Keith Crank, ASA Research and Graduate Education Manager

Keith Crank earned a BS in mathematics education and an MS in mathematics from Michigan State University and a PhD in statistics from Purdue University. Prior to joining the ASA as research and graduate education manager, he was a program officer at the National Science Foundation, primarily in the probability program.

I want to begin this article with three questions, addressed to three audiences:

1. Department chairs/heads: How difficult is it to fill your faculty vacancies?
2. Faculty with PhD students: How difficult is it for your PhD students to find a job when they graduate?
3. Recent PhD graduates: How difficult was it for you to find a job?

A recent issue of Nature contains a number of articles about the PhD. Many question the value of a PhD, especially as it now exists. A primary concern of the authors is whether there are too many PhDs being produced and whether there are enough jobs for them. A problem with these articles is that all disciplines are lumped together and the difficulties of a few are viewed as a sign that we need to change everything.

Are we producing too many statistics PhDs? I don’t think so. Let’s begin by looking at some numbers. (Various reports from the American Mathematical Society (AMS) are used here) According to the AMS, there are about 1,800 statistics and biostatistics faculty in PhD-granting departments of statistics and biostatistics. If the typical faculty member has a 40-year career, we would need 45 PhDs per year to replace those who are leaving (in a steady-state scenario). But, these are not the only statistics faculty. There are also faculty at colleges and universities that do not produce PhDs in statistics or biostatistics. To estimate the number of faculty needed in those places, I will use student enrollment data.

At the universities that have PhD-granting departments of statistics or biostatistics, there were about 2 million students enrolled in the fall of 2009. There were more than 3.5 million additional students at PhD-granting universities that do not have a PhD-granting department of statistics or biostatistics. This suggests we need an additional 78 PhDs to fill those faculty positions. That would bring us to a need for 123 new PhDs each year.

But, we haven’t included colleges that don’t offer a PhD. There are roughly an equal number of students enrolled in those institutions as there are at the PhD-granting universities. Recognizing that teaching loads are generally higher in these institutions, let’s say we would need half as many faculty members (61) to teach the same number of students. That brings us to 184.

Based on the AMS surveys, nonacademic employment absorbs another 125 or more new statistics PhDs each year and roughly 50 leave the United States. So, the minimum number of statistics and biostatistics PhDs needed each year is slightly more than 350.

But, there are many reasons to believe this is not enough. To begin with, steady state suggests PhD-granting departments of statistics and biostatistics should be hiring about 45 new faculty members each year. Using AMS data again, these PhD-granting departments of statistics and biostatistics actually lost only 35 faculty to retirement and death (total for the last two years), but they filled approximately 115 faculty positions in each of the past two years. While some of these are likely due to movement of faculty from one university to another, that alone would not explain the difference. According to the AMS, the PhD-granting departments of statistics and biostatistics increased by more then 30% between 2000 and 2009 (or an increase of about 50 faculty members per year).

These numbers suggest 400 new PhDs in statistics and biostatistics is the minimum we should be producing. I view it as a lower bound because I did not include growth of statistics faculty in other mathematical sciences departments, nor did I include academic jobs in departments outside the mathematical sciences. (In 2009, 75 PhDs from statistics and biostatistics departments took academic jobs in departments outside the mathematical sciences.) In addition, U.S. government statistical agencies say they would hire more, if they were available. We could probably produce 500–600 per year without too much concern about them finding jobs.

To contact me, send an email to keith@amstat.org. Questions and comments about this article, as well as suggestions for future articles, are always welcome.