Wharton’s Stephanie Creary speaks with experts about how increasing access to accurate data can help drive changes in DEI policies.

The Hershey Co. has created something even sweeter than its world-famous chocolate. The Pennsylvania-based confectioner has a female chief executive, 40% of its board members are women, and it offers equal pay for women in the U.S.

Carolynn Johnson, CEO of DiversityInc, a data and media firm that promotes the business benefits of diversity, said Hershey is proof that companies can build more fair, equitable, and inclusive environments for their employees.

“If you really want to do this work, there’s a roadmap. What there isn’t is an excuse any longer,” she said.

Johnson spoke during a recent Leading Diversity@Wharton event hosted by Wharton management professor Stephanie Creary. They were joined by Shujaat Ahmad, director of People Analytics at LinkedIn, and Wharton management professor Matthew Bidwell to discuss trends in diversity analytics and data disclosure. (Watch a full video here or listen to the podcast at the top of this page.)

The conversation focused on how data analytics can help companies drive change in their diversity, equity, and inclusion (DEI) policies. But before that can happen, they need reliable, accurate data. And that’s where the trouble begins.

“Ninety percent of the time, people see analytics as the enemy of diversity. I think that everybody’s terrified that analytics will be used to create automated decisions that encode existing biases and perpetuate them,” Bidwell said. “We want to be absolutely worried about that, but we also want to think about how we can use analytics to understand the problem better, so we can start to solve it.”

“If you really want to do this work, there’s a roadmap. What there isn’t is an excuse any longer.” –Carolynn Johnson

Biases are encoded when the people who write the algorithms bring their own prejudices and stereotypes to the programming, explained Ahmad. Removing those biases requires diligent re-coding including being clear on the guiding principles.

“The same system that limits you on data is the same system that built workflows and processes that, knowingly or unknowingly, can amplify bias and potentially lead to discrimination,” he said. “You’re trying to fight the same system. You actually have to re-code and be creative to define a future of work that is not limited by the same system.”

A Rigged System

In many cases, the experts said, problems begin at the point of data collection. Many employees are reluctant to respond to survey questions about their feelings and experiences in the workplace. And many companies are reluctant to release their diversity data because it shines a light on their DEI shortcomings. Corporate attorneys often stop the release because of liability concerns.

“To be honest, no organization does as well as they would like,” Bidwell said. “They’re operating in a society that is deeply rigged, so it’s incredibly tough to swim upstream from that.”

He said diversity survey questions are socially constructed categories in which employees are asked to classify themselves. Some find those questions intrusive and worry about how much they want to reveal about their gender, sexual orientation, ethnicity, and other factors.

“The more you do with the data and the more you’re going to use it to drive insights, potentially the more nervous they’re going to be to share it with you,” Bidwell said.

“You actually have to re-code and be creative to define a future of work that is not limited by the same system.” –Shujaat Ahmad

Ahmad agreed, adding that it takes just one person to misread a data point for corporate attorneys to step in.

“From a liability perspective, when you’re analyzing diversity data, most legal departments won’t even let you put an insight on it,” he said. “It’s very closed-door, highly privileged, and rightfully so because a slight misinterpretation by anyone who is not trained in interpreting diversity data can end up with an incorrect insight that can easily take a life of its own as a fact. Then everybody’s running after a rabbit hole from a PR perspective.”

For 23 years, DiversityInc has produced a top 50 ranking on diversity for companies with more than 750 employees. Johnson said she’s seen over time how business leaders have become more receptive to the survey, understanding how benchmarking can help them make improvements.

The data that DiversityInc collects is designed to be informative rather than prescriptive for companies. Yet Johnson said there are still attempts to manipulate it.

“When we benchmark, it is very interesting to me how people want to change our benchmarking reports to tell what they want their leaders to hear and not what’s actually happening,” she said.

Making the Data Work

Despite the obstacles, data analysis is an effective tool for firms that want to find their DEI potholes and fill them. Creary asked the experts for advice on how leaders can overcome the obstacles and move their businesses from practice to outcomes.

“To be honest, no organization does as well as they would like.” –Matthew Bidwell

Bidwell encouraged companies to work with academic researchers because they are seeking to understand both correlation and causation, which can yield valuable insights. Ahmad suggested picking one project as a “beachhead” that can build momentum for meeting other DEI goals through data. And Johnson said to “meet people where they are” by having the right conversations with changemakers at the right level in the organization.

“This work is not for the faint at heart,” Johnson warned. “If they like you, you’re not doing your job. Because even the best companies have a long way to go.”

Creary, who has conducted DEI research with numerous companies, echoed that sentiment.

“We’re not in this business to make companies feel better about what they’re doing. We’re here to help them do a better job,” she said. “Sometimes you have to be OK to be the bearer of bad news because that’s the only way they can effectively change.”