During the past two years, the UCAR Board of Trustees and other members of the
UCAR university community have become increasingly concerned about the number
of students applying for graduate school (Anthes, 2000a). A number of us have
seen the pool of applicants shrink at our own Ph.D. programs in recent years.
Furthermore, participants in the 1999 UCAR Forum on the future of the atmospheric
sciences confirmed that there is a serious problem regarding the quantity and
quality of students entering the atmospheric and related sciences (including
oceanography, space science and earth science). In a subsequent UCAR survey
on broad topics, conducted during June and July of 2000, we received more comments
on the quantity and quality of graduate students than on any other issue (Anthes,
2000b). Apparently many university departments around the world are experiencing
similar difficulties recruiting highly skilled students in the environmental
sciences (Brasseur, 2000).
Anecdotal evidence suggests that this decline has led to increased competition
nationally for good students, a shortfall of graduate students to assist in
teaching, and a smaller pool of students to contribute to the university research
efforts. If the decrease is truly widespread and persistent, the number of PhD
graduates could be insufficient to meet national needs in the near future. To
gain a more objective view of the situation, we surveyed UCAR universities specifically
about this issue and investigated related statistics available elsewhere. To
our knowledge, this is the first such study of the Ph.D. pipeline in atmospheric
sciences in recent years, and the results reinforce our concern. The community
may face a significant shortfall in Ph.Ds by the year 2011.
UCAR Enrollment Survey
Our questionnaire, posted on the web in April 2000, requested data from UCAR
member institutions on graduate student enrollments from the 199596 academic
year to 19992000. While five years is a short period for such a study, we wanted
data that were relatively easy to access to assure a large response. We asked
for the number of students who applied, who were admitted, and who actually
entered the program, as well as for the GRE scores of the admitted students.
The students who matriculated were separated into two groups: those with and
without financial support. In addition, we asked respondents to rate on a scale
of 15 how significant they felt this graduate enrollment problem to be, with
1 being insignificant and 5 being very significant. We promised to keep the
identity of the respondents confidential.
Out of 63 institutions contacted, 36 respondeda rate of 57%, which is good
given the relatively large amount of effort required. The average rating on
the 'significance' question was 3.25 and the median was 3.75, indicating that
the respondents considered this issue to be an important one, but not yet extremely
serious. For the 5year period, the annual average of the total number of students
applying to graduate schools at the 36 institutions reporting was 2765, and
the average number entering graduate school was 388.
For all schools combined, applications declined over five years at an average
rate of 8.7% per year, admissions decreased by 1.3% per year, and 3.2% per year
fewer students entered the programstrends computed by least squares fit to
the data in Fig. 1 (see also Table 1). We stratified the data by school size
by sorting the schools in order of the average number of students who entered
in the 5 years. We divided the respondents four groups, each with roughly one
quarter of the total number of entering students^{2}.
The 3 largest schools admitted a fourth of the total number, the next group
of 6 schools another quarter, the next 8 another quarter, and the smallest 19
another quarter. We refer to the four groups here as largest, large,
medium, and small. The average numbers of entering students per
institution for the four groups were 30.7, 18.0, 11.4 and 5.0.
Figure 1
Table 1: 5year trends (% per year) computed from leastsquares fit to data in Fig. 1
APPLY

ADMIT

ENTER


Combined 
8.7%

1.3%

3.2%

3 largest schools 
4.1%

+0.5%

+1.3%

6 large schools 
8.2%

0.8%

6.1%

8 medium schools 
13.4%

6.9%

8.1% 
19 small schools 
4.6%

+4.5%

+3.0%

Because only 57% of the universities contacted responded, the total numbers
of students applying to, accepted by and entering the departments of atmospheric
and related sciences at UCAR universities are larger than these numbers. We
multiply these numbers by 1.75 to estimate the totals for the entire UCAR community,
on the assumption that the sample is representative.
The 8.7% decrease per year in the number of applications represents a decline
of about 240 applications per year at the responding schools alone. The annual
average decrease in the number of entering students is about 12. The rate of
decrease of applications varied by group, but all groups experienced negative
trends. Admissions and matriculations show small and probably insignificant
positive trends for the largest and small schools, but negative
trends overall.
Because the number of applications is decreasing faster than the number of admissions,
the universities are having fewer choices and are therefore possibly becoming
less selective (Fig. 2). The GRE scores, however, reveal no clear temporal trends.
For all schools combined, variations from the mean values shown in Table 2 over
the five years are just a few percent.
Figure 2
Table 2
Mean GRE scores and standard deviations
VERBAL

