Daniela Witten writes:

I recently read a blog post by a (non-statistical) science professor whose opinions I respect and value. Paraphrasing a bit, the thesis was this: If you are a professor, then you should not be surprised that the majority of your trainees (PhD students and postdocs) are less talented /hard-working/brilliant/extraordinary than you were at their career stage. In fact, you should logically expect this to be the case: if you have been successful then you are — almost by definition — exceptional.

Superficially, this argument makes sense, and I certainly cannot speak to whether it is an accurate representation of other areas of science, or of other statisticians’ experiences. But on a fundamental level it does not describe my lived experience as (bio)statistics faculty for almost 14 years.

I believe that academic success in our field requires two things: hard skills and soft skills.

Hard skills are the technical skills that you learn in your academic coursework and employ in your PhD qualifying exams. You continue to develop, expand, refine, and apply these hard skills in your dissertation research, and over the course of your career. These hard skills include proof techniques, programming skills, data analysis experience… whatever it takes for you to go from “Point A: here is a statistical task” to “Point B: that statistical task is now completed”.

It would be hard (read: impossible) to have a successful academic career without hard skills. It goes without saying that if you want to be an applied biostatistician, then you’d better be good at data analysis. And if you want to publish in the Annals of Statistics, then I hope you’re great at proving theorems.

But I’m going to let you in on a secret: you don’t need to have all of the hard skills to have a successful academic career (assuming you can pass quals, of course). If you’re an applied biostatistician then it’s probably okay if you’re not big into measure theory. And if you publish regularly in the Annals of Statistics then there’s no need for you to be even marginally competent at statistical consulting. If you happen to have all of the hard skills, then yay for you!!! But most of us don’t, and that’s 100% fine.

And know that hard skills are learned. Nobody comes out of the womb proving theorems or knowing how to program. Figure out which hard skills align with your interests and abilities, and then develop those skills. And recognize that you will continue to hone those skills over the course of your career: by reading papers, attending seminars and conferences, etc.

What, then, are the soft skills of the job? Everything else. They include the written communication skills to craft a paper that is both clear and interesting; the verbal communication skills to give a good talk; the salesmanship to write a grant proposal or convince a department to hire you; the creativity to pose a good research problem; the vision to know which statistical pursuits will have the highest return on investment; and the grit to see a project through.

Soft skills are often undervalued in our field, and I can understand why: they are difficult to teach via formal coursework, and they are hard to quantify. But they are incredibly important, since the hard skills can only get you so far: what good is proving the theorem if you can’t give an interesting talk about it; if you can’t package it into a paper that other people can understand; and if you can’t follow through on the painstaking (and often multi-year) review process required to shepherd the paper to publication?

What’s that I’m seeing: an eye roll from a reader who thinks that proving minimax bounds is all you need? Well, I’m going to double down. You can never have too many soft skills, and the stronger your soft skills, the better. And guess what: if all you care about are hard skills, then you should really work a bit more on your soft skills, because strong soft skills enable you to quickly and effectively complete the “other” parts of the job so that you can get back to using your hard skills.

Just as we develop and improve our hard skills over the course of our careers, our soft skills are also a work in progress. The first draft of the first paper that I wrote in grad school was a literal mess, and I still cringe when I think about the first talk I gave in my PhD advisor’s group meeting. But I have worked on improving my soft skills alongside my hard skills, and I continue to do so to this day.

Given the huge number of hard and soft skills that go into a successful academic career, it seems clear to me that nobody will “uniformly dominate” anyone else at all of the skills: instead, each person has skills at which they particularly excel, and other skills that remain an area for growth.

So this brings me back to that blog post, which said that as a professor, I should expect that none of my trainees are as talented as I was at their career stage. That is not my experience. My trainees come to me with a variety of skills. Without exception, each has been extraordinary — far better than I was at their career stage — at one or more of these skills. It is my job as an advisor to help each of my trainees further develop their existing strengths, and improve the skills that are not their strengths (yet).

If you are a professor, then every day is a great day to celebrate your trainees’ current strengths, and to help them build up their “not-yet-strengths” (but let’s not call them weaknesses), both hard and soft. And if you are a trainee, then know that there is no one way to build a successful career in (bio)statistics. While there’s always room to improve, you’re amazing, just the way you are.