Measuring gender differences in personalities through natural language in the labor force: Application of the 5-Factor Model

Gender stereotypes still play a major role in the perception and representation of people in the workplace. Measuring the effects of those stereotypes quantitatively is very hard though. Traditional methods, such as questionnaires, struggle to provide the full picture, for example through misunderstanding, omission or incorrect answering of questions. However, evidence-based policy making requires accurate indicators of gender inequalities to promote equality. We present a framework measuring gender stereotypes on company level using publicly available big data. Specifically, we analyse the one million websites of all German companies using natural language processing with regard to differences in their portrayal of genders through the use of certain terms. We then contextualize the gender stereotype measures following the personality traits of the Five Factor Model and their sublevels. Statistical analysis of the results indicates significant stereotypes within personality traits for large portions of the sample. The qualitative differences in gender presentation are mostly consistent with those found in the literature, which serves as a validation for the presented framework. The presented approach complements traditional quantitative measurement techniques by capturing a mainly latent level of inequality. The fully automated and comprehensive analysis of the linguistic portrayal of gender stereotypes in a corporate context is at low cost, with little delay and at a granular basis.