ABSTRACT
Women are underrepresented in leadership positions – academia is no exception. Using data on careers of doctoral graduates in Germany, we study gender differences in the decision to stay at university as a postdoctoral researcher and in the intention to become a professor. We find that gender gaps related to aiming for a professorship can be fully explained by observable characteristics other than gender. On the contrary, even after adding controls for an array of characteristics relevant to academic careers, we find female graduates to be 5.9 percentage points less likely to hold a postdoctoral position which allows them to qualify for professorship.
Acknowledgement
We thank the following for valuable comments: an anonymous referee, Maria Marchenko, Ulf Rinne, Stella Capuano, Celeste K. Carruthers, Hans-Jörg Schmerer, Joachim Grosser, Michael Möcker, and participants at the VfS Annual Conference 2020.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Notes
1 C.f. Federal Statistical Office of Germany (Citation2019). For other countries, see Catalyst (Citation2020) and references therein.
2 See the official documentation (accessible via doi.org/10.21249/DZHW:phd2014:2.0.0) and Jaksztat et al. (Citation2017) for further information on the data set. Since the items differ between the two waves, it is not possible to examine within‐subject changes over time.
3 Information on these academic positions can be found on this website maintained by the German Federal Ministry of Education and Research (BMBF): https://tinyurl.com/y7dxflqe.
4 Note that we also categorized individuals who stated ‘I already hold a chair’ as professorship = 1 when this did not include ‘Fachhochschulen’ (universities of applied sciences). The reason is that ‘Fachhochschulen’ aim to qualify their students for a more practical career. Being a professor there requires work experience outside academia and typically does not qualify postdocs for a professorship at a traditional ‘academic’ university.
5 For personality traits, individuals were asked to which extent they agree with trait-related statements on a five-point scale (1 = strongly disagree, 5 = strongly agree). Since several statements correspond to each personality trait, we calculate the arithmetic mean across the relevant statements so that our personality trait variable still displays values between one and five.
6 As a robustness check, we replicate the analysis using a Logit estimator. The main results are unchanged (available upon request).