Machine Learning Text Analysis of Corporate Diversity Statements Predicts Employees’ Online Ratings

In response to increased public consciousness around racism after George Floyd’s killing, many organizations released public statements to condemn racism and affirm their stance on diversity, equity, and inclusion (DEI). However, little is known about the specific thematic contents of various diversity statements and their implications on important organizational outcomes. Taking both inductive and deductive approaches, the current research conducted two studies to advance our understanding in this area. We first employed novel unsupervised machine learning techniques and comprehensively analyzed the texts of diversity statements publicly released by Fortune 1000 companies in early June 2020. Our structural topic modeling uncovered six underlying latent semantic topics: 1) general DEI terms, 2) supporting Black community, 3) acknowledging Black community, 4) committing to diversifying the workforce, 5) miscellaneous words, and 6) titles and companies. Furthermore, drawing from the identity-blind and consciousness diversity ideologies framework and leveraging tens of millions of employee rating data points on Glassdoor.com, we further found evidence that companies that released (vs. did not release) diversity statements and companies whose diversity statements emphasized identity-conscious (vs. identity-blind) topics were more positively evaluated by their employees online. Our findings shed light on important theoretical implications for current diversity research and offer practical recommendations for organizational scientists and practitioners in diversity management.