Portrait of a Decade: Results from the 2003 CSWA Survey of Women in Astronomy

By Jennifer L. Hoffman and Meg Urry

Jennifer Hoffman is an NSF Astronomy & Astrophysics Postdoctoral Fellow working with
research groups at UC Berkeley and Lawrence Berkeley Laboratory on 3-D spectropolarimetric
modeling of supernova ejecta. She also organizes local networking, outreach, and social
activities aimed at developing a diverse support structure for women in physics and astronomy
across educational and career levels. Meg Urry is Professor of Physics and Director of the Yale
Center for Astronomy & Astrophysics. Her research is on active galaxies, notably blazars and
obscured black holes. She was the organizer of both the 1992 and 2003 Women in
Astronomy conferences.

June 2004

In the early 1990s, the organizers of the first Women in Astronomy conference at STScI
realized that although anecdotal evidence and their own experience suggested that women were
underrepresented among astronomers, no statistical data existed to help them define or quantify the
problem; past demographic surveys had always combined astronomy with physics or other
sciences. The STScI group therefore set out to conduct the first survey of the gender distribution
of astronomers at major U.S. institutions. Ethan Schreier presented their results at the Women in
Astronomy conference (Proceedings of the Conference on Women in Astronomy, 1992;
www.stsci.edu/stsci/meetings/WiA/schreier.pdf).

In 1999, Meg Urry repeated the survey and compared her results with those from 1992 (STATUS,
June 2000; http://www.aas.org/cswa/status/status jun00.pdf).

As the 10-year anniversary of the first Women in Astronomy meeting approached, the AAS
Committee on the Status of Women in Astronomy (CSWA) decided to conduct the survey again, so
that the results could be presented at the second Women in Astronomy conference at Caltech in
July 2003. Meg Urry provided the original survey’s data, questions, and list of institutions;
Karen Kwitter contacted department chairs and other colleagues for help in gathering the data;
and I compiled and analyzed the results and presented them at the WiA II meeting
(www.grammai.org/astrowomen/stats/poster.html). Full data from all three surveys are now
available online in HTML and Excel formats at www.grammai.org/astrowomen/stats/ (1). Each of the
surveys had a near 100% response rate; together, the three comprise a uniform sample of the major
astronomy programs in the U.S over the past decade.

(1) Those already familiar with these web pages should note that they have recently moved along with me; I am no longer maintaining the old site at Rice University. Please update any links or bookmarks.

With three comparable surveys spanning 11 years at our disposal, we have the opportunity to
investigate whether and how quickly the representation of women in astronomy is changing.
This article presents the results of that investigation. We first discuss the 2003 survey and other recent
snapshots of women in astronomy. In section 2, we compare the 2003 CSWA survey results with
those from the previous STScI surveys to assess changes over time. In section 3, we examine the
past decade’s results for individual institutions. Finally, in section 4, we address the limitations of
these surveys and suggest strategies to guide future efforts in assessing the status of women in astronomy.

1. 2003 snapshot

To ensure that the 2003 data would be comparable with the earlier surveys, we used the
same list of institutions and gathered the same information. Specifically, we asked each department
chair to send us the numbers of men and women in his or her department at the following career
ranks as of May 2003: graduate student, postdoc, assistant professor–faculty, assistant professor–
research, associate professor–faculty, associate professor–research, full professor–faculty, and
full professor–research. When addressing chairs of combined physics and astronomy departments,
we asked for data representing astronomers only. Decisions regarding precisely whom to include
and how to classify them were made by the individual chairs. We received responses from 35
of the 36 institutions in our sample (32 universities and 4 national research centers; see Table 1 for
the full list), representing over 1600 Ph.D. astronomers and nearly 800 graduate students.

Figure 1 shows our 2003 results along with the data from the 1992 and 1999 STScI surveys,
which we discuss in section 2. “Research” and “faculty” tracks are combined in this figure. In
Table 2, we compare our 2003 results for “faculty” only with the most recent survey data for astronomy
from NSF, AIP, into and the Nelson Diversity Surveys conducted by Dr. Donna Nelson at the
University of Oklahoma. (The NSF and AIP studies of bachelor’s and Ph.D. students were comprehensive,
covering several hundred degree-granting institutions each. The AIP study of astronomy faculty surveyed
38 stand-alone university astronomy departments. The Nelson Diversity Surveys considered the "top
50" astronomy institutions, drawn from NRC rankings and the 2003 AIP guide to graduate
programs, including combined astronomy/physics departments but no national research centers.
Although differences in category definitions, years covered, and institutions included give rise to
variations in results among the surveys, the overall picture is relatively coherent.Women currently earn
over a third of the bachelor’s degrees in astronomy awarded in the U.S., make up about 30% of the
graduate population in astronomy, and receive one fourth to one fifth of the astronomy Ph.D.’s
awarded. Fluctuating populations and differing categorization of postdocs and lower-rank faculty
members make for larger uncertainties in these statistics, but women’s representation at these
levels appears to be comparable to that at the current Ph.D. recipient level. However, at the
full professor rank, women make up at most one tenth of the population.

