Being a Statistician in the Age of Data Science: Fleeting Advice from a Mid-Career, Card-Carrying Statistician

white male, beard, mustache, button down shirt, slight smileNick Beyler is the senior director of survey analytics, logistics, and techniques at Fors Marsh. He holds a doctorate and master’s degree in statistics from Iowa State University and a bachelor’s degree in mathematics and economics from Lawrence University. He also serves as a board member for the Wisconsin Chapter of the American Statistical Association.

I love being a statistician. I’m proud to have a career in such an important and highly regarded profession. When I first entered the job market, I was excited to take on one of the hottest jobs out there, according to JobsRated.com. I was constantly saying to friends and family, “That’s what I do! Aren’t you impressed?”

Then about 10–12 years ago, I blinked and data science—not statistics—was the sexy profession, according to Harvard Business Review. And the US Bureau of Labor Statistics Occupational Outlook Handbook says it is projected to continue its dominance. Friends and family often comment that statistics doesn’t seem to be the shiny new object it once was—ouch!

They’re not wrong. In many ways, data science has engulfed statistics. When you search for “statistics” on job boards, you may find a few roles with “statistician” in the title, but the majority are data scientist positions. When you search for graduate programs in “statistics,” programs in data science will inevitably pop up. It wasn’t like this a decade ago. A November 2024 article in Amstat News, “Data Science, Analytics Degrees See Explosive Growth,” shows just how massive the growth in data science has become.

What’s going on? Are statisticians no longer relevant or in demand? Have they been replaced by data scientists? Do companies and research institutions no longer care about using fundamental statistical theory and methods to solve complex problems?

These are rhetorical questions, of course, with a dash of cynicism sprinkled in for good measure. My take: Statisticians are still relevant, and there are actually more opportunities for statisticians to make an impact and build a meaningful career than there were five, 10, or even 15 years ago. Statisticians—especially those just entering the job market or early in their career—just have to know how to adapt. And I have some suggestions.

A few disclaimers before I get on my soapbox:

  • First, my advice (“advice” may be the wrong word; this will be more like a hodgepodge of stories pieced together with some degree of fluidity) is framed largely by how to navigate being a statistician in a data scientist’s world. This includes how to thrive, grow, and make lasting impressions in a world in which we are engaging—and sometimes competing—with colleagues who do not consider themselves statisticians but do consider themselves knowledgeable in advanced statistical concepts and methods.
  • Second, I don’t pretend there is a single tried-and-true approach to navigating a career as a statistician. I have a unique perspective shaped by nearly a decade as a full-time student, followed by 15 years in the professional consulting services industry, with the last five focused on managing other statisticians, analysts, and programmers. I can’t speak about the experience of being a statistician in academia, government, or the tech or pharmaceutical industries. I suspect (actually, I’m 99% confident) my advice will not be applicable to all early-career statisticians, so please take it with a grain of salt.
  • Finally, I know some—maybe even many—will disagree with my advice. I can already picture other mid-career statisticians reading this and rolling their eyes or shouting at the page, “What is this guy talking about?” That’s fine. If there are other stories out there, let’s hear them. More (quality) data is always better for making inferences, right?

Suggestion 1: Take pride in being a statistician, just do it with humility.

I take pride in being trained as a statistician and being a practitioner of statistics. I still remember when I first truly understood the power of the central limit theorem (mind blown). I slogged through courses in measure theory and probability theory (sigma algebras, anyone?), pushing through because I knew it was all foundational to statistics. I thought it was so cool I was taking classes taught by faculty who literally wrote the book on this or that statistical concept. I had a doctoral adviser who once crossed out an entire paragraph of my dissertation because I didn’t accurately specify my model—par for the course, I thought, this will make me a better statistician (and writer) in the long run. I loved hearing stories from the statistical seminars hosted by the Iowa State University Department of Statistics about when hardcore frequentists would argue with up-and-coming Bayesians about where they got their prior distributions from. One such retort was apparently, “I made it up, just like you make up your likelihood function.” I think they’re still scouring Snedecor Hall for the microphone that was dropped after that one.

In my current role at Fors Marsh, I love contributing to impactful research, providing insights into optimal sample designs, suggesting more rigorous approaches to tackling problems effectively, and being a trusted partner among colleagues who respect what I bring to the table. And I love seeing other statisticians making their mark, adding critical value, and getting the praise and recognition they deserve.