QUANTITATIVE

ANALYSTICAL


Combined 
548 ± 55

717 ± 33

652 ± 50

3 largest schools 
571 ± 55

722 ± 21

676 ± 38

6 large schools 
534 ± 56

706 ± 40

631 ± 50

8 medium schools 
551 ± 29

731 ± 13

668 ± 24

19 small schools 
547 ± 79

709 ± 53

646 ± 66

The scores in Table 2 would represent roughly, the upper 32, 17, and 33% of
all those taking the GRE and indicating earth, atmospheric, and marine sciences
as their intended field (ETS 2000; Table 4). Interestingly, the average scores
of students heading into the earth, atmospheric and marine sciences are nearly
identical to scores of those going into biological sciences. However, the average
quantitative scores are about 80 points (~13%) below the scores of those in
physics, chemistry or computer sciences (ETS 2000; Table 4).
Alarming Signs
Even though GRE scores don't yet show consequences of a smaller applicant pool,
there are general trends in the sciences that may intensify the size and impact
of a future shortfall of new Ph.D.s. The generally unsatisfactory state of science
education in this country is one of these trends. The proportion of 24yearolds
with degrees in the natural sciences or engineering was near 5.5 % in the United
States compared with 9.5 % in the United Kingdom, 8.9 % in South Korea, and
8.2 % in Germany. Japan and Taiwan are also ahead of the U.S. in this measure.
Of the same group of six countries, the U.S. held second place on this score
in 1975, slipping to last by 1997 (NSB, 2000a, Fig. 415).
Another relevant indicator is the number of bachelor's degrees awarded in various
disciplines. The number of graduates in physical sciences in 1995 was about
20% lower than in 1981 even though the overall number of bachelor's degrees
grew by nearly 20% over the same period (NCES 1998, Indicator 29). This was
the only decreasing field in the survey, which ranged from the humanities to
engineering. A small bit of positive news can be added to this: after ten years
of steady decline, the number of physics bachelor's degrees remained unchanged
in 1998 from the previous year (AIP 2000).
Fortunately, the decline in numbers of physical science graduates is not uniformly
duplicated in trends over the past two decades in the number of graduate degrees
awarded in the atmospheric and related sciences (NSB 2000b, Tables 423 and
425; NSF 2000b, Tables 12, 19, 41 and 43). As Fig. 3 shows, the trends are
quite different in atmospheric sciences than in oceanography. A longterm upward
trend in atmospheric science doctoral degrees appears to have been accompanied
by an equally strong downward trend in M.S. degrees earned.
Figure 3
The trend in Ph.D.s since 1989 is an average annual increase of 3.8%, but the
last two years (199798) are down for the Ph.D. degrees. Trends are more complex
for graduates in oceanography, but these numbers also decreased in the last
two or three years. The rate of increase of Ph.D.s in atmospheric science was
considerably greater than the increase of earthatmosphereocean science Ph.D.s
as a whole or for physicalearthmathematicaloceanbiological sciences overall
(Fig. 4). The decreasing rate of growth in the latter part of this period is
also evident for the other groups.
Figure 4
The AMSUCAR Curricula Guides for 1998 and 2000 (Curricula 1998, 2000) provide
independent data for the number of B.S., M.S. and Ph.D. degrees for twoyear
periods centered around 1996 and 1998. These data (Table 3) complement and extend
the NSF data, although they are not exactly comparable because the time periods
and schools sampled are not identical and the AMSUCAR data include graduates
in oceanography, hydrology and in some cases earth sciences as well as atmospheric
sciences. Nevertheless, the message from the data in Table 3 is consistent with
that from the NSF data. There is a significant decline in the number of undergraduate
and M.S. degrees over the period 19962000.
Table 3 Data from AMS/UCAR Curricula Guide 1998 and 2000
Table 3a: Degrees in atmospheric, oceanic, hydrologic and related sciences granted over periods 19951997 (Fall 1995 through summer 1997) and 19971999 (Fall 1997summer 1999). Number of schools reporting in ( ). "1996" and "1998" numbers are obtained by normalizing the number of degrees granted in each twoyear period to a constant number of schools (50 for B.S., 55 for M.S. and 46 for Ph.D.s) and dividing by two.
Degree

