Lab Events
Bocconi

Seminar: Gender differences in reference letters

Image media image 2
-

Abstract 

Gender disparities in the workplace have received significant attention in the public debate. Academia, especially in the field of economics, is facing increased scrutiny to determine whether gender discrimination is widespread. Recent empirical work has studied many aspects of the “leaky pipeline” –uncovering gender bias in teaching evaluations, the publication process, conference admissions, and tenure and promotion decisions. However, little is known about the first step of an academic career, the junior “job market”. Our research fills this gap by analysing the text of over 9,000 reference letters written in support of almost 2,900 candidates who applied for entry-level positions at a research-intensive UK university between 2017 and 2020. We study differences in the language used in these letters, according to the gender of the candidate they recommend.  Borrowing from existing methods in cognitive psychology, linguistic and finance, we use machine learning algorithms to quantify sentiment, tone and stereotyping. A first, descriptive, analysis uses LASSO estimators to isolate a number of expressions in full letters and letter ends predominately associated with male or female candidates, respectively. We then develop `bags' of signal expressions in six categories, related to candidates' ability, research, recruitment prospects, graft (“grindstone”), and teaching and citizenship. We study the gendered determinants of categorised signal expressions while conditioning on applicant characteristics and letter writer attributes. Our regression results confirm the findings from the LASSO analysis indicating that female candidates are characterized with recourse to more “grindstone” expressions and furthermore with a greater emphasis on their teaching skills and good citizenship qualities. Importantly, we find significant differences depending the research reputation of their institution.


Bio

Valeria is a lecturer in Economics at the University of Nottingham's School of Economics, and a Research Affiliate at the CEPR. Her research interests are: Economic History, Development Economics, Political Economy, and Economic Geography. Valeria is particularly interested in institutional and cultural long-term determinants of  spatial inequalities.