Figure 2 shows the distribution of male and female astronomers across ranks for the 1992,
1999, and 2003 surveys with tenure and research tracks combined. In 2003, nearly half the women
in astronomy (48%) were graduate students, while a tenth (11%) were full professors or equivalent.
Men in astronomy were evenly divided between graduate students and full professors (29%
each). Postdocs, assistant professors, and associate professors made up
roughly the same fractions of the male and female populations. We discuss changes in these
distributions over time in section 2.

Table 3 shows the 2003 CSWA survey results for astronomy faculty members divided between
“research” and “faculty” tracks. We see a difference between the two only at the assistant professor
level, where women appear to be better represented among tenure-track professors. Although the
definitions of these two categories may vary from place to place, especially between universities and
national research centers (the distinction in each case was made by the departmental representative),
we cautiously interpret these data to suggest that women are not currently more likely than men to
hold soft-money positions—in fact, the reverse may be true at the assistant professor level.

2. Changes over time

How has the situation evolved since 1992? Figure 1 shows the numbers and percentages of
women and men at each rank in the three years of the survey. The percentage of women has been
steadily, if slowly, increasing with time at nearly all ranks. The only exception in our surveys is at
the assistant professor rank, where the inclusion of research-track astronomers results in an overall
decrease between 1999 and 2003; see below for further discussion. The number of female graduate students increased from 1999 to 2003, even as the total graduate population decreased to near
1992 levels. Since 1992, then, the increase in graduate women has nearly exactly compensated
for the decrease in graduate men, indicating a true change in the composition of the field (as
opposed to a change in the number of students overall). The total number of astronomy postdocs
increased sharply from 1999 to 2003, with similar growth rates for men and women. At the same
time, the number of assistant and associate professor positions both declined, perhaps
suggesting a tightening job market.

We can use these three datasets to assess whether there have been discrepancies in the past
11 years between the rates at which men and women advance through the academic career
ranks. If it takes roughly 6 years to advance from graduate student to postdoc and from postdoc to
assistant professor, we can assume that in the years between the 1992 and 1999 STScI surveys,
the majority of both the “grad student” and the“postdoc” cohorts who remain in the field had
the opportunity to advance to the next career stage. If the factors governing this advancement
affected men and women equally, the percentage of women should have remained constant across
this transition. That is, we should find that women made up the same proportion of postdocs
in 1999 as grad students in 1992 and the same proportion of assistant professors in 1999 as
postdocs in 1992. (Beyond the assistant professor level, the progression takes varying amounts of
time and is no longer as predictable.) Table 4 shows that this is approximately the case. Dividing the
number of 1999 postdocs by 1992 grad students, we find that 60% of the men advanced to the
postdoc level while roughly half the women did; the error bars are large enough that the difference
is not significant (note that these figures have been corrected from the report by Urry in STATUS,
June 2000). We also see no significant difference in the fraction of men and women (~70% in each
case) who advanced from the postdoc to the assistant professor level between 1992 and 1999. These
data indicate that the women in this sample neither experienced widespread discrimination nor
received preferential treatment as a result of affirmative action policies; instead, they obtained
postdoc and assistant professor positions at a rate proportionate to their presence in the candidate pool.

Since only four years elapsed between the 1999 and 2003 surveys, the situation is not as
clear-cut—that is, the time interval is too short for all members of the 1999 grad student cohort
to have moved on. To perform the same analysis using the 2003 data, therefore, we estimated the
numbers of grad students and postdocs in 1996 by linearly interpolating between the 1992 and
1999 values. Table 5 shows the results, which suggest that the situation may have worsened for
women in recent years. While a larger fraction of women (65%) advanced from the grad school to
the postdoc level between 1996 and 2003 than between 1992 and 1999, the fraction of men who
advanced was even higher (78%). At the transition to assistant professor, both men and women in
our survey saw their chances dwindle: between 1996 and 2003, only 55% of men advanced from
postdoc positions, while the fraction of women was even lower at 40%. We stress that while the
1996 figures are reasonable estimates based on the 1992 and 1999 data, they do not reflect actual
measurements. Nevertheless, these results are troubling; they suggest that even after approximate
gender equity is attained (as seems to have been the case between 1992 and 1999), it does not
automatically persist.

