Missing Women in Tech: The Labor Market for Highly Skilled Software Engineers

Self-reporting of technical skills was one of the most important factors in recruiters’ decisions to reach out to job candidates, however women were less likely to self-report their programming languages than men. 

• When recruiters have objective evidence of candidates’ experience in a coding language, candidates who self-report knowing the programming language are predicted to be approximately 30% more likely to be recruited. However, the benefits of self-reporting drop off for more experienced candidates. 
• Compared to male programmers, female programmers with experience in a programming language are 7.86% less likely to self-report that they know the programming language. Conversely, female programmers who do self-report programming languages have more average experience with the language than their male counterparts. 

• Recruiters are predicted to be 6.47% less likely to reach out to a woman than a man with comparable qualifications, suggesting that recruiters do not adjust for gender differences in self-reporting of technical skills.  

In short, these findings show that female engineers are less likely to self-report programming skills, and recruiters are not adjusting their recruitment strategies based on these gender differences, which may hinder companies’ efforts to improve gender diversity of their workforce. Although gender-blind resume review can potentially eliminate some discrimination, employers need to think carefully about how to correct for gender differences in candidates’ approaches to the recruitment process.

Using unique data from an online recruiting platform, the author investigated gender differences in job candidates’ decisions to advertise their technical skills to potential employers. On a job seeker website, candidates were able to self-report software programming skills. Job seekers were linked with open source computer code they had uploaded, allowing the platform to verify programming skills. The author collected a dataset of 170,886 candidates (7% female) on the platform who had verified experience in at least one programming language. This enabled the author to quantify the gender differences in a candidate’s decision to self-report programming skills, given their verified programming skills. Next, the author observed which candidates the tech company recruiters indicated they were interested in messaging. By comparing candidates with and without self-reported programming skills, the author investigated whether self-reporting of programming skills increased the likelihood of recruiters showing interest in male or female candidates. Finally, the author tested whether recruiters adjusted to gender differences in candidates’ decisions to self-report programming skills.