A Report on the Age Distribution of Fusion Science Faculty and Fusion Science PhD Production in the United States

University Fusion Association

July 29, 2003July 25, 2003July 24, 2003

Report authored by

Earl Scime, Department of Physics, West Virginia University, Morgantown, WV

Adil Hassam, Department of Physics, University of Maryland, College Park, MD

Ken Gentle, Department of Physics, University of Texas at Austin, Austin, TX

Adil Hassam, Department of Physics, University of Maryland, College Park, MD

Executive Summary

A survey of the demographics of fusion science faculty at US universities is reported.

An assessment of production rate of PhD’s in fusion science is also included.

The age distribution of fusion science faculty (including both engineering and physics faculty) at universities shows a marked imbalance weighted towards older faculty. In contrast, the age distribution for all physics faculty is fairly flat. As one measure, the ratio of faculty in the 55-750 age bracket to faculty in the 30-5045 age bracket (“skewness ratio”) is 1.5about 2 for the fusion science faculty and 1.105 for all physics faculty. This aging of fusion science faculty is even more pronounced when only largerthe(more than five fusion faculty) older, more established (and larger) institutions are considered (MIT, U. Maryland, U. Wisconsin, U. Texas, UCLA, and UCSD). For this grouping, the skewness ratio is about 2.43. The skewness ratio for physics faculty doing fusion science at the “Top 25 in Physics” universities (as defined by USNews) is about 7.

Hiring trends at the above larger institutions suggest that recent and projected fusion science hiring at larger institutions is down: in the last ten years, only 10% of all Assistant Prof. hires at these institutions were in plasma science; and, based on self-reported departmental strategic plans, the hoped-for hiring in fusion science over the next five years indicates a hiring-to-retirement ratio of at most two hires for every three retirements.

The production rate of PhD’s in plasma and fusion science shows a steady decline starting 1986. This decline starts approximately 3 years after the onset of a similar steady decline in the funding level of the US Fusion Energy Sciences Program.

Although objective evidence has been lacking, it has been a common assumption within the fusion science community that the median age of university and college faculty involved in fusion science research has risen significantharply in the past decade and that the number of students obtaining PhD’s in fusion science has likewise significantlyharply decreased. This brief report summarizes a recent self-reporting study (completed in July 2003) of faculty demographics and PhD production in the United States.

Faculty Demographics

A histogram of the estimated age of fusion science faculty[1] at colleges and universities in the United States isn shown below in Figure 1a (in blue). The age calculation is based on the self-reported time since the PhD was obtained plus 27 years. The distribution of actual ages of college and university physics faculty as determined in a recent study by the American Institute of Physics is shown in Figure 1b. The two distributions shown in Figure 1 suggest a significant imbalance towards older ages amongst the fusion science faculty in comparison with the age distribution of all physics faculty in the United States. The most probable age for a fusion faculty member is 65. The median of both distributions is constrained by an effective upper bound on the highest possible age of a faculty member and is 55 for both distributions.

The median age of fusion faculty at Ph.D. granting institutions with relatively new fusion science research programs, (e.g., University of Alaska, West Virginia University, and the University of Montana) is 38. Thus, many of the younger faculty included in Figure 1a are not at institutions with historically large (more than five fusion faculty) plasma research efforts. To distinguish between the fusion faculty age distributions in these newer programs from institutions with larger (and, presumably, older, more established), fusion programs, the fusion faculty age distribution for six institutions (MIT, Univ. of Maryland, Univ. of Wisconsin, Univ. of Texas, UCSD, and UCLA)[2] has been overlaid in red atop the overall distribution shown in Figure 1a. Within these larger r, older fusion programs, the skew towards older faculty is clearly pronounced and suggests that a significant fraction of the fusion workforce at these institutions will disappear without adequate replacement in the next ten years. This imbalance in the faculty age distribution is particularly significant as new fusion science research personnel would typically be trained in academia.We have not yet determined if a similar age distribution exists within the national laboratory

Figure 1. (A) Estimated age (years since PhD + 27 years) distribution of fusion science faculty at colleges and universities in the United States obtained from 24 institutions with a total of 100 faculty. Fusion faculty data from six major centers of plasma physics (MIT, Univ. of Maryland, Univ. of Wisconsin, Univ. of Texas, UCSD and UCLA) are shown in red as an overlay. (B) Age distribution of physics faculty at colleges and universities in the United States [obtained from the talk, “Enrollments and Faculty in Physics” given by Roman Czujko, Statistical Research Center Director, American Institute of Physics, at the University of Maryland in June, 2002]

and private industry fusion science workforce. The imbalance in the faculty age distribution is particularly significant as new fusion science research personnel would typically be trained in academia.