19951997  19971999 
#  # Schools  "1996"  #  # Schools  "199  
B.S./B.A.  1157  (53)  546  816  (45)  453 
M.S.  629  (54)  320  502  (55)  251 
Ph.D.  313  (47)  153  305  (46)  153 
Table 3.b Projections. First column are projections from 1998 Curricula Guide, second column are projections from 2000 Curricula Guide.
Year  B.S.  M.S.  Ph.D. 
9798  511  220  113  
9899  531  226  119  
9900  527  431  221  208  122  118 
0001  599  455  230  219  114  128 
0102  621  514  234  213  132  119 
0203  580  221  128  
0304  583  219  132 
Fig. 5 shows the trends in the numbers of doctorates in the earthatmosphericocean sciences that were granted domestically to U.S. citizens and permanent residents and to nonU.S. citizens with temporary visas (NSF 2000a, Table 3). Since 1994 the former group decreased moderately and the latter group grew more sharply, but overall the 10year trends are close to insignificant.
Figure 5
An Aging Science
Another general trend that could affect a future Ph.D. shortfall is demographic.
The sciences generally are aging. As Fig. 6a shows, the percentage of fulltime
doctoral faculty in science and engineering under age 45 has declined from 62%
in 1973 to 38% in 1997; Fig. 6b shows this shift in age distribution as well.
The percentage of faculty under 35 has decreased even more, from 26% in 1973
to 8.3% in 1997. The same data show that the average age of science and engineering
faulty has risen from 42.5 in 1973 to 48 years in 1997 (Fig. 7).
Figure 6
Figure 7
The most recent data available (Fig. 8) show some similar trends for earthatmosphericoceanic
sciences (NSF 1997, Table 15; NSF 2000c, Table 15; and NSF 2000d, Table 6).
The total number of Ph.D.s in the earth/atmosphere/ocean sciences increased
from 15,580 to 18,360 (17.8% or about 4.5% per year) from 1995 to 1999, but
this growth was not even across all age groups. The under35 yearold category
decreased while the 3539yearold category increased in 199599. This may reflect
a trend of students being somewhat older when embarking on Ph.D. programs and/or
a longer time spent completing the program. In the 4554yearold groups, the
relative proportions decreased, but the actual numbers increased during the
period. The 55andolder groups increased over the period. There is a small
but distinct shift in age distributions, amounting to 0.9 years in average age
over the 4 years (Fig. 9). The lower panel (NSB 2000b, Tables 319 and 625)
illustrates that there is no appreciable difference between the ages of the
overall science and engineering doctoral workforce and the fulltime Ph.D. faculty
holding in the earthatmosphericoceanic sciences.
Figure 8
Figure 9
The aging of Ph.D. scientists overall in this country will continue in the near
future unless more young Ph.D.s enter the field. The spike in the 5559year
age group portends a significant increase in average age over the next ten years
or so. Assuming that younger scientists contribute much of the creativity in
the field, these trends suggest that the vitality of the scientific workforce
may ebb. They also suggest that retirements will increase rapidly in the next
decade, causing a shortfall of Ph.D.s unless the number of Ph.D.s entering the
field increases.
If the 5year decline in graduate applications to departments of atmospheric
science becomes widespread and persistent, the implications are significant
and disturbing and there may not be enough doctorates to meet needs. In an extreme
scenario, the viability of some departments may be threatened. More generally,
the opportunities and challenges presented by the environment, where the atmospheric
sciences play a critical role, demand a steady supply of young, energetic, and
creative talent.
Modeling Future Needs
We have developed a simple model (see sidebar) that estimates the need for future
Ph.D. graduates for the field given a number of quantitative assumptions that
can be varied to produce different scenarios. Obviously any predictions of this
type are uncertain, but the scenarios are instructive. For the base case we
assume that the need for Ph.D.s will grow by 15% over the next decade (NSB 2000b,
Appendix table 328, p. A208). The different assumptions and results for three
scenarios are listed in Table 4. The number of new Ph.D.s needed over the next
decade varies from 675 in the leastneed case to 1,642 in the greatestneed
case.
Let N_{t} be the total number of Ph.D.s needed
in M years, N_{p} the total number of Ph.D.s presently
in the field, and N_{r} the number of Ph.D.s retiring
or leaving the field for other reasons in M years. Then the number of new
Ph.D.s needed over the next M years, N_{g}, is
N_{g} = N_{t}  N_{p}
+ N_{r} (1)
The total number of Ph.D.s needed in M years, N_{t},
is given by
N_{t} = N_{p} (1 + P_{n})
(2)
where P_{n} is the assumed percent increase (or decrease)
in the need. The number of retirements, N_{r}, over
the next M years can be estimated as a function of the number of people in various
age ranges times the probability that the people in this range will retire during
this period. For example,
N_{r} = [P_{R‹56} x P_{‹56}
+P_{R5665} x P_{5665 }+ P_{R66}+
x P_{66+}] N_{p} (3)
where P is the percentage of N_{p} in the three age
ranges and P_{R} is the probability of retirement of
the people in these ranges.
We create three scenarios based on different, but plausible, assumptions about
the need for Ph.D.s in 2011 using the above model. We take N_{p}
to be the number of Ph.D.s in atmospheric sciences in 1997, estimated by NSF
(1997) to be approximately 2,700. From the 1997 age distribution of all science
and engineering Ph.D.s (NSB 2000a, Fig. 313, p 323) we take the percentage
of N_{p} presently 66 years old or older to be 5% and
the percentage of N_{p} in the age range 5665 to be
21%, which leaves 74% of N_{p} aged less than 56 years.
Table 4: Three scenarios based on projected needs for Ph.D.s
in the
atmospheric sciences in the year 2011.
Scenario 
Least Need