Assuming the percentage of women did in fact decline in both these cases as the 1996 grad
student and postdoc cohorts moved upward, it is logical to ask at what stage(s) the decline
occurred. We estimated from the STScI surveys that 25% of the astronomy graduate students in
1996 were women; the NSF reported this figure to be 26% (NSF, Graduate Students and
Postdoctorates in Science and Engineering: Fall 2001). The NSF further reported that 19–22% of
the 2002 doctoral degrees in astronomy went to women between 1997 and 2002 (NSF, Science
and Engineering Doctorate Awards: 2002), suggesting that these female students may have
experienced greater attrition rates during graduate school than their male classmates. Another possible
explanation for the decline we observed is that the women of the 1996 cohorts remained
disproportionately long at the graduate and postdoctoral levels. The increase in the postdoc
population from 1999 to 2003, combined with the decrease in the assistant and associate professor
populations, suggests that a less favorable job market may also have contributed to the loss of
women from the field during this period.

Our data suggest yet another contributing factor. We see from Table 3 that in 2003 there were
proportionately fewer women in assistant research professor positions than in assistant
tenure-track professor positions; using these disaggregated data, we find that 111, or 33% (±
4%), of the estimated 1996 male postdocs became assistant research professors, while only
13, or 17% (± 5%), of the female postdocs did. By contrast, approximately equal percentages of
men and women (71 men, 21% ± 3%; 18 women, 23% ± 6%) became assistant tenuretrack
professors. These results suggest that while tenure-track hiring still occurs at equal rates for
men and women, either there exists a bias in hiring for research-track positions or women now
have a lower tolerance than men for these positions. Perhaps the women who would
otherwise round out these proportions are deciding to leave the field rather than take soft-money
jobs. However, because we have no information regarding the gender makeup of either the faculty
applicant pool or the group of people who leave the field each year, we cannot assess the relative
roles of these effects in shaping the percentages presented here. In section 4, we discuss what
additional information would allow us to refine our analysis of the academic pipeline in future surveys.

The distribution of male and female astronomers across ranks has changed steadily
since 1992 (Figure 2). Among both groups, the percentage of full professors has increased, while
the percentage of grad students has decreased. Interestingly, the rates of change in both groups
are very similar, so that the population of women in astronomy is still more heavily weighted
toward the young end of the spectrum than is the population of men.

Our survey results indicate that the situation for women in astronomy has improved in many
ways over the past decade. Women’s representation continues to increase at all levels, and the pipeline
appears to have treated women and men equally at least part of the time. This is good news to
those concerned about the status of women in the field. However, we found troubling indications
that women and men do not always progress through the career ranks in proportion to their
representation at lower levels. There are many possible reasons for these differences, which we
can only begin to analyze; our data suggest that one factor may be underrepresentation of women
on the research track. We feel these results should serve as a warning that despite welcome gains in
many arenas, the work of building an equitable environment for women in astronomy is not yet
done; hard-won improvements in the status of women in astronomy require continual maintenance.

3. Individual institutions

We can also use the STScI and CSWA surveys to track the representation of women over time at
particular institutions. Our compilation of 1992, 1999, and 2003 survey data sorted by institution
is available online in HTML and PDF format at www.grammai.org/astrowomen/stats/. For simplicity
of presentation, we combined the research and faculty tracks in creating the compilation table;
disaggregated data for each institution may be found in the 2003 database on the same page. We
stress again that all decisions about classification was done by the departmental representatives.
We also note that although this table includes percentages, these are not useful indicators in
many cases where the sample size is small. We urge caution and common sense in evaluating
these numbers.

We invite all interested parties to read the compilation table carefully and discuss their
institution’s results with their colleagues. In the discussion that follows, we will not name particular
institutions, but we hope that the public availability of these data will encourage widespread dialogue
about the environment for women, both negative and positive, at institutions across the country.
We hope students and job seekers will use these results to inform their decisions, and we hope
those who believe their institutions’ records need improvement will look for ways they can help
effect that change.

