An article in the BBC earlier this week described how a 12-year-old girl joined Instagram and was immediately bombarded with makeup and fashion advertisements despite actually being interested in academics and athletics. Bias in advertising has received considerable attention in recent years, especially the ways in which it can lead to illegal discrimination. This raises the question of whether targeted advertising in general can ever be minimally biased or whether its very existence reinforces the harmful stereotypes that damage society.
In the aftermath of the BBC article, I asked female friends and colleagues across the science and engineering disciplines what kind of ads they saw when they logged into social media platforms. While only a few dozen replied in time for this article, the result was nearly unanimous: Advertisements related to their professional disciplines were few and far between. In contrast, those with male spouses or partners in engineering or science disciplines reported that their spouses received quite a few science and technology-related advertisements, ranging from certificate programs to products to upcoming events, including career fairs.
Speaking with many of them, a common theme was that the dearth of STEM-related (science, technology, engineering and mathematics) ads was consistent over time. Several noted that even in college they received few of the ads related to their discipline that their male peers did. Others noted that despite being in science and technology majors, they typically saw the same ads as their female peers in non-STEM fields like business or English.
Those that eventually got married often mentioned seeing far more ads relating to their impending wedding and encouraging them to start a family than their now-husbands.
As they embarked upon their STEM careers, the ads they saw frequently revolved around stereotypical interests like clothes and makeup and glamorous Instagram-ready travel. A few noticed fashion items that featured “cutesy” STEM slogans like pink t-shirts either mentioning or parodying engineering such as pink glittery “Math is hard” shirts being marketed to women mathematicians. Their male peers typically saw far more ads related to their field, from technology events to product announcements to beta tester opportunities to leadership seminars to career coaching to books and services to advance their careers.
Those that had children noticed that their ads rapidly transitioned to motherhood and childcare. In contrast, their male spouses’ ads often experienced far less of a transition, continuing to focus primarily on their careers and professional development, with most experiencing fatherhood only as an undercurrent in those ads.
While this is admittedly only a very small sample and it is unclear to what degree these trends hold nationally or even globally, the strong alignment in their experiences raises the question to what degree targeted advertisements reinforce and perpetuate society’s harmful stereotypes that we as a nation increasingly agree have no place in our society?
Targeted advertisements are supposed to deliver messages closely attuned to the viewer’s interests. Why then might STEM women receive different advertisements than STEM men?
One interesting anecdote from many of the ones I spoke with was that they use social media to communicate with friends and family, while they turned to LinkedIn and other platforms for strictly professional communications and interactions related to their jobs. One concern that was raised was establishing a separation between their work lives and their personal lives, one boundary of which was that being added as a Facebook or Instagram friend in particular required more than merely being a random coworker. Several also mentioned combating unwanted interest in the workplace and at professional meetings by entirely excluding work-related individuals from their social accounts.
In essence, these approaches create “sanitized” social accounts that reflect very little of their work-related interests.
In such scenarios, interest-related advertisements based on social behaviors would reasonably conclude that these users had little interest in STEM topics, since their social usage reflects few STEM activities. This would leave offline data from data brokers as the primary targeting mechanism to recognize their STEM interests, but the up to three-quarters error rate of data broker data means they are unlikely to be recognized as STEM using broker-acquired data either.
Interestingly, however, the failure of interest-based selectors to capture their STEM interests may not fully explain their dearth of STEM-related ads.
Looking at my own social media feeds, the overwhelming majority of STEM-related ads I see report either that the ad was delivered to me because I was on a list provided by the advertiser (typically derived from data brokers) or that I was “similar” to the audience they were targeting, while many others were targeted only to “males” in a particular age range. Similarly, many I spoke with also saw that their ads were targeted by lists, audience similarity of their gender as a female. In fact, several of those I spoke with clicked through all of the ads they saw and noted that many were gender-targeting.
Targeting of STEM-related content to males is something that social platforms could easily regulate, prohibiting gender as a selector option or audience influencer for ads relating to fields that have severe demographic imbalances, whether based on gender, race, age or other factors.
The reliance on external data broker-derived selectors, audience similarity and attributes like gender are extremely harmful for the ways in which they artificially circumscribe the opportunities for those that do not conform to society’s outdated and harmful stereotypes.
Do social media platforms have a social responsibility to curb such advertisements?
One could imagine social platforms enforcing rules that any career-related topic that suffers from strong demographic biases should be restricted from advertising campaigns that mimic those biases, whether using the platform’s own selectors or using outside audience or marketing data.
However, the economic cost of such bans means it is highly unlikely that platforms would ever embrace them.
Putting this all together, behaviorally-targeted and interest-based advertising have become the defacto lifeblood of the modern Web, yet they risk reinforcing all of the harmful stereotypes our societies are working so hard to move past.
Should such ads be regulated to prevent bias?
Asked about this, Facebook did not respond to a request for comment.
In the end, the economic forces behind targeted advertising means there is unlikely to be any genuine movement on addressing bias absent a societal sea change.
*One of Foreign Policy Magazine's Top 100 Global Thinkers of 2013 and a 2015-2016 Google Developer Expert for Google Cloud Platform, Kalev Leetaru is a Senior Fellow at the George Washington University Center for Cyber & Homeland Security. This article was first published in Forbes magazine.