Physics and astronomy have a representation problem. Despite decades of recognition that there is a problem, persons who lie on at least one axis of non-privilege—e.g., race, gender, disability status, sexual orientation, socioeconomic status—are poorly represented at all levels in these fields, but especially at the top. The reasons for this are many, and those reasons interact in complex ways. The fact that we haven’t seen a lot of positive movement is, in my mind, an enormous civil rights issue, as well as a catastrophe for the progress of ideas in my fields. My goal with this series is to bring together some data and analyses that I’ve found particularly helpful as I’ve been framing problems in my own head for my own institution, and continue conversations about how to make change.
The goal with this first part is to explore the demographics of today. Before diving into the numbers, I want to highlight my go-to sources for data. My top pick for a collection of documents with clear presentation and insightful analysis is the American Institute of Physics‘ Diversity Initiatives webpage. This has been my go-to source ever since my eyes were opened at an American Astronomical Society meeting dinner sponsored by the Committee on the Status of Minorities in Astronomy in 2010 (note that the CSMA has an EXCELLENT website–I’ll refer to some of their work below; I learned today that my colleague Laura Lopez edited their publication, SPECTRUM, for a decade). I’ve been consistently impressed by the work of Rachel Ivie in particular. The most remarkable document I’ve read to date, though, is the report by the TEAM-UP Taskforce, whose charge was to study the crisis in the percentage of bachelor’s degrees in physics awarded to African Americans, and how to make change about it. I will have more to say on it throughout the series. But truly, it is a powerful document outlining an important problem, and recommending not only solutions but the process by which individual institutions can make change. Another professional organization, the American Physical Society, has wrapped data from the Integrated Postsecondary Education Data System (IPEDS) within an interactive query form so that you can track some demographic data from physics departments throughout the country. I’d like to thank former OSU graduate student Mike Lopez for teaching me about this resource. The National Science Foundation (NSF) runs the National Center for Science and Engineering Statistics (NCSES), which publishes data related to STEM fields. This is the most useful resource for data related to disability status (e.g., this report from 2002) and on postdocs (e.g., postdoc demographics by field from 2013) I’ve found to date. Finally, I also find a lot of good ideas and places to start for literature searches on the AAS Committee on the Status of Women in Astronomy and Committee on the Status of Minorities in Astronomy webpages. Out in STEM supports the LGBTQ+ community, and has a fantastic blog featuring members of the community. VanguardSTEM is by and for women of color, and you can watch many conversations on their website. I learn a lot by watching and listening, and would encourage you to do the same.
I acknowledge that it is easiest to find data related to gender (on the binary), race, ethnicity, citizenship, and (to a lesser extent) disability status (NCSES does track this). I have not been able to find systematic data sets related to other privilege markers. If you know of such a data set, please let me know where to find it in the comments below.
I’m going to explore demographics by academic level.
I’ll start at the top of the power hierarchy: faculty. I start here because faculty are both gate-keepers and role-models, they play a key role in setting the tone and priorities of the department and the field. In examining faculty in the United States, Anne Marie Porter and Rachel Ivie of the American Institute of Physics found that in 2014, only 10% of full (most senior), 18% of associate, and 23% of assistant (most junior) professors in physics departments are women (Women in Physics and Astronomy, 2019). The statistics for astronomy departments are slightly better, with 15% of full, 29% of associate, and 29% of assistant professors identifying as women. The numbers have been changing with time, with the 2014 percentages being about 5 percentage points higher in each rank since 2002. However, PhD-granting departments, the ones training the next generation of physics and astronomy faculty, have significantly lower percentages of women faculty than at master’s and bachelor’s -granting institutions—14%. At my own institution, about a quarter of the astronomy faculty are women, and about 10% of the physics faculty are women. Astronomer James Guillochon compiled a list of institutions with astronomy programs and estimated the year in which those departments would have 50% women faculty, assuming that only male faculty retire (no retirement of women), and that 50% of new hires are women. He predicts that our astronomy department will reach parity in 2033, and our physics department in 2044.
At PhD-granting institutions, Porter and Ivie found that only 2% of women faculty members are Black, and 4% of women faculty members of PhD-granting departments are Hispanic or Latina. I could not find reference to Native American/Alaska Native/Hawaiian/Pacific Islander women in this report, which I take to mean that there are effectively none at the faculty level in the United States. Altogether, of the 14% of physics faculty in PhD-granting departments who are women, only 6% are from underrepresented minority groups (URM). I focus on URM groups because of the structural inequality faced by those populations. Asian-American women also face discrimination in ways that white women do not (and have representation problems in particular in astronomy), but in the United States, they are typically not counted as URM in the types of studies performed by, for example, AIP and NSF, for various reasons. For both white and Asian-American women, their representation tends to be better (although clearly still not good) than of URM women relative to the US population as a whole. Back to the URM statistics: URM women are a fraction of (0.14)*(0.06) = 0.0084 = 0.84% of PhD-granting institutions in the United States. It’s less than 50 total women.
