80% of Data Scientists supposedly hold a Master’s/PhD — Is It Fair As A Minimum Requirement?
4 contributing factors, and what we can make of it moving forward.
The job application process does feel like the above sometimes. And we’re not talking about making it to the top of the company. This is just to get a foot in the door.
There are increasingly more data scientists roles published that require applicants to have a Masters degree, or even a PhD. Some jobs explicitly declare this as a requirement outright, while others may indicate that possessing such degrees would be ‘preferred’. (More on this in Section 1)
Matters which involve differential treatment between different groups of people might be deemed taboo by some. This could be why there’s little discussion on this topic. Specifically, the difference in opportunities between those who hold higher degrees, and those who do not.
I seek to explore this topic from a neutral perspective. You have every right to feel that my perspective would be biased. All I can say is that my name goes out there when publishing this opinion piece. Furthermore, both my parents do not have any university degrees, and I will also put myself in their shoes of having to navigate through the job market without a degree (though the market expectations decades ago were certainly very different).
1. Statistics
According to this LinkedIn post by Bob Moore (over 8000 followers), an analysis of 27,000 education records years ago shows that close to 80% of the data scientists have a Masters or PhD. The same information is shown by Stitch.
Meanwhile, Zippia, which is a source that had been cited by Coursera (though a different article), showed that 51% of the data scientists hold a Bachelor’s, while most of the others hold a Master’s/PhD.
This goes nicely in line with the information stated by indeed.com that 49% of the data scientists openings (on Indeed in the United States within the last three years) require a Bachelor’s, with most others requiring a Master’s/PhD. It won’t be a surprise that a higher proportion of such candidates successfully secured the job.
I am not vouching for the accuracy of these references. (If this were part of my full-time job, I would certainly dig deeper, but for the purpose of an article here, that’s as far as it goes). The information would also change with time. But, without evidence to the contrary, I think the numbers sound plausible and the truth would be somewhere around this range.
2. Setting The Tone
Just as people who pursue higher degrees do so for a variety of reasons, the same is true for people who do not pursue higher degrees.
At no point should we be mistaken that this invisible ‘line’ is drawn between different intelligence levels. In my opinion, it is a line of circumstances as well as preference.
Now that things are nicely set, let’s explore the following factors which may contribute to the fact that Master’s/PhD holders are sought after for data scientist roles. I will explore the reason such a distinction exists to begin with, and give my take on it.
- [3.1] Likelihood of Satisfactory Proficiency
- [3.2] Playing It Safe
- [3.3] Skin In The Game
- [3.4] Company Branding
3.1 Likelihood of Satisfactory Proficiency
If you ask any HR or hiring manager why they explicitly require candidates to hold a Masters/PhD before being eligible to apply for the role, I think a natural answer would be that the job entails highly technical skills, which are expected to be developed through higher degrees.
Consequently, the HR or hiring manager (or other proponents in general) would deem that graduates who have completed a Masters or PhD are better suited to perform the task.
Some opponents would counter that with this strawman argument: “Not all Master’s/PhD graduates are good employees and able to execute their tasks effectively”. That goes without saying, and it is a weak argument. The chance of a less-than-ideal encounter cannot mean that we turn a blind eye to all the potential benefits — the lack of a guarantee does not void a good correlation (if it exists). A more measured discussion will therefore be whether it is more likely for those with a Masters or PhD to be competent Data Scientists. That is, we should compare two complete distributions, rather than selectively singling out specific examples.
However, there is no single universally acceptable metric to begin with. This makes it difficult to perform an objective evidence-based study.
- In terms of comprehension of advanced mathematics and concepts, it is indeed more probable that graduates of a (relevant) higher degree will be able to comprehend the technical details better, because that is what the university programme has trained them for.
- In terms of performance on the job, though, it depends on many factors such as the ability to solve the ‘real-world’ problems according to the company’s constraints of costs, resources and time. These skills can be acquired by those with or without a higher degree, be it from the workplace or outside.
This likely means that at the very least, it doesn’t hurt selecting from a pool of candidates that already possess the higher degrees.
And, if it is indeed true that the company wants only Master’s/PhD holders (justified or not, let’s leave that aside), isn’t it more fair to everyone that it is just specified upfront, rather than waste the candidates’ time?
3.2 Playing It Safe
The hirers are salaried employees themselves.
The interviewer is likely to be a team lead, or at least a senior, and looking for someone to share the workload in their team. The HR, which is probably responsible for posting the job, is trying to fulfil the hiring requirements and getting the required headcount.
In both cases, they need to justify their hiring decision, and would be implicated in one way or another if the hired staff turn out to be a poor fit, or worse, cause problems.
There is less room for blame if the hirer had selected someone certified by a reputable university to have met the minimum academic requirements of the corresponding degree. After all, that candidate had made it through the selection process of the university to even be enrolled in the higher degree, and he/she had passed the examinations involving highly technical contents. By issuing the graduate certificate, the university had, to a certain extent, already done the filtering.