Base

Greatest Need

Assumptions  
Pn (% inc needed) 
0

15

20

Prob retirement in 10 yrs 


90

95

100


80

90

100


5

10

20

Results  
Ph.D.s retiring in 10 years 
675

838

1102

Ph.D.s needed in 2011 
2700

3105

3240

New Ph.D.s needed 
675

1243

1642

Avg new Ph.Ds. per year 
68

124

164

Annual turnover rate^{3} 
2.5%

3.1%

4.1%

Another way to look at the situation is to interpret the relatively small variabilities
of the curves in Fig. 9 as indications of a slowly varying age distribution.
If the age distribution were steady state, then, the slope of the lines in Fig.
9 are equivalent to the rate of flow into and out of the population. This slope
is 3% per year, which implies that about 3% of the total number of Ph.D.s in
the field must enter and exit each age interval. According to this rule, for
a steady state total of 2,700 Ph.D. atmospheric scientists, about 81 new Ph.D.s
would have to enter the field each year. This number is between the base case
and "least need" scenarios shown in Table 4. The distribution curves
are slightly steeper, up to 3.8%, if data are used for employed doctoral scientists
only. However, if one assumes no change in retirement practices and in other
factors governing employment, then the total population of degree holders is
the relevant population to consider here. If the number of Ph.D. scientists
grows, then additional assumptions will have to be made about the source of
that growth, namely whether it consists of new graduates or transfers from other
fields. Maintaining the recent (past 4 years) rate of increase of about 4.5%
per year (Fig. 8) entirely from new graduates would require 120 new Ph.D.s per
year in addition to the 81 representing the replacement rate; clearly this is
an upper limit estimate.
The average number of Ph.D. graduates per year in recent years (19951999) varied
between 100 and 150 (Fig. 3) and current projections for the next few years
are also in this range (Table 3b). If this rate persists for the next decade,
there will be a sufficient number of Ph.D.s by the year 2011 for the base and
least need cases. Should either the number of Ph.D. graduates continue to decline
or the "greatest need" scenario occur, however, there could be a significant
shortfall in the number of Ph.D.s.
Implications and conclusions
Monitoring graduate school application and entry numbers can reveal trends well
ahead of measures (such as graduation rates) that are normally used to examine
the supply of a research workforce. This paper can be considered a baseline
study to be updated in about five years, and UCAR will continue to monitor the
situation through annual surveys of members.
In the meantime, shortterm trends and projections into the future are highly
risky. Only a few years ago, the UCAR community was worried about overproduction
of graduate students (Mass, 1996). The universities and the National Weather
Service were hiring very few new graduates. It looked like there might not be
enough suitable jobs for the large and increasing number of graduates in the
atmospheric and related sciences. Some people even suggested that departments
should limit the number of graduates. The community eventually decided that
it was not up to the universities to artificially limit graduates and that growing
opportunities in the private sector could make up for the reduced hiring by
the universities and the government. This decision may turn out to have been
wise, but not necessarily for the reasons given at the time.
It is important to understand the reason for the drop in graduate school applications
and, if it continues, to address the underlying causes. As discussed at the
UCAR Fora in 1999 and 2000, more practical issues, such as the long time and
great effort needed to obtain a doctorate and the relatively meager financial
rewards at the end of the process, may be among the most important reasons for
the declining interest. For example, the salaries of new doctorates in the earth
