Being inclusive is very important to us; using the terms “male” and “female” here is intentional. By using the terms “male” and “female”, we do not intend to assume either a binary gender dichotomy or a binary sex dichotomy. We use these terms because, in our data collection, it has proven impossible to track down individual people’s gender. Very few people make this information publicly available on their CVs or websites. It was not feasible to send out surveys asking for people to disclose their gender to us: we would have had to send it to the invited speakers at dozens of conferences over many years, all the linguistics faculty, and all of the undergraduate, MA, and Ph.D. students at 50 institutions. Instead, we decided to simply assume people’s sex based on their names and pictures. Admittedly, this is quite a crude method of data collection. However, we believe that the process of assuming people’s sex is exactly what people do every day in their interactions with others and that these assumptions contribute to bias in every aspect of daily life.
We refer the reader to this link for further information regarding the difference between gender identity and sex.
We believe that bias, recognized or not, manifests in specific actions which result in patterns of discrimination. This study quantifies the pattern of lower proportions of women in each stage of the academic career. The observations we make on this website are based on our data, and care should be taken in extrapolating the results to the whole field.
We do not quantify the actions which cause this incrementally decreasing proportion of women academics. However, we can speculate on what causes this imbalance. For example, women in academia are disproportionately burdened with professional service duties (Guarino and Borden, 2017) and housework/childcare (Mason, Goulden, and Wolfinger, 2006); don’t benefit from male privilege in abstract reviews (Roberts & Verhoef, 2016); receive poorer student evaluations (Mengel, Sauermann, and Zolitz, 2017); are written poorer letters of recommendation (Trix & Pskena, 2003, Madera, Hebl and Martin, 2009); and experience outright misogyny such as the sexual harassment and assault. All of these are manifestations of implicit gender biases and potentially contribute to the attrition rates that we see. Because we are not directly measuring these actions, only the symptom, we urge the readers to be careful in interpreting our results. However, our data is useful for providing a before/after metric to determine if interventions have the desired effect.