Are companies to blame for unrepresentative hiring?
I don’t think so. Tech companies have earned a reputation as a problem child for racially discriminatory hiring practices. They have, of course, created some real buzz over real acts of discrimination, to include a cyber-security firm posting an ad that asked for “Preferably Caucasian” applicants, or recent developments to hiring protocols that inadvertently create difficulty for disabled applicants. These are problems faced by the tech industry, but they don’t paint a full picture of the more common frustration that the demographics of the engineers who design and produce tech products are not representative of the population that those products serve. At its core, the problem that tech engineers do not represent the population starts with earlier cultural influences, and then manifests itself as a skewed adult workforce.
This problem begs three questions: What demographics do tech hiring practices represent? Whose responsibility is this lack of representation in tech? And what should be done to correct the demographic spread of our developers?
First, who are Big Tech’s targets to hire? If we compare data from the United States Census Bureau with Diversity, Equity, and Inclusion (DEI) Reports from Google, Amazon, Apple, and Facebook, it does not look like our tech firms balance these decisions well. We see in every circumstance a higher proportion of Asian employees and proportionate under-representation of every other racial group relative to the US population,, and a devastating lack of women in the workforce. In Amazon’s and Facebook’s cases, we do also see a disproportionate skew of white workers compared to black and Hispanic ones.
But tech giants don’t hire from the general American population, they hire from primarily the group of college graduates with degrees in computer, information, and data sciences. If we compare those companies’ reports with the demographics of graduates in the information technology field, we see a different picture. According to Data USA’s aggregation of degree data, almost 78% of graduates in the field are men. All four of these giants actually hire a higher proportion of women in their tech departments than graduate from university with the relevant degrees. In terms of race, Facebook leaves black and Hispanic representation in employment wanting in comparison to graduates. Google, Apple, and Amazon, on the other hand, all roughly match or beat minority representation in their employment though, which runs counter to the notion of a discriminatory trend throughout the industry.
It’s worth noting that international hiring skews corporate relations to American demographics. On top of that, representative hiring is a step from these companies towards creating representative organizations. While their hiring has been recently made representative, with years of under-hiring minorities, overall tech employment is not.
That’s good? Maybe. It’s good that companies are doing all they can given the hand they’ve been dealt, but why are we here in the first place? Where does it begin?
Let’s start with some speculation. What would be the consequences of the tech industry taking it upon themselves to represent the population? We could quickly see a total demographic match at massive companies. Meta, Google, Amazon, Intel, and the like see enough applicants that they could easily hire within national proportions for race, gender identity, etc.. However, the giants’ decision to match the population would catastrophically worsen the proportions of the remaining applicant pool. In the wake of representative giants, we would watch smaller companies skew even more overwhelmingly white, Asian, and male.
From an optimistic perspective, that kind of move by leaders in the tech industry could signal outward that the field is ready to accept diversity in the work force in an almost trickle-down fashion. But given the constitutional violations it would take to execute this, and the damage that would resonate through the industry, it isn’t quite reasonable to blame companies for hiring in relation to the applicant pool.
If we pass the buck from companies, we have two main agents to investigate for the disparate set of applicants that show up at company doorsteps. These are colleges and culture.
Starting at the next highest level we have colleges, where the problem of disproportionate demographics is clear as day. Every semester I start new classes, and am consistently a little bit surprised by how different my computer science classes look from my humanities classes. Not only do they lack diversity in the visibly apparent demographics like sex and race, but in categories like class and experiences we also all fall into similar buckets.
Colleges don’t overtly prevent anyone from joining the major of their choice, but there are less direct forces that could add preventative pressure from entering STEM majors. At the University of Michigan, the College of Engineering has a higher tuition cost than its liberal arts school. While many students take a computer science major through the liberal arts school, all CS students pay the higher rate.
This table from Pew’s research shows the additional costs for engineering programs within the Big 10 coalition of schools. Since 2017, that gap has increased for engineering. National research taken in 2022 shows that the average tuition for 4 year university (considering in-state, out-of-state, private, and public institutions) is just over $20,000 annually. For engineering, that cost jumps up to $32,000 a year. This is skewed by a large international student population (about 20%) in engineering, but overall represents a monetary barrier to entry.
According to the U.S. Bureau of Labor Statistics, the median annual income for Black, Hispanic, and Native American households fall $20–30,000 below that of White households, and $30–40,000 below Asian households. Those disparate incomes, just on a monetary basis, serve as a preemptive force for Black, Hispanic, and Native American students interested in engineering fields.
The cost prohibitions in place at the collegiate level provide a solid explanation for the racial discrepancies between the U.S. population and the tech industry, but not so much on the gender discrepancies. While women represent 60% of college students, they only represent about a quarter of students in computer science and information types of degree programs. Men and women don’t see nation-wide differences in class so greatly as members of different races, so that part of the blame lies elsewhere.
Ultimately, what the gender gap comes down to is a lack of cultural incentivization for women interested in joining technology-related career fields. The first chapter of Catherine D’Ignazio and Lauren Klein’s Data Feminism discusses the nuances of power within the tech industry. Here they define a notion of “privilege hazard,” wherein which a heterogenous group drives innovation, leaving developers blind to harms and biases present in their products. A secondary effect is that the responsibility to prove and remove these problems often falls onto members of the impacted minoritized groups.
Neither internet companies nor the internet itself are places where women hold much power. From my point of view, asking women to change the environment of the internet in organizations where they are dramatically underrepresented just doesn’t seem appealing. The culture of the industry is toxic to those not represented at the solutions table, and that toxicity hinders those unrepresented from entering the field.
According to the American Association of University Women, there are other cultural factors exclusionary towards women. There is a recursive problem of a lack of role models, and pervasive, wholly untrue stereotypes about women’s abilities to perform in these roles. Additionally, in a survey reported in Fortune, 85% of women reported maternity leave as a factor for exiting the industry. We can see reason after reason for women not to join the tech industry, so it’s no surprise that we have a 3:1 ratio of men to women.
So now we’re here. We have drastic disparity among the creators of the online space that skew its development towards groups who already carry undue power in American society. We can’t blame the hiring practices of giant tech companies because they actually do a pretty good job of matching the demographics of their applicants. We can blame some of the racial misrepresentation on systemic class discrimination between the costs of liberal arts degrees and engineering degrees; and we can blame a lot of the gender discrepancies on toxic power dynamics in the work environment. Where do we go?
Colleges nationwide have made a decent start with financial aid packages based on need, and government student loan assistance has followed suit. The problem of disproportionate racial representation is induced, at least in part, by disproportionate class opportunity when entering university. As such, the clearest solutions include equitable provisions along class lines. To level the representation of women in tech, America needs a cultural shift. The problems for women range from a life of stereotypes to bad options for parental leave to the nature of the platforms they’d work on. There’s no law to wave away misogyny, but we have to create an environment where women know they are welcome and have a chance at a long, fruitful career.