and space sciences run considerably below those in other disciplines (CPST 1998;
NSF 1997, Table F1).
Whatever the reasons for the decline, we as a community should seek ways to
increase the number of qualified applicants. Because the number of atmospheric
scientists required under any reasonable scenario is small compared to the total
number of students in undergraduate education, a modest increase in the effort
to recruit students from other disciplines could have a major impact in a relatively
short period of time. An important aspect of any strategy is to increase the
diversity in our field. Caucasian males have long dominated the field, but compose
only 33% of the primary and secondary students in the United States. In addition
to all the other good reasons for increasing diversity, the demographics say
we need it. UCAR will respond to these challenges by increasing efforts to recruit
students from diverse backgrounds into the field.
The intellectual excitement of the atmospheric sciences, the importance of the
field to society, and the availability of powerful observational and theoretical
tools to advance the science have never been higher. Our community is not communicating
these opportunities to enough students.
References:
AIP 2000: Enrollments and Degrees Report. American Institute of Physics,
Education and Employment Division. Pub. R151.36 (March 2000).
Anthes, R. A., 2000a: Crises in graduate enrollments? UCAR Quarterly,
Spring 2000, p. 23. http://www.ucar.edu/communications/quarterly/spring00/president.html
Anthes, R. A., 2000b: Results from the UCAR survey of the community. UCAR
Quarterly, Fall 2000, p. 23. http://www.ucar.edu/communications/quarterly/fall00/president.html
Brasseur, G., 2000: IGAC and Education, IGACtivities Newsletter, No.
22, December, 2000. Published by the IGAC Core Project Offfice, MIT, Room 24409,
Cambridge, MA 021394307.
CPST 2000: Salary and Employment Survey 1998. Commission on Professionals
in Science and Technology. American Association for the Advancement of Science.
http://nextwave.sciencemag.org/survey/
Curricula 1998: 1998 Curricula in the Atmospheric, Oceanic, Hydrologic and Related
Scienes. American Meteorological Society and UCAR, 567 pp.
Curricula 2000: 2000 Curricula in the Atmospheric, Oceanic, Hydrologic and Related
Scienes. American Meteorological Society and UCAR, 501 pp.
ETS 2000: Guide to the use of scores. Educational Testing Service, Princeton,
New Jersey. http://www.gre.org/codelst.html
Mass, C.F., 1996: Are we graduating too many atmospheric scientists? Bull.
Amer. Meteor. Soc., 77, 12551267.
NCES 1998: The Condition of Education 1998. U.S. Department of Education,
National Center for Education Statistics. http://nces.ed.gov/pubs98/condition98/
NSB 2000a: National Science Board, Science and Engineering Indicators 2000.
Vol. 1, National Science Foundation, Arlington ,VA.[NSB 001].
NSB 2000b: National Science Board, Science and Engineering Indicators 2000.
Vol. 2, National Science Foundation, Arlington ,VA.[NSB 001].
NSF, 1997: SESTAT (Scientists and Engineers Statistical Data System). Available
on web at http://srstats.sbe.nsf.gov/
.
NSF 2000a: Science and Engineering Doctorate Awards: 1999 (Early Release
Tables). National Science Foundation, Division of Science Resources Studies.
Arlington, VA [November, 2000]. http://www.nsf.gov/sbe/srs/srs01405/start.htm
NSF 2000b: National Science Foundation, Division of Science Resources Studies.
Science and Engineering Degrees: 196697, Arlington, VA [NSF 00310]
http://www.nsf.gov/sbe/srs/nsf00310/start.htm
NSF 2000c: Characteristics of Doctoral Scientists and Engineers in the United
States, 1997. National Science Foundation. Division of Science Resources Studies.
Arlington , VA. Publication NSF 00308. http://www.nsf.gov/sbe/srs/nsf00308/start.htm
NSF 2000d: Characteristics of Doctoral Scientists and Engineers in the United
States, 1999. Early Release Tables. National Science Foundation http://www.nsf.gov/sbe/srs/srs01406/start.htm
.