Working Paper: CEPR ID: DP16960
Authors: Markus Eberhardt; Giovanni Facchini; Valeria Rueda
Abstract: Academia, and economics in particular, faces increased scrutiny because of gender imbalance. This paper studies the job market for entry-level faculty positions. We employ machine learning methods to analyze gendered patterns in the text of 12,000 reference letters written in support of over 3,700 candidates. Using both supervised and unsupervised techniques, we document widespread differences in the attributes emphasized. Women are systematically more likely to be described using ‘grindstone’ terms and at times less likely to be praised for their ability. Using information on initial placement we highlight the implications of these gendered descriptors for the quality of academic placement.
Keywords: gender; natural language processing; gender stereotypes; diversity
JEL Codes: J16; A11
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| gendered differences in language (J16) | descriptors related to grindstone attributes for women (J21) |
| gendered differences in language (J16) | ability-related terms for men (J14) |
| descriptors related to grindstone attributes for women (J21) | academic placement outcomes for female candidates (I24) |
| ability-related terms for men (J14) | academic placement outcomes for male candidates (I24) |
| letter language (Y20) | misinterpretation of women's capabilities (J16) |
| language bias (J15) | placement recommendations for women (J16) |
| language bias (J15) | placement recommendations for men (J62) |