In another report coauthored by Rachel Ivie, Garrett Anderson, and Susan White, representation of Hispanic and Black physicists and astronomers were reported, with the latest data analyzed being 2012. In that year, about Black scientists were 2% of physics and astronomy professors nationwide, and Hispanic scientists were 3%. By contrast, non-Hispanic Blacks make up over 12% of the US population, and Hispanics of any race make up 18%, according to the last census. Even in academia, Black professors made up 7% of the professor pool, and Hispanic professors were at 4% of the US professoriate in 2012. So, academia is not representative by race, and physics and astronomy are even less representative. If we focus on PhD-granting institutions again, in 2012 there were about 5,400 physics and astronomy faculty members, of whom about 210 were Black or Hispanic. In 2012, about 34 of those were women.
One key takeaway: if you have applicants in your faculty search from underrepresented groups in your field, PLEASE FOR THE LOVE OF SCIENCE give those applicants a close look. Thinking through the hiring process and identifying places where bias and discrimination can keep people out is important not just for the outcome of your particular search, but for the field as a whole. There is a large literature on how to run a good hiring process; American Physical Society’s webpage on the subject is one place to start. It is essential to make sure that a diverse pool of people are applying to your job. There is significant research on the topic of who applies to what job, but the short of it is that different demographic groups respond to job ads in different ways. Women, on average, tend to underestimate themselves and only apply to job ads where they think they match most of the criteria. Men, on average, will apply even if they don’t match as many of the criteria.
As demonstrated by the numbers above, the hiring of even a few more persons from underrepresented groups can make a noticeable difference to the field. Of course, it does no good to hire people and them not support them, so PLEASE also consider all the ways you can support your new hires. If you are unsure what kind of support to offer, just ask and LISTEN. I will devote a later post on the ways in detail in which representation matters, but I highly recommend reading the concerns and recommendations in the Women of Color in Astronomy and Astrophysics article by Dara Norman et al. in the 2014 volume of the CMSA’s SPECTRUM publication (p. 11) as a place to start on the subject.
As with many things in academia, postdocs draw the short end of the stick when it comes to demographic data and analysis of their demographic. I’ve found a few pieces of data here and there, but I’d love to hear about other sources if you know about them (drop me a line in the comments below). In 2012, the AAS CWSA published a demographic survey in their publication, STATUS. They found that in 2013, 28% of astronomy postdocs were women–a hair higher percentage than of assistant professors (26%) and a bit lower than of the graduate student population (34%). One interesting tidbit from that survey is that the fraction of students and postdocs in a department correlates strongly with the fraction of women faculty. Another reason why representation matters. Unfortunately, while this survey does have a lot of historical data, it does not break down demographics by race, ethnicity, sexual orientation, etc. I came across a publication from the NSF from 2013 on postdoc demographics by field. In 2011, there were 529 postdocs in astronomy, and 2,704 in physics. There were 148 women astronomy postdocs, and 456 women physics postdocs, percentages in line with PhD graduates. Over half of postdocs were foreign nationals. There were only 4 Hispanic and 1 Black (!) postdocs in astronomy that year, and 33 Hispanic and 16 Black postdocs in physics. Hiring committees take note: pay attention to your applicant pool. The fraction of postdocs who are from URM groups is even smaller than in the graduate student pool, as we will see below. It’s difficult to populate faculty jobs with a diverse group of people if there are so few postdocs of color in the pipeline.
There is relatively plentiful quantitative data on graduate and undergraduate students. Again, AIP and NSF publish studies and tables frequently. I’ve also learned a lot by reading reviews in the CSMA’s SPECTRUM publication (see the article by Rachel Ivie here, as well as the article on tribal colleges that immediately follows that article, and my colleague Laura Lopez’s article on faculty retention).
Let’s look at the graduate student data first. The good news is that the percentage of PhDs in physics and astronomy going to women has grown with time; by 2017, about 20% of physics PhDs, and 35% of astronomy PhDs, went to women. This is in line with what we would expect based on bachelor’s degrees going to women, and is an improvement from a decade ago, when PhD rates trailed bachelor’s degrees rates to women. There is one striking difference between graduate and undergraduate astronomy and physics programs, which matters for some of what I describe below. Between 2005 and 2015, the vast majority of bachelor’s degrees in physics and astronomy awarded in the US went to US citizens and permanent residents. There is some evidence this has moved a lot since then, but prior to 2015, over 90% of bachelor’s degrees went to domestic students. However, while astronomy graduate programs are still overwhelmingly populated by US citizens (more than 2/3 in the 2005-2015 window), physics programs are filled by about half domestic students, half international students.
