Leading up to International Women’s Day today, my social media feeds have been overflowing with videos and images of female leaders and women gathering together to network with, learn from, and support one another. Female executives have been sharing their personal career development stories and how they were able to break the glass ceiling on their way to the boardroom. These stories should be told and amplified as much as possible, but with one caveat.

In many cases, the women speak about what they did differently from other women that contributed to their success. Some shared that they “leaned in,” worked longer days, networked heavily or took more risks. Others shared how they dressed less feminine, learned to play golf, took a short maternity leave, or even hid the fact that they had a family at home. The stories range from inspiring to deeply troubling as you see the challenging rise to leadership through the eyes/experiences of women who are making history. But the unintended consequence of sharing these stories is it implies that in order to advance, women need to behave differently. Specifically, the message is “if you want to become a leader, you need to act like a man.”

We could write a book on why that’s a terrible argument, but instead, let’s turn to science. In 2017, a team of researchers and People Analytics experts took a data-driven approach to identifying which behaviors lead to men receiving a disproportionate number of leadership roles compared to women. Do they lack access to senior executives? Are they more vocal and aggressive in seeking out opportunities? Do they work more hours? What would the data say?

The researchers examined every piece of data they had access to including email communications, meeting schedules, location, etc. They even went further and had employees wear sensors that monitored where they went, who they spoke to, who dominated the conversation, how loud they spoke, and what their tone was.

The data revealed almost no recognizable differences between the behaviors of women and men. They interacted with senior leadership at the same frequency, worked similar schedules, had the same number of meetings, spent as much time in face-to-face conversations, and even had identical scores on performance evaluations. The data showed that women and men do not behave differently in any of the ways often used as reasons for why more men are promoted.

And yet, in this company, where males and females behaved in statistically similar ways, women were not being promoted.

This study highlights the danger of sharing stories of what women did differently to be promoted to a leadership position. It implies the solution to the gender gap in leadership is for women to change. It misses the big picture. The behavioral pattern that must be addressed is how women are perceived.

For example, surveys show that male leaders who have children are seen as more responsible, whereas female leaders are seen as less committed. The answer to this isn’t for women to hide the fact that they have a family at home. The answer is to address the bias that is the root cause of gender inequality.

Identifying, addressing, and removing bias in perceptions is inherently difficult and requires both complete commitment and an openness to admit fault where necessary. Because bias occurs subconsciously, many people find it difficult to accept until they’ve been presented with data. As a first step to addressing the gender leadership gap in your organization, conduct a bias audit on performance evaluations. A strong people analytics approach will go deeper than just the ratings, and also identify the contextual clues derived from a semantic analysis of open text responses. Suddenly the differences between the word aggressive in a male context (read: driven) and a female context (read: argumentative) become clear and it becomes impossible to ignore the real behaviors that must change.

Creating a more diverse leadership team isn’t about identifying more women that act like men. It involves asking the questions and letting the data tell the story of which behaviors need to change. Then the real work of gender equality by way of bias reduction begins.