"We propose that using simulation experiments with random assignment of players to roles presents a useful way to test and expand organization theory and elucidate the interplay between micro-processes and macro phenomena. In this paper, we discuss the advantages of using these simulations to conduct organizational experiments at scale and illustrate the usefulness of these experiments by looking at theorized causes of entrepreneurial gender bias using The Startup Game, a role-playing simulation of capital raising in Silicon Valley. In this game, we randomly assigned 27,082 players in 259 organizations to founder and investor roles involving fictional companies. We thereby generated multiple "worlds" with different features, which enabled us to look at how player role assignment influenced organizational outcomes. We found that assigning identical startups to female (vs. male) founders systematically resulted in 11 percent lower valuations from investors. We looked at variation across game runs using data from multi-founder teams to understand why. We found that assigning one percent more female players to the investor role resulted in lowering the gender gap in startup funding by 272 percent. These results suggest that equalizing the investor pool potentially holds the key to reducing entrepreneurial gender bias. We discuss the implications of our findings for the value of using simulated experiences to design more equitable organizations and markets."