A quick measure of the skewness of the distributions can be obtained by taking the ratio of all faculty in the 55-750 age bins to those in the 30-5045 age bins[3]. For all fusion science faculty (Fig 1a, blue), this ratio was about 1.52; for the “larger institution” data (Fig 1a, red), the ratio was 2.4about 3; in contrast, for all physics (Fig 1b), the ratio was 1.075[4].

Another issue facing fusion science is the decline of programs at top ranked institutions in the United States. At least 15 of the top 25 engineering programs (according to the 2004 U.S.News survey for Engineering) in the U.S. have substantial plasma science programs within their engineering colleges. Of those same 25 engineering programs, only 7 have active fusion science research programs (MIT, U. Illinois-UC, U. Michigan-Ann Arbor, Cornell, UCSD, U. Wisconsin–Madison, and UCLA). Nine of the top 25 physics programs (ranked according to the 2000 U.S.News survey for physics) have active plasma science research programs and all7 of those 9 programs involve a significant fusion science component (CalTech, MIT, Princeton, UCLA, UCSD, U.niv. Maryland–College Park, Univ.U. Texas–Austin, Univ. Wisconsin–Madison, and Univ. of Colorado–Boulder)[5].

We note that the “larger institutions” list separated out above very closely overlaps the Top25 list. Thus, the skewness in the “large institutions” data is representative of the trends at the Top25 institutions[6]. In addition, we have obtained hiring data for recent hires as well as for projected hiring from 11 of the 12 different institutions with fusion science programs at top 25 institutions (as defined above). Those institutions report hiring at least 115 assistant professors in the past ten years (hiring data from Princeton University was unavailable). Of those assistant professor hires, 11 were in plasma physics. Those same ten institutions expect to hire 4 additional faculty in plasmas in the next year and perhaps a total of 10 over the next five years. Given that these institutions currently have 68 faculty engaged in fusion science research and over 15 of those faculty are older than 65 (according to Figure 1a), it is clear that fusion science research at top 25 institutions is not expanding and will likely shrink over the next decade.

Fusion Science PhD Production

Shown in Figure 2a is the number of PhD graduates per calendar year in plasma physics as reported by thirteenen institutions. For those institutions able to provide separate figures for basic plasma science PhD’s and fusion science PhD’s, only the PhD’s earned for fusion science research were included in Figure 2. For the remaining institutions, the total number of plasma science PhD graduates per year was included in Figure 2a[7].

A declining trend in PhD production rate after 1986 is clearly evident. It is relevant to examine, in the same time period, the budget history of the United States Fusion Energy program. This is shown in Fig 2b. The downturn in PhD production after 1986 follows the drop in fusion program funding in 1983 with a three year time lag – roughly the time required for students in the middle of PhD research to complete their degree and graduate.

For comparison, the overall PhD graduation rate in physics for the same time period is shown in Figure 3a and the total funding level for the Office of Science in the Department of Energy is shown in Figure 3b. Comparison of Figures 2a and 3a indicates that the steady decline in plasma science PhD’s is uncorrelated with changes in PhD production in the overall physics community. However, the overall funding level of the Office of Science appears to be significantly correlated (with a 3 – 4 year time lag) with the total number of physics PhD’s produced each year in the United States (Figure 3). The decrease in fusion science funding was also much steeper (more than a factor of two) than the decrease in overall science funding awarded by the Office of Science within the Department of Energy. These comparisons suggest that an important, if not the most important, driver in the production of plasma science PhDs in the United States is the funding level of the Fusion Science program.

Figure 2. (A) Total plasma science PhD graduates per year from thirteen institutions. Data from after 1997 in red includes information from each responding institution. Pre-1997 data is incomplete as not all institutions had data for those years. To account for missing data, PhD production in previous years for those institutions is assumed to equal the PhD production level for the most recent year with data, this likely underestimates the pre-1997 PhD graduate numbers. The contribution from the extrapolated PhD data is shown as open blocks. (B) Funding level of United States Fusion Energy Sciences program in constant FY00 dollars.

Figure 3. (A) Total physics PhD graduates per year (data obtained from the American Institute of Physics PhD graduation statistics web site). (B) Funding level of United States Department of Energy’s Office of Science in constant FY02 dollars (scaled by consumer price index. Data provided by the Office of Science – US DoE).

