New research has found that Female Google users in search of a job are far less likely to be shown advertisements for highly paid jobs than men.
Researchers from Carnegie Mellon constructed an automated test system named AdFisher that acted like male or female job seekers. The nearly 20,000 fake profiles had only visited job related sites and were shown over 600,000 ads which the team tracked and analyzed.
The study's authors concluded that “in particular, we found that males were shown ads encouraging the seeking of coaching services for high paying jobs more than females.”
An example of this bias was a case where Google displayed ads for a coaching service for “$200k+” executive jobs. The ads were displayed to 1,852 members of the male group yet just 318 times to the female group.
Google’s ad targeting system is complex, taking into account various factors of personal information, browsing history and internet activity and also pairing this information with data from advertisers on when ad clicks led to a sale.
The fake users began with completely fresh profiles and behaved in the exact same way, aside from visiting female and male oriented websites respectively. Yet while this could show gender discrimination it is not necessarily the ad networks or advertisers purposely doing this.
Advertisers show ads where they make the most money. Clicks need to be translated into sales and advertisers respond to this clicks to sales ratio. If females are less likely to respond to online ads for spammy career advancement services advertisers will simply stop showing ads and that may just happen to be for all females.
A Google spokeswoman confirmed this fact, saying that: “Advertisers can choose to target the audience they want to reach, and we have policies that guide the type of interest-based ads that are allowed.”
Yet the study authors warned that even if its just about ad performance the discrimination is still occurring and its effects will still be felt. “Male candidates getting more encouragement to seek coaching services for high-paying jobs could further the current gender pay gap. Even if this decision was made solely for economic reasons, it would continue to be discrimination," said the authors.
Yet the study is problematic because the authors did not control for re-marketing, where the ads they see will be based on pages they have visited. If the list of websites they visited was skewed female or male it could have influenced the results of the study.
“One possible reason why Google served more male ads could be re-marketing, a marketing strategy that encourages users to return to previously visited website,” the authors confirmed.
Multiple examples of sites were found that had been observed serving ads and were also included in the list of sites the fake profiles visited. Such behavior would trigger ad serving not based on gender but based on the site visited previously.
Google did not comment directly on the discrimination issue but said that it prohibits ad targeting within its “sensitive category policy”, which includes substance abuse and other health issues. It also does not allow re-marketing within the same sensitive areas.