Women of color are very underrepresented in physics and astronomy PhD programs. Between 2005 and 2015, the number of PhDs in astronomy awarded to women went from ~50/year to ~85/year (compared to ~140/year to ~200/year to men). Of US citizens and permanent residents, only about 2 astronomy PhDs per year were granted to Hispanic women in that time, and one every two years to African American women. There were NO astronomy PhDs awarded to Native American, Alaska Native, or Hawaiian women between 2005 and 2015. The representation of URM women is only somewhat better in physics. Between 2005 and 2015, the number of PhDs going to women went from 200 per year (of whom about 100 were US citizens and permanent residents) to about 370 per year (of whom about 170ish were US citizens and permanent residents). In that time, the number of PhDs going to Hispanic women went from about 2 per year to about 8 per year. In that ten-year window, 6 PhDs total went to Native American, Alaska Native, or Hawaiian women, mostly to Hawaiian women. About 2-3 PhDs per year went to African American women, and that number didn’t change throughout the decade. As only 14-30 PhDs per year went to Asian American women, we find that most PhDs going to women in physics go to white women. For some even more startling historical data and interpretation, see the article by Rachel Ivie in this golden 2010 edition of SPECTRUM. It is worth mentioning, and saving for a future discussion, that a lot of “diversity” initiatives disproportionally benefit white ladies like me.
Overall, Hispanic, Black, Native American, Alaska Native, and Hawaiian physicists are highly underrepresented in US physics and astronomy PhD programs. The three-year average between 2014 and 2016 is only 1% of physics PhDs awarded in the US going to Black US citizens/permanent residents (on average 16/year), and 2% to Hispanic Americans (38/year). There are about 4 PhDs in physics per year going to Native American, Alaska Native, and Hawaiian physicists. The rates are comparable for astronomy programs, although the number of PhDs awarded in astronomy every year is an order of magnitude less than in physics.
One observation is that these numbers overall are LOW. So low, that the action of an individual person or institution to support an individual graduate student matters quite a bit. The guiding philosophy of the APS Master’s to PhD Bridge Program is that investing in and providing multiple pathways for students into PhD programs can transform the field as a whole. Of course, a (the?) major underlying issue for poor representation is the systemic racism in US culture, the K-12 education system, college access, and academia (and physics and astronomy) as a whole. I don’t mean racism in the way that white people often mean it (“being mean to people of different races”), but in the deep structural inequality that privileges white people vs. persons of color. Look up “redlining” if you are curious for a moderately contemporary example. But the point for this discussion is, investment in relatively few people goes a long way. It is ESPECIALLY important to have a careful look at the structure of graduate admissions as a field. Even though every institution runs its own process, there are a lot of similarities that run across institutions. The TL;DR is that most of these processes have structural biases that disproportionally affect underrepresented groups. I’m saving a deeper dive on admissions for another day, but I strongly recommend this article in STATUS by Casey Miller on better practices for graduate admissions, as well as this comprehensive book by Julie Posselt.
And finally, we get to undergraduates. The “leak” in the pipeline begins already in high school, especially when it comes to race and socioeconomic status. The high-school stats are sufficiently out of my wheelhouse that I need some time to process it coherently. But yes, when we consider WHY representation for undergrads is so poor, seeing what goes on in high school (and before) matters enormously. Even so, representation is worse for bachelor’s degrees recipients than one would expect based on who takes physics in high school, and so figuring out what to do at the undergraduate level, is to my mind, one of the single most important things that physics and astronomy can do. I mean, we should be advocating for equity in public primary and secondary school as well. But that is a long haul. Doing something about representation and equity at the undergraduate level is, however, something we could do if it were a priority of the field. And moreover, if we want to get a more diverse population through our pipeline, it is essential that we fix undergraduate physics and astronomy.
