How Small Differences Can Have Big Implications for Organizations
Ever heard of the Butterfly Effect? According to mathematician/meteorologist Edward Lorenz, small effects – like a butterfly flapping its wings – can lead to much larger effects within a broader system – such as changing the course of a tornado occurring weeks later. Pretty dramatic, sir, but interesting, nonetheless.
For the Skimmers:
Research on gender bias shows only small differences between how men and women are evaluated at work, so why are there so few women in executive leadership?
Computational modeling shows how a difference as small as 1% in performance ratings adds up to a 15% gender disparity at the top of an organization.
Organizations can use this approach to model the long-term effects of inequities that otherwise look insignificant.
How might the butterfly effect emerge in complex systems like large organizations? Could something tiny (say, a 1% boost given to men on performance scores) alter the makeup of CEOs in America’s top companies years later?
A large body of research indicates only small and inconsistent differences in the leadership performance of men and women. At the same time, researchers have been working tirelessly to understanding the barriers to women’s advancement in this domain (for example). Rightfully so: recent data from Catalyst indicates that women represent a mere 5.8% of S&P 500 CEOs and hold 21.2% of board seats in those companies. Even though the findings are mixed, it’s safe to say gender differences in leadership are extremely small. Is it possible that even a small amount of bias against women could contribute to the large disparity that we see in top management in organizations?
Math geeks, economists, and accountants everywhere are probably nodding vigorously at this point. Economists readily accept that seemingly small changes to supply and demand can have huge consequences for the economy. For example, a 25% tariff on Chinese auto part imports can increase average domestic car prices by $4,400. Likewise, a recent drop in mortgage rates has led to a 79% spike in refinance applications because homeowners know even a 1% change in their APR could mean thousands of dollars in savings over time.
Can small differences like these add up within organizations? If women are subject to a small disadvantage, what is the long-term impact to the organization, and top management specifically? This seems like a difficult problem to address. There is a lot to consider when it comes to organizations (e.g., hiring rate, turnover, promotions, the gender makeup of the hiring pool, etc.). All these factors obscure the process, making it hard to pinpoint which really matter.
To answer these questions, we can use computational modeling to predict the impact of small differences in performance evaluations between men and women over time. In 1996, Richard Martell and colleagues devised a simple computer model to illustrate how the experiences of men and women differed as they advanced through a fictional organization.
Martell’s shell company had 8 levels of leadership, like a pyramid, with 10 positions at the highest level, and 500 positions at the lowest level. An equal number of men and women were in the “hiring pool,” waiting in a virtual “queue” to be hired as employees in the layers above them left. The simulation had simple rules: 15% of the employees would leave each year, to be replaced with the top performers from lower levels in the organization. And each employee was given a performance rating (based on a bell curve) with one noteworthy exception: Male employees got a 1% advantage in their performance ratings. That’s it. That is the thrust of the entire model. The researchers ran the simulation until all the employees in the organization were replaced by new ones from the queue. This allowed them to see how the makeup of the employees in the organization changed as time passed.
What they found was astonishing. If a mere 1% of bias was given to male employees, the result was a company in which top management was only 35% female. This model provides a great demonstration of how a tiny 1% difference, over time, can result in a 15% disparity in female leadership. It turns out little differences can matter. A lot. The subtle effects of gender stereotypes compound over the span of a woman's career, and this classic computational model illustrates how.
Adapted from Martell, R. F., Lane, D. M., & Emrich, C. (1996). Male-female differences: a computer simulation.
Of course, each organization is different. The underlying causes of a lack of female leadership is likely due to many factors. In considering any one organization, we need to think about how that organization’s conditions could impact leadership success over time. The important takeaway is that a seemingly small difference can matter a lot to organizations.
HR practitioners now have tools like the ones used in the Martell model at their disposal, which provide a data-driven way to understand how these kinds of changes play out over time. Management can alter different factors and examine, using a model, the potential repercussions down the road.
Given the current climate, its vital that organizations use the tools at their disposal to understand the impact of organizational conditions on gender and racial equality at the top levels of organizations. Luckily, we are at a point in history when we have cutting edge tools at our disposal to do just that. To the mangers of top companies, I say: it’s time to catch some butterflies.
A few takeaways:
It’s vital for companies to measure/audit things like differences in performance ratings and/or selection rates between genders. This is core to building a diverse pipeline of talent
Computer models are a useful tool for isolating/adjusting organizational conditions and seeing the larger impact over time
Don’t fall into the trap of thinking small differences are immaterial. Seemingly small differences can have a large impact over the life of an organization!
Special thanks to my mentor Dr. Goran Kuljanin for introducing me to the work of Martell et al. (1996), and the beauty of computer simulations. And to my colleague Alyssa (Meyers) Green for her help generating them!