Despite significant investments of time, energy and capital, the most common approaches to improving workplace diversity, equity and inclusion (DEI) have had little success changing employee behaviour.

New research from Sonia Kang, a professor at the Rotman School of Management and director of the Institute for Gender and the Economy (GATE), recommends a different approach.

In a paper recently published in Nature Reviews Psychology titled “Bridging individual-level and system-level approaches to advance psychology-based diversity initiatives,” Kang and her co-authors argue that many DEI efforts focus too heavily on changing individuals, and not enough on changing systems.

“There are two broad traditions,” says Kang, who also serves as Canada Research Chair in Identity, Diversity and Inclusion. “One comes out of the psychological sciences and focuses on individuals. The other comes out of the sociological sciences and looks at more macro-level systems and structures. What we argue for is bringing those two perspectives together — integrating the psychological and the sociological — so that organizations can design environments that actually change behaviour.” 

She explains that organizations have historically relied on two dominant strategies for addressing workplace inequities, both centred on individuals rather than systems.

“One approach focuses on fixing the people who experience discrimination, the ‘lean-in’ model, which for example suggests that women could overcome the effects of sexism if they negotiated better and spoke up more,” Kang explains. “The second focuses on fixing the minds of people who may hold bias based on race, gender, or other social identities, often through unconscious bias training.”

While well-intentioned, Kang suggests these efforts have done little to move the needle and may have even inadvertently reinforced negative feelings and backlash against DEI initiatives. Such programs have also proven susceptible to political pressure and budgetary constraints in recent years.

“One of the reasons that programs might not work is that people generally don’t want to be told how to think or what to do,” she says. “You might see small shifts in attitudes, but those shifts happen within systems where it’s still really hard for people who belong to groups that haven’t traditionally held power to break through.”

Instead, Kang and her colleagues recommend a more systemic approach, one that combines a series of small, evidence-based changes that are designed to reduce institutional bias. These interventions may seem modest individually, but they can make a big difference when added together.

For example, rather than requiring employees to actively apply for advancement opportunities, organizations could automatically consider all qualified internal candidates by default and allow them to opt-out.

“Promotions processes are typically very active – you have to apply, you have to tell someone that you're interested, somebody has to put you forward,” says Kang. “But those requirements can make the pool of candidates less diverse because they reinforce existing social norms. When we shifted the process from opt-in to opt-out, where everyone above a certain threshold was automatically considered for the promotion, we were able to completely eliminate the gender gap.”

Another strategy involves changing evaluation scales. Research by Rotman professor András Tilcsik shows that rating systems that use scales out of numbers traditionally associated with brilliance or perfection (e.g., a 10-point scale) can amplify bias.

“In a natural experiment involving university teaching evaluations, women tended to receive lower scores than men. [The university] just so happened to change its evaluation scale from being out of 10, to being out of six,” Kang says. “People seem to have a harder time giving women that 10 out of 10 perfect score than they do for men, but when that was changed to a six-point scale, the gender differences disappeared.”

Bias can also be reduced by evaluating candidates’ qualifications and experiences side-by-side, rather than evaluating individual profiles one by one, which the researchers refer to as “choice bracketing.”

“This is similar to how we grade exams,” Kang says. “You evaluate everyone’s answer to question one, then question two. When you isolate pieces of information and compare them directly, it reduces the over-influence of certain signals—like where someone went to school—that might otherwise spillover into your overall evaluation.”

Altogether the paper outlines eight tactics that help chip away at biases inadvertently baked into everyday organizational systems, like removing gendered language in job postings, masking identity information in candidate evaluations and standardizing interview procedures.

“No single solution is going to fix sexism or racism on its own,” says Kang. “Changing an evaluation scale won’t solve everything. But when you combine many small, evidence-based changes, they can add up to meaningful shifts.”

Importantly, Kang emphasizes that system-level reforms should complement — rather than replace — traditional individual-focused efforts such as mentorship programs or bias training. “You don't necessarily need to get rid of those approaches, but we should think of them as more limited tools than many organizations have assumed in the past,” she says.

Looking ahead, Kang believes organizations will eventually renew their focus on DEI. When they do, she hopes the conversation shifts from framing DEI primarily as a financial investment. Arguments that emphasize the business case for DEI can make such programs vulnerable to short-term budgetary constraints during difficult economic times.

Instead, Kang encourages organizations to emphasize how these kinds of changes strengthen meritocracy, ensuring that the most qualified individuals are selected for opportunities, ultimately levelling the playing field for everyone, not just members of marginalized groups.

“It’s the idea of ‘universal design,’” Kang says. “When you improve systems so they work more fairly, they end up working better for everyone.”