Measuring Global Gender Inequality Indicators with Large-Scale Online Advertising Data

Ridhi Kashyap, University of Oxford
Florianne Verkroost, Nuffield College, University of Oxford
Reham Al Tamime , University of Southampton
Masoomali Fatehkia, Qatar Computing Research Institute
Ingmar Weber, Qatar Computing Research Institute

This paper demonstrates how anonymous, aggregate data from the online advertising platforms of LinkedIn and Google can be repurposed as a ‘digital census’ to measure gender inequality indicators in skilled occupations and education globally. Although these data have widespread geographical coverage, are available with high frequency, they are non-representative. We compute gender gap indicators (female-to-male ratios) of these online populations and use them to predict ground truth gender gap indicators. We further explore the different types of biases of indicators generated using these online populations. We find that the LinkedIn gender gap for users 18+ is able to predict global gender gaps in skilled occupations well (adj. R-squared 0.64). In contrast, despite a positive correlation (r = 0.4 to 0.6) between AdWords gender gaps for those searching and planning for post-secondary education and global post-secondary educational gender gaps, the predictive performance of AdWords is weaker compared with LinkedIn.

See extended abstract

 Presented in Session P3. Poster Session Migration, Economics, Environment, Methods, History and Policy