Where are the female physicists?
Robert Ehrlich
The existence and reasons behind gender and racial
under-representation among scientists has been a long-standing concern of both
scientists and policymakers in the
Introduction.
Recently,
considerable attention has been given to the matter of gender bias in science, including
its causes and consequences.1
Even with the continuing existence of such bias, it is worth noting,
however, that women science Ph. D’s are not a rarity. In fact, projections from the data show that women
might be in the majority of science doctorates by the year 2008. Already by 2003 women received 45.8% of
science Ph.D.’s – a 9.8% rise since 1991 – according to the National Science
Foundation.2 Since the percentage of women BS degrees in
science rose 4.1% from 1994 to 2001,3 a corresponding rise at the
Ph.D. level (given an average seven year delay between receipt of BS and Ph.D.
degrees), could put women in the majority of science Ph.D. recipients by the
year 2008 or 2009 at the latest.
Admittedly, women science faculty at universities will still be greatly
outnumbered by their male peers, given that on the average science faculty
received their doctorates many years ago, when women science Ph.D.’s were
indeed a rarity, but that situation also will change over time.
But what is
true for science generally is untrue for certain fields of science, notably
physics, where in 2003 women still earned only 18% of all Ph.D.’s in the U.S. –
albeit a record high (up from a meager 4% in 1972).4 On the other hand, given a rise of only 3.3%
per decade in the percentage of physics Ph.D.’s going to women, it probably will
be many decades if ever before the gender gap begins to close in physics. Given this reality, it is worth trying to
understand better which universities are especially successful in attracting
female physicists (faculty and graduate students), and what factors are
responsible for their success. Two
recent studies in particular are especially helpful: (1) “Women in Physics and
Astronomy, 2005,”4 prepared by the American Institute of Physics
(AIP), and (2) an ongoing online survey conducted by the Committee on the
Status of Women in Physics (CSWP),5 a committee of the American
Physical Society. The latter survey has
elicited 145 responses from departments
of physics or physics and astronomy that grant physics Ph.D.’s, as of July 24,
2007. Both studies relied on
self-reported data supplied by individual physics and astronomy departments at
Ph.D.-granting institutions. The AIP
study makes it clear that the U.S. is not alone in having few women in physics,
and that most countries award less than 20% of their physics Ph.D.’s to
women. With regard to physics degrees at
the baccalaureate level, some readers may be surprised to learn the identity of
the nation having the largest percentage (39%) of physics degrees awarded to
women.6 The AIP study also
identifies ten U.S. schools which have a particularly healthy output of female
physics Ph.D.’s (over 25% during the years 1999 – 2003), and suggests that it
might be worthwhile to learn what these schools are doing right.
Analysis of Data.
In the CWSP
survey respondents provide both written paragraphs describing what makes their
departments “female-friendly,” and quantitative data on graduate students and
faculty. Using the quantitative data we
have looked for factors that may account for high percentages of female
graduate students and faculty in physics in those schools and others. Our results are illustrated in figures 1,2
and 3. As the trend line of Fig 1 shows,
a correlation can be found between schools having higher than average percentages
of female physics faculty and female physics graduate students. Although the correlation is weak (small slope
of the best fit trend line), it is statistically significant at a level of p =
0.0009. It seems unlikely that females
would apply to graduate schools based on their knowledge of how many women
faculty they had, and most of them probably decide on a graduate school using
the same criteria as their male peers.
However, some women might be attracted to schools that had a
“female-friendly” atmosphere based on a visit there, word of mouth, or their
having attended the school as an undergraduate.
Two specific
groups of departments have been flagged in figure 1: (1) those who were
highlighted in the AIP survey as giving more than 25% of Ph.D.’s to women
during 1999-2003 (seven schools shown with open squares), and (2) those six departments
(shown with open triangles) having a female chair based on her first name, according
to a 2007 AIP directory.7
Most of these 13 departments with one or two exceptions would seem to
have unremarkable percentages of female faculty and graduate students, which
suggests that (a) the schools identified in the AIP report as having a high
output of female physics Ph.D.’s during 1999-2003 may simply represent a
statistical fluctuation for many of them, and (b) the presence or absence of a
female department chair is also not highly correlated with large percentages of
female faculty or graduate students.
In order to
learn what factors in fact influence the percentages of females among the
physics faculty and graduate student population, we have looked the departments
of fig 1 which are “above average female representation,” i.e., have more than
10% females among the faculty and also more than 20% among the graduate
students, and those that are “below average female representation,” i.e., have
less than 10% females among the faculty and also less than 20% among the
graduate students. These departments
show an interesting geographic correlation: only one of the 29 “above average”
schools are in the south (not counting the border state of Virginia or the
Western state of New Mexico), while as many as 12 out of 39 “below average”
schools are in the south.