Here are some interesting results from this dataset:

• Of the 36 institutions surveyed, there are now (as of 2003) six without a woman
  at a full professor rank. Of these six, two have fewer than three full professors;
  the others have 4, 6, 7, and 12). Four of the six are combined physics/astronomy departments.
• As of 2003, the largest number of female full professors at any institution is 3; the largest number of male
  full professors at any institution is 42.
• Some institutions have dramatically improved the number of women among their
  professorial ranks since 1992, while others have not. Table 6 shows two extreme examples.
  Institution A had no female professors in 1992. It has since promoted two women and eight men to
  full professor rank, and currently has three more women and four more men at lower ranks.
  Institution B has had two female full professors since 1992, which is unusual and admirable.
  However, since then it has hired nine men and promoted five men to the full professor level. No
  women have apparently been hired during this time.
• Table 7 illustrates two other interesting situations. At some institutions, the overall
  percentage of women has increased, but the long-term situation has not improved; at Institution C, the
  number of female faculty members remains nearly unchanged since 1992, while more postdocs of both
  genders have been hired and male full professors have left or retired. At others,
  women move up the professorial ranks, but then apparently leave; Institution D is an example.
• There seems to be no relation between the prestige of an institution and the representation of women
  among its ranks. Neither the 2003 percentage of women among faculty members nor the 2003
  percentage of women among all Ph.D.s in the university departments we surveyed correlates with NRC ranking (NRC, Research-Doctorate Programs in the United States: Continuity and Change, 1995). All
  examples listed here refer to top-30 institutions as ranked by the NRC, and the most extreme
  positive and negative examples given by Fran Bagenal in this issue both come from Ivy League
  schools. We also find no clear-cut trends with institution size.
• Women tend to be slightly better represented at public universities and in standalone astronomy
  departments than at private schools and in combined physics/astronomy departments. The differences
  are not large, but they persist over the three surveys.

4. Comments and recommendations

The 2003 CSWA survey had several limitations. We encountered difficulties in standardizing the
professorial ranks and faculty/research distinction from institution to institution. Though we
attempted to keep the sample as similar as possible to the 1992 and 1999 samples, this occasionally
proved difficult as well; it was not always clear which criteria had been applied in the past, or
even which scientists at a given institution should be considered astronomers! Statistical complications
included small numbers and the possibility of counting people with multiple affiliations more than once.

However, we feel this survey, especially in combination with the previous two, provides a
robust view of the overall progression of women and men into our profession. We urge the CSWA
to continue conducting this survey with the same sample on a regular basis. Ideally, the next survey
should occur in 2005-2006, at which time the 1999 grad student and postdoc cohorts should
have moved up to the next career rank. We entertain fantasies of a world in which each astronomical
institution submits yearly reports to the AAS containing the modest amount of information we
collected for this survey; until this ideal comes to pass, however, one repetition of the survey every
6 years would still provide a valuable long-term record of the gender makeup of the field. We
consider it important to continue gathering information on research or soft-money astronomers, and urge
the CSWA as well as other organizations to include them in future surveys. As we have seen,
variations in the characteristics of these groups can have significant effects on the overall composition of
the field. We also recommend collecting data on yearly Ph.D. production at surveyed institutions;
this would not only allow better analysis of institutional hiring practices, but also inform the
decisions of prospective graduate students.

Like any good scientific study, this survey raised many new questions that were beyond its
scope. Future studies can complement this one by focusing on more narrowly-defined groups and
seeking more detailed information. We add ours to the many voices raised at the Women in
Astronomy II conference in support of a long-term longitudinal study sponsored by the AAS. Only
by tracking the specific career paths of welldefined cohorts of subjects can we address such
issues as the possible differential attrition of women in graduate school or the gender makeup
of faculty applicant pools in a given year. We also point out the importance of soliciting information
from people, especially women, who leave the field; if we simply assume that only the “best”
people remain, we blind ourselves to the potential discovery of forces detrimental to the community
as a whole. Finally, we reiterate our main finding, which agrees with that of Bagenal in this issue:
despite many improvements, men and women in astronomy still experience differential attrition as
they advance in their careers. Simply monitoring the situation with surveys and demographic
analysis will not solve the problem; we urge astronomical organizations and individual members
of the community to continue their efforts to encourage and retain women in the profession,
and to use the information in these surveys as a guide to assessing the environment for women
within the ranks of individual institutions.

Back to June 2004 Contents

Back to STATUS Table of Contents