Morality Report: Precrime Comes to the Office

Hiring decisions have long relied on résumés, cover letters, and interviews, but evidence suggests these tools are crude predictors of future behavior. Recruiters spend an average of six seconds reviewing a résumé, and early positive impressions often bias interviews that follow. The result is a hiring process built on a series of imperfect judgments.

The upside of making a good hire is obvious. The downside of a bad one, however, is often underestimated. In the United States, dishonest retail employees steal more from employers than shoplifters do, and research has suggested that employee deviance may contribute to as many as one-third of business failures.

Research from Cornerstone OnDemand highlights how damaging a single problematic hire can be. Their analysis found that good employees are 54 percent more likely to quit when they work closely with a coworker who engages in toxic behavior.

Given these stakes, organizations are increasingly searching for better ways to predict misconduct before it occurs. Traditional signals such as personality traits have shown some promise. Employees who are more prone to feeling guilt, as well as older workers, tend to be less likely to engage in behaviors like absenteeism, theft, or abuse of sick leave.

The rise of people analytics and Big Data has expanded this effort beyond broad correlations toward more individualized predictions. One notable example comes from Cornerstone’s research on toxic employees.

Applicants who significantly overestimated their own technical proficiency were found to be 43 percent more likely to engage in toxic workplace behavior after being hired.

Other academic work has explored how language itself can serve as a predictive signal. In simulated workplace studies, researchers monitored employee communications and found that individuals on the verge of acting maliciously changed how they spoke. They used more singular pronouns, fewer cooperative phrases, and reduced linguistic mimicry—signals that allowed researchers to identify up to 93 percent of harmful employees in controlled settings.

Not all screening tools are supported by evidence, however. Credit reports are still used by many employers as a proxy for honesty and responsibility, yet empirical studies show they are largely ineffective at predicting workplace misconduct. Poor credit often reflects life circumstances rather than character, and credit reports themselves are frequently inaccurate.

The broader lesson is not that résumés are useless or that analytics provides perfect foresight. Rather, every hiring method has limitations. Data-driven tools can meaningfully improve decision-making when used carefully, but they cannot eliminate uncertainty entirely.

Hiring will always involve risk. The challenge for employers is to combine human judgment with validated, evidence-based signals—using analytics to reduce blind spots without assuming that misbehavior can ever be predicted with complete certainty.