Being a card-carrying statistician for so long has also taught me you shouldn’t flaunt your statistical abilities or background. For example, don’t act like you’re smarter than someone else just because you know the difference between a t-test and Z-test. Treat the statistics field as a big tent and be open to other ways people think about statistics or find passion for it, even if you think they are misguided. You will have a more meaningful and fulfilling career as a statistician if you focus on building bridges and not talking down to or around people who aren’t statisticians.

I personally like to make fun of statisticians any chance I can; it can be a good ice breaker. When an economist colleague asked if there really was a dance party during the Joint Statistical Meetings—as if it was crazy statisticians would be associated with a party, let alone one involving dancing—I didn’t take offense. I chuckled and confirmed the rumor was true and said they should come see it for themselves. To kick off a meeting about selecting an optimal sample size for a study design, my favorite quip is, “Go with the biggest sample you can afford. Any other questions?” When someone asks about how statistics fits into training artificial intelligence (AI) models, I like to share the Scooby Doo “Let’s see who this really is!” meme, which reveals the word “statistics” underneath the mask with “AI” written on it. My all-time favorite joke is the one about the three statisticians who go hunting—I’ll let you Google that one.

Suggestion 2: It’s OK not to know everything but be sure you know something.

Undoubtedly, you will encounter someone during your career who thinks that because you are a statistician, you can immediately and accurately answer any kind of statistical question. This isn’t unique to statistics. The other day, I asked my company’s general counsel a question about a lawsuit I saw in the news, and he kindly reminded me he isn’t an expert in all kinds of law.

It’s OK to say “I don’t know” when someone asks you a statistical question. Better yet, say “I don’t know, but let me look into it and get back to you.” This will come across better than pretending you know something or trying to make something up that you think sounds right. Sure, after you say, “I don’t know,” you might get the occasional, “But aren’t you a statistician? Shouldn’t you know that?” My response to that is typically, “Yep, I am a statistician, but I have expertise in some areas of statistics and not much in others. So, I’ll do some research and get back to you. By the way, have you heard the joke about the three statisticians who go hunting?”

Just be sure you know something. You are a statistician after all, and you can’t say “I don’t know” every time a statistical question comes your way (I learned this lesson the hard way during my master’s thesis defense when I forgot the formula for a t-test statistic). Make sure you have a statistical topic, approach, or method you are passionate about and can speak to on a whim. One of mine is survey statistics—or, more precisely, sampling statistics. I have a handy elevator pitch about the different ways to select a sample from a population and what stratification, clustering, or multistage sampling buys (or costs) from a statistical power perspective. Have an elevator pitch—or two—in your back pocket!

Suggestion 3: Be a statistician who knows they sometimes need to be a data scientist.

I have lost my way at times being a statistician in the age of data science. I think it really hit me when I saw colleagues from my graduate program in statistics (including my spouse, whom I met in graduate school) touting data scientist job titles and promoting courses on data science methods such as regression and clustering. I thought, “Aren’t those statistical methods? I thought you were all statisticians. What’s going on? Is this a Twilight Zone episode?”

Then, it happened. I was pulled into a proposal for which one of the key positions required was a data scientist and I had the right educational background and experience to be bid in the role. I went along with it, begrudgingly, and we won the work—either because or in spite of my accolades as a data scientist. Had it happened? Had I joined the cohort of statisticians turned data scientists?

I probably hit peak ‘freak out’ during the spring of 2023. At my company’s annual meeting, I presented a crash course on statistical methods and spent more time than I probably needed to reminding my colleagues that the awesome work Fors Marsh does involving data science is only possible because of statistics. I felt in some ways like the frequentist who screamed at the Bayesian, “Where’d you come up with that prior?!” during a statistical seminar in Ames, Iowa, many decades prior.

After my presentation, I was mingling with colleagues, including one I knew was a proud, card-carrying data scientist (probably just as proud as I am of being a statistician). During an awkward pause in the niceties, I said, “Sorry if my presentation earlier was harsh toward data scientists.” She looked at me, clearly confused, and reassured me I hadn’t said anything offensive. Data science, she pointed out, is a descendant of statistics—just with cooler toys to play with to solve problems. My freak out from earlier quickly subsided and I calmly replied, “Good point. … Want to collaborate on something?”