Who would dare to take the risk and hire a Mike Ross (from Netflix Suits)?
The hirers are obviously aware that candidates without a higher degree, or even one without any degree, may be a good fit as a data scientist. However, keep in mind that interviewing candidates is just one of their many responsibilities at the workplace. There will be enough candidates for the hirers to tolerate ‘false negatives’ (ie. missing out on what would have been a match) — it is simply not feasible to interview all the applicants.
Is it fair? Maybe not.
Is it a natural step to take? Probably yes.
In fact, it has long been the case that many professional positions require candidates to possess a relevant degree. The chances of people without a degree getting hired for roles meant for graduates are close to nothing. By shifting the line now to ‘Master’s/PhD only’, the only difference is that a different group of people become affected.
3.3 Skin In The Game
Let’s talk about something tangible, that can absolutely supported by concrete evidence. Something that cannot be disputed.
Masters and PhD graduates showed that they have their skin in the game. Masters students typically spend USD20,000 to USD40,000, as well as at least one year of full-time or two years of part-time studies. Meanwhile, PhD students spend four years of full-time studies, where there’s huge opportunity costs as mentioned here.
So much had been invested. So much had been sacrificed.
Clearly, those with such higher degrees make a clear statement that they are in this game for the long haul. It simply wouldn’t make sense otherwise.
Now, don’t get me wrong. No one is entitled to have an easily-sailing journey simply because they had made an investment. However, this aspect ought to project confidence to potential employers. These employees are not going to throw away their reputation by screwing up in the game. They have much more to lose.
Let me give you an analogy. Suppose you are having a vacation in South East Asia, and are considering between two food places. One is a well-furnished restaurant, and the other is a street stall. (I am not undermining anyone, be it people of different educational backgrounds, or hawkers selling food in humble conditions; just laying out facts as they are).
Sure, food at the restaurant may cost more. And sure, the street stall may possibly be serving very delicious food. Are you going to take the chance? I think we can all agree that the probability of getting a bad stomach after dining at the restaurant is much lower — they risk getting their license revoked and suffering heavy losses.
Is it fair to conclude that Master’s/PhD holders have more skin in the game, and consequently can be expected to persevere through the tough times? To me, it is a yes.
3.4 Company Branding
I personally know of a company with a team of 8 data scientists, all PhD holders. Not that the job requirement mandates a PhD. But it just happens that way. Maybe it is due to the supply of suitable candidates at that time, or a preference on the part of the interviewer, or just pure chance. I don’t know. But it is what it is.
If you have a team where everyone possesses a higher degree, wouldn’t it be more effective in giving confidence to the other stakeholders?
Impression matters a great deal in businesses, and people want to look good. The company’s leadership may want to be able to claim that their data scientists have all gone through higher education in top universities. Call it marketing if you like. We all know how much money is spent on advertisements, as well as how much money advertisements generate for the advertisers.
Can you fault the company for this?
4. What Next?
Having said all these, we need to keep in mind that job descriptions are wish lists from companies. Companies can post any sort of requirements as they deem fit. Whether or not they adhere to it strictly during the hiring process is another matter. If the company is looking for ‘someone that knows 10 programming languages, has 20 publications in top-tier journals, and able to build a secure application useable by a million concurrent users with state-of-the-art model performance hosted on the cloud with fully automated CICD’, chances are that not a single candidate satisfies that.
But, having the bullet point ‘minimum Master’s degree’ in the JD is no deal-breaker at all, and lots of applicants meet this requirement.
There are indeed many legitimate reasons for companies to hire people according to their education background. Fairness aside, what matters is how we move forward. Knowing that many data scientist positions require a higher degree, and are indeed filled by people with such degrees, what can we do about it?
Please don’t get me wrong, but I have to simply call a spade a spade. Well, suppose you are well aware that December is a rainy season, and that the chances of rain is high. You can choose to:
- Stay strictly indoors, or
- Buy an umbrella and carry it along with you to wherever you want to go.
It’s perfectly okay to choose the first option and stay indoors. Honestly. Maybe you have no intention of being outdoors at all, or maybe you are fine either way. The choice is yours to make, but do not blame the weather.
If you want to give yourself the flexibility, simply take the effort (and the money) to buy an umbrella — you only pay for it once and get to keep it with you for a good part of your life.
Options #1 and #2 are about changing yourself, and there’s of course a third option which is changing the world. If you have this action space available to you, for example if you run a large company, you can change the hiring practices. But until then, as a ‘price taker’ the best you could do is play by the rules of the game.
Disclaimer: All opinions and interpretations are that of the writer, and not of MITB. I declare that I have full rights to use the contents published here, and nothing is plagiarized. I declare that this article is written by me and not with any generative AI tool such as ChatGPT. I declare that no data privacy policy is breached, and that any data associated with the contents here are obtained legitimately to the best of my knowledge. I agree not to make any changes without first seeking the editors’ approval. Any violations may lead to this article being retracted from the publication.