There are some things to cheer about when it comes to undergraduate representation, but there are even more things that are sad. First, the good news. The rates of bachelor’s degrees going to Hispanic students has gone up in the past decade; as of 2017, about 8% of physics bachelor’s degrees were awarded to Hispanic students. Hispanic students, especially women, are still underrepresented, but the derivative has the right sign. The number of bachelor’s degrees in physics and astronomy has also grown rapidly. Now, the bad news. Very little of the growth in bachelor’s degrees is on account of women. In fact, the percentage of physics and astronomy bachelor’s degrees awarded to women has gone down since I graduated. Just a hair under 20% of bachelor’s degrees in physics go to women. At some point, nearly half of astronomy degrees were going to women, but it’s back down to around 35% again. I’ve been studying this in the context of large public universities, which is the subject of another future long post; what I can say already is that the big undergraduate programs, which are disproportionally located at large, flagship public universities, have significantly WORSE diversity metrics than average. Possibly most alarmingly, the rate of degrees going to Black Americans has gone down with time (now hovering at about 3.5%, but still higher than the rate of Black PhD graduates); this was the reason for the creation of the TEAM-UP taskforce. The rates of Native American participation are very, very low–only 5 bachelors degrees in astronomy, and 26 in physics, went to Native American students in 2013. There is an interesting opinion piece by Calvin John Ortega, Jr., in the 2014 edition of SPECTRUM on the topic. If you are someone who has a lot of money and wants to fix representation problems in physics, I would recommend spending your dollars to support URM physics students, Historically Black Colleges and Universities (HBCUs), and other Minority-Serving Institutions (MSIs). I am very worried about HBCUs.
We can consider how the undergrad stats line up with those of graduate students. The short version is: women overall are basically going to grad school at the rates we would expect based on undergraduate degree recipients, but that means that white women are doing relatively OK compared to women of color. Persons of color (including Asian Americans, but most acutely URM students) are still underrepresented relative to undergraduate physics and astronomy populations. One of the bits of analysis I did to understand the problem was to look at data from 2005 to 2015. Specifically, I looked at the numbers of US citizen students who were awarded bachelor’s degrees in physics and astronomy in 2007 and 2008, and compared the numbers to doctoral degree recipients in 2013-2014, which reflects a mean time to graduation of 6 years (approximately the median degree time), assuming no break. An additional caveat is that I only consider US citizens, because that’s what NSF tracks.
Interestingly, the “yield” in astronomy is quite high–PhD rates are about 40% of the bachelor’s degree rates. This may not be so surprising–many students come in with physics degrees instead of astronomy. However, the rates are quite different between white astronomers and persons of color. White astronomers have a “yield” of 40%–the PhD rate in 2013-2014 is 40% that of the bachelor’s degree rate in 2007-2008. However, for persons of color (all non-white astronomers), the yield is about 25%. It’s even lower for URM students (just under 20%). So, astronomy grad programs, at least as of 2015, were substantially whiter than expected given the undergraduate cohort leading into the program.
The overall “yield” for physics is quite a bit lower–20% overall. This is an interesting stat–it is certainly lower than the number I quoted for astronomy. One thing than my physics department prides itself on is that 30%-40% of our undergraduates go to graduate school. They don’t all go into physics, but still: that’s a lot of OSU undergrads going to grad school, especially given the size of our program. Again, we do see racial differences in cohort, although they are different than in astronomy. The “yield” of Hispanic students in physics is what we would expect based on undergraduate populations, unlike astronomy, where the yield is a factor of two (or more) lower than expected. However, Black and Native American “yield” is about a factor of two lower than expected. The yield of Asian Americans is slightly higher than that of white students, but there are so many more white students than students of other races that they dominate the overall yield statistic.
I have some hypotheses of why this might be. One fact I can point out is that the bachelor’s institutions of PhD recipients in science and engineering depend strongly on race and ethnicity. A large percentage of Black PhD recipients attended HBCUs as undergraduates. As the TEAM-UP Taskforce report points out, this is changing with time, as more degrees are awarded to Black students at Predominantly White Institutions (PWIs). For Hispanic PhD recipients, the major bachelor’s institutions are two campuses each of the University of Puerto Rico, University of Texas, and University of California systems. American Indian and Alaska Native students are few in number, and the two leading bachelor’s institutions are the University of Oklahoma and University of Arizona. By contrast, the major institutions feeding into science and engineering programs overall (which largely means white students) are Berkeley, Cornell, UCLA, University of Michigan, and UW Madison. The big players for students overall are also big players for Asian American students. I don’t know what the data are for physics and astronomy specifically, but I worry that the elite institutions that graduate a lot of white and Asian American science and engineering majors are dominating the game, and those institutions tend to be whiter and with fewer women than average. This is something I will look into, as it meshes well with my current interest in the demographics and structure of undergraduate programs at large public universities.
In summary, persons of color and white women are underrepresented in physics and astronomy, at every level. The leaks in the pipeline are somewhat different between physics and astronomy. In both cases, representation starts out poor for undergraduates, and persists on up. The participation of