To find
other correlations we have investigated the relationship between the number of
tenured and tenure-track faculty in a department and the percentage of faculty
who are female. Although, we would
expect that larger physics and astronomy departments tend to have more women
faculty in absolute numbers, it is less clear what to expect concerning the
percentages. In fact, as fig 2 shows,
larger departments tend to have lower percentages of female faculty. Given
that the number of females must be integral, the data points of fig 2 all lie
on a set of hyperbolas indicating those departments having 1, 2, 3, … tenured
or tenure-track faculty. My own
institution (flagged with an open square) happens to have the largest
percentage of tenured or tenure-track faculty of all 145 physics and astronomy
departments surveyed (as of July 24, 2007) who have 10 or more faculty members,
although it does not have the largest number, an honor belonging to the
University of Michigan, Applied Physics Department (flaged with an open
triangle), which has 14 female tenured and tenure-track faculty out of a
faculty of 70. The factors accounting
for the high percentage of women faculty at George Mason University are not
unlike some other schools in the CWSP survey.8 The
reason that larger departments tend to have lower percentages of female faculty
is due to simple demographics – they tend to have a larger percentage of senior
faculty, who received their Ph.D.’s many years ago, when female physics
doctorates were a rarity, and when discrimination against women was not
forbidden by law, i.e., before passage of the civil rights act.9
One final
correlation we have investigated is between the “selectiveness” of a graduate
program and its percentage of women graduate students. Selectiveness is shorthand here for schools
that accept a low percentage of applicants to their physics and astronomy Ph.D.
programs. Since the physics Graduate
Record Exam (GRE) plays a significant factor in evaluating graduate applicants,10
and since it has been found that females score more poorly on this exam by on
the average 150 points,11 one might expect that female graduate
students are scarcer at more selective institutions – at least to the extent
that they rely on GRE scores as a significant factor in admission decisions. Surprisingly, however, the trend – albeit a
weak one – is in the reverse direction, as can be seen in fig 3, with more selective
institutions having on the average higher percentages of female graduate
students. Thus, suppose we arbitrarily
define “more selective” institutions as those admitting fewer than 20% of
applicants, and we also define “many” female graduate students, as more than 25%
women. Figure 3 shows that among more selective
schools 36% have many female graduate students, while among other schools only 21%
have many females. One can imagine many
possible explanations for this surprising correlation, including (a) greater
percentages of females than males applying to more selective schools, (b)
greater recognition by selective schools that the physics GRE has limited
utility in predicting success in graduate school,12 and (c) greater
possibilities of scholarships for talented female Ph.D. students. (We put little credence in the idea that the
better programs are unfairly accepting more women than are justified from GRE test
scores, just for the sake of making
their statistics look good.)
If the
correlation between program selectivity and percentage of women graduate
students is genuine,13 it is
a piece of good news, since females and males have comparable dropout rates in
physics graduate programs.4
It means that future women physics Ph.D.’s will on the average come from
more selective schools than their male peers, and will be at a competitive
advantage in the hiring process. Additionally,
such a correlation undermines the misguided belief that women tend to be
ill-suited to be in the forefront of physics.
Fig 1. Dependence of the percentage of physics graduate students who are female on the percentage of faculty who are female. The meaning of the data points shown by squares and diamonds is discussed in the text.
Fig 2. Dependence of the percentage of women physics faculty on the department size.
Fig 3. Dependence of the percentage of
graduate physics students who are female on the selectivity of the program.
References
1. Maxine Singer, et al., Beyond
Bias and Barriers, Science 314, 893 (2006).
2. Chapter
2, “Science Indicators, 2006,” National Science Foundation,
3. See ref 2. Note that “science” as defined by the
National Science Foundation includes social science as well as natural science,
however, the percentages of women in the social sciences are actually lower
than those in science generally at the bachelor’s level.
4. R. Ivie
and K.N. Ray, “Women in Physics and Astronomy, 2005,” AIP publication Number
R-43002, Statistical Research Center, The American Institute of Physics, College Park, MD, February, 2005.
5. “Survey
by the Committee on the Status of Women in Physics,” See: http://cswp.womeninphysics.org/results.php. For each institution, the questionnaire asked
the following: How many tenure-track or
tenured faculty -- male/female? How many graduate students -- male/female? Is there a family leave policy for graduate
students? If so, describe. Is there family health insurance available
for graduate students? Is it included in
the stipend? In a paragraph, please
describe why someone applying to graduate school who is interested in a
female-friendly department should choose your institution.
6.
7. “2007
Graduate Programs in Physics, Astronomy and Related Fields,” American
8. The main
three factors include a welcoming attitude by both male and female faculty to
women in senior positions of leadership, an existing critical mass of highly
successful female faculty, and an
understanding policy regarding maternity and paternity
9. E.
Mielczarek, letter to the editor, Science 27, October 2006 314: 592.
10. This
claim can be verified based on the large number of schools that either require
or strongly recommend that graduate applicants take the physics GRE in the 2007
A.I.P Directory of graduate programs. An
informal sampling of 10 department chairs indicates that in all but one case,
“some” or “major” weight is given to the GRE score in admissions decisions –
this despite the fact that a majority of the chairs polled were aware of the
research claiming that the GRE score is not a good predictor of later success
in graduate school or in research. The
chairs were evenly split on whether the score is useful for all applicants or
only those who seem marginal or come from schools with an unknown reputation
11. APS
News, The American Physical Society, July 1996.
12. A quick informal survey of the web sites of a
number of physics departments showas that many do require or strongly recommend
the physics GRE exam for applicants.
However, they also usually say that there is no minimum score required on
the GRE, and that top grades and strong recommendations outweigh GRE scores. In a thoughtful essay, Howard Giorgi, former chair of the
Harvard University Physics Department offers a number of reasons why this may
be the case in the newsletter: “STATUS, A Report on Women in Astronomy,”
January 2000. Moreover, as reported in
ref 6, a study of Harvard graduate students in physics found that “while there
was a slight correlation between GRE scores and graduate course grades, there
was no correlation with other measures of success in graduate school, including
oral exam scores, and overall completion time for the Ph.D. degree.” Giorgi further notes that the GRE was
most useful in judging applicants from small colleges where the University had
little or no experience with previous students.
13. It may
not be because it depends largely on four schools having acceptance rates
greater than 60%, and it might depend on confounding variables, such as the
percentage of foreign applicants.