Appendix A: Survey Form

Fusion Science Demographics Survey

for the Burning Plasma Advisory Committee

Ken Gentle, Adil Hassam, Earl Scime

Institution (including all departments) ______

Person completing survey (please print) ______

Return completed form to xxxxxxxxxx by fax at 304-293-5732 or email at xxxxxxxxxxxxx

  1. How many full-time, tenured or tenure-track, professors at your institution are doing what you would call laboratory plasma physics research (theory and experiment)? Please do not include faculty engaged in exclusively in space plasma physics or high powered lasers used in inertial confinement oriented research.
  1. How many of these professors are doing fusion-science-relevant research (more than 5% of their time) ?
  1. Of the professors doing fusion-science-relevant research, in what years did they obtain their PhD’s ?

______

______

______

______

______

______

  1. How many of these are Assistant Professors ?
  1. How many Assistant Profs has your Dept hired in the last 10 years ?
  1. How many of these have been in a fusion-science-relevant field ?
  1. If your department has gone through some sort of strategic planning exercise recently, how many fusion-science related faculty do you expect will be hired in the next one year ?

7B) In the next five years ?

Year / Total number of laboratory plasma PhD’s: theory and experiment / Number of fusion related PhD’s / #of fusion related PhD’s still in fusion (to the best of your knowledge)
2002
2001
2000
1999
1998
1997
1996
1995
1994
1993
1992
1991
1990
1989
1988
1987
1986
1985
1984
1983

Fusion Science Pipeline Survey

for the Burning Plasma Advisory Committee

Institution (including all departments) ______

Person completing survey (please print) ______

Please note that we realize few institutions will have records going back twenty years. Our hope is to get data for the at least the last five years, the last ten would be great, and the last twenty fantastic. Traditionally, plasmas physics PhD graduates that could most easily transition to work on a burning plasma class experiment have been trained as theorists or experimentalists with a focus on laboratory plasmas or plasma technology. Thus, PhD students trained in space plasma physics, high power lasers used for inertial confinement fusion, etc. should not been included in these numbers. Return completed form to Earl Scime by fax at 304-293-5732 or email at

Appendix B: Significance of Skewness

How significant is the apparent skew in the demographic histograms as compared with randomness in sampling? A crude measure was arrived at as follows:

For each histogram, we estimated skewness by finding the net number of faculty above the mean in the upper age bins (55 to75,60,65,70) and adding this to the net number of faculty below the mean in the lower age bins (30 to 50,35,40,45)[8]. A “skewness factor” was then defined by normalizing this number to the total faculty in the sample. Thus, if all faculty were equally distributed in the four upper bins, the skew factor would be 1. If the distribution were flat, the skew factor would be 0.

As defined, the “skewness factor” for the two histograms in Fig 1a and the histogram in Fig 1b are as follows:

All fusion science institutions (Fig 1a, blue):0.1622

Large fusion science institutions (Fig 1a, red):0.3541

All physics institutions (Fig 1b):0.0302

“Maximum” skewness:1

To get a measure of the statistical randomness in the sampling, we estimated that the statistical fluctuation in faculty count per bin was of order the square root of the mean number of faculty per bin. We then calculated a random skewness factor by assuming that the net number above (below) the mean in all the top (bottom) half bins was this fluctuation number.

The corresponding “random skewness factors” were as follows:

All fusion science institutions (Fig 1a, blue):0.2725

Large fusion science institutions (Fig 1a, red):0.3834

All physics institutions (Fig 1b):0.30

Note that the random skewness factor is somewhat of an overestimate. With this in mind, it seems that the large fusion institutions are significantlyhighly skewed, and all the fusion institutions are significantly skewed compared with all physics institutions though the statistical sampling error for the latter may not be insignificant.

[1] “fusion science faculty” is defined as tenured or tenure-track professors doing fusion-science-relevant research more than 5% of their time; see Appendix A for the survey questionnaire

[2] we defined “larger” as “more than five fusion faculty”

[3]the over-75 and under-30 bins were left out of all skewness measuresthe age 75 and 80 bins were left out of all skewness measures

[4] a measure of how statistically significant the skewnesses may be was also taken. These findings are reported in Appendix B.

[5] Note that this method of counting fusion science programs does not include those institutions that have a highly ranked physics program and a fusion science effort in a lower ranked engineering program (and vice-versa), e.g., the University of Washington.

[6] The 55-750 to 30-5045 skewness measure was also applied to only physics faculty doing fusion science at the “Top 25 in Physics” universities (as defined by USNews). The ratio was about 7.

[7]Note that the pre-1997 data is incomplete; we have used an extrapolation (see Figure Caption)

[8] the overage- 75 and under-30 80 bins were left out of all skewness measures