The war for data science and AI talent is intensifying.
In the midst of this high-stakes race, we are often asked by clients "Which school produces the best data talent? Where should we concentrate recruiting efforts?"
Most prestigious organizations focus on a small set of elite schools. Many job postings list "degree from a top university" as a prerequisite, and the 'prestige of educational credentials' is the most common criteria employers use when reviewing resumes. This approach skews competition for talent towards a very small group of students from top-tier universities.
The result of this hiring practice is consequently reflected in employee demographics within a company. Even in the tech industry, which is thought to be less elitist than the corporate world, Stanford graduates constitute more than 5% of the workforce of Facebook and Google, along with over 10% of Dropbox and Snapchat. Meanwhile, Carnegie Mellon alumni constitute more than 7% of employees at Uber and Pinterest, and 11% of Yelp employees are UC Berkeley alumni (according to Paysa).
It is not surprising that hirers focus on students from elite universities.
Selection bias and/or ingroup bias could explain why people from prestigious schools are cyclically inclined to focus on top-tier schools (or worse, candidates from just their own school). In an HBR article about firms wasting millions of dollars spent on recruiting, one manager perfectly summed up the problem:
"I'm just being really honest, it [an application] pretty much goes into a black hole. And I'm pretty open about that with the students I talk to. It's tough. You need to know someone, you need to have a connection, you need to get someone to raise their hand and say, "Let's bring this candidate in ...There's not an easy way to get into the firm if you're not at a target school."
Companies have good reason to see elite universities as a dependable source of elite talent. Our data, captured from assessing 50,000+ students from 200+ universities, shows that the most prestigious technical universities do produce the highest-quality data prospects. Students from Carnegie Mellon and Columbia for instance had a higher-than-normal probability of showing up in our top 10% of global test takers. While students from Harvard (ranked #1 in our undergraduate rankings), Princeton ( #2), and University of Chicago ( #3), on average, outscored students from other schools.
Even though our findings show prestigious schools to be consistent sources of top-tier data talent, we also found high quality talent to be very widely distributed. Beyond the brand-name schools, there are great places for undervalued talent. More importantly, these less obvious schools can provide a long tail of elite data talent that in many cases goes untapped and undiscovered.
However, while our findings confirm that prestigious schools nestle large pools of top-tier data talent, it also reveals that high quality talent is very widely distributed. Beyond the brand-name schools, there are great places for undervalued talent. More importantly, these schools provide a huge long tail of elite data talent that is untapped and undiscovered.
Regional schools like Baruch College and the University of Pittsburgh, for instance, are great places for data talent, while Statistics, Computer Science, and Aerospace Engineering students from Purdue University (which topped our PhD program rankings) all performed extremely well on our data science assessment.
The question hiring managers should be asking isn't 'Which school is the best for data talent?', but rather 'Where can I find the great data students that other companies don't know about?'.
The chart below (based on rankings published by U.S. News and World Report, The Economist, and the Princeton Review) illustrates test performance across three tiers: students from 30 top technical schools (tier 1), students from the next 50 top technical schools (tier 2), and students from 120 schools typically ranked outside the top 80 (tier 3).
Although tier-1 schools have a higher concentration of elite talent, the total volume of top-scorers is much greater among tier-2 and tier-3 schools, due to smaller student populations at tier-1 schools. To put these pools in perspective, Ivy League schools comprise roughly 59,000 undergraduate students per year, just 0.3% of the 20 million U.S. undergraduates. By maintaining the recruiting process status-quo, firms not only miss out on great untapped talent sources, they also use drive up costs and drive down conversion rates.
The top 20 data science and AI recruiters spend more than $650 million per year. As a result, firms that cannot afford to compete with deep-pocketed tech giants, who decide to focus on tier 1 schools, will likely end up hiring average students. Firms who understand how to assess candidates, and exploit the tier 2/3 school opportunity, could land students with far superior data science and computing skills. Students from tier 1 schools may also command higher salaries and receive multiple offers upon graduating.
Companies should thus devise innovative ways to create a more egalitarian recruiting process, and implement benchmarks that can help recruiters evaluate all candidates on a level-playing field. Doing so, would expand the organization's pipeline and result in better hires.
Correlation One built a platform that allows companies to turn a job description into a tailored assessment, built precisely for the role. By using C1 Assessments, companies can fairly and objectively evaluate candidates, and reduce university bias. This approach is also crucial to maximize the return on investment in the recruiting process. With an expanded funnel and a turnkey assessment process, companies can easily target undervalued candidates that are elite, and more likely to convert into new hires.
For more information about tapping into undiscovered data talent, download our Future of Data Talent report here or get in touch.