Catherine Tinsley on How Statistical Discrimination Impacts Bias in the Workplace

In a recent working paper, Statistical Discrimination Against Minority Groups, Catherine Tinsley, Raffini Professor of Management, and her colleagues explain how utilizing systemic statistical analysis to assess performance inherently impacts bias of certain minority groups. 

As the push for diversity, equity, and inclusion (DEI) in the workplace gains momentum, with projected spending exceeding $15 billion in the coming years, it is crucial to ensure resources are used effectively. To achieve substantial progress, companies must adopt innovative methods and procedures that address the current obstacles to reaching DEI goals. Merely allocating more resources is unlikely to bring the desired success, says Tinsley.

Tinsley argues that a leader’s beliefs about how members of a certain demographic group will perform can be highly influenced by the very salient information about the top performers of a job. Then, if the underlying population is demographically imbalanced, such as more males than females or more Whites than non-whites, when this imbalance is neglected it leads to skewed inferences about how good the majority demographic group is relative to the minority group. This creates a false belief that majority members perform better, which helps to perpetuate a uniform, majority-favored workforce.

The study introduces a novel idea called “population neglect” in which decision makers ignore the fact that different demographic groups within a population are unevenly represented and that this can bias managers’ expectations about the success rate of minority candidates.

“The investment to hire from diverse races, genders, and ages is a noble pursuit. However, current training is heavily focused on fighting direct bias and discrimination,” said Tinsley. “The reality is that managers may be making their choices based on a misunderstanding of the data they’ve been given on candidate performances. Hiring managers are misinterpreting the performance data on who would be most successful at a given job.”

In the paper, Tinsley and her colleagues highlight the following example to support this phenomenon: “Suppose a law firm wants to hire new associates. It seems that across its top 20 senior associates, 16 are white, three are Black, and one is Asian. If leaders at the firm failed to adjust for the demographic proportions of the U.S. population, they would incorrectly infer that white senior associates are more productive at the task because they make up 80% of the top performers.”

The problem is that while there is an effort among managers to diversify their staff, Tinsley says the results are lacking. “The reality is when managers are looking to hire for a given position, they must look to who will be most successful at the job. Despite the motive to hire and advance women and minorities when they are underrepresented within certain industries or organizational levels, managers make their final hiring decision based on who they believe will be most successful, and these beliefs can be wrong.”

“The bottom line is that organizations are spending a lot of money on diversity training that may not address the full problem,” said Tinsley.

Tinsley highlights the fact that failure to incorporate differences in group size when receiving information about the composition of top performers thus means members of minority demographic categories may be continually penalized across a wide range of industries. No matter the source of the imbalance, information about top performers is only informative about differences when one adjusts for relative group size.

“DEI training should include teaching people how to evaluate and interpret the data on who would perform best at any given position. Managers want to hire the best people; they just need to fully understand how statistics play into the evaluation of which candidates would best fit the bill,” said Tinsley.