics have appeared at The Huffington Post, US News & World Report and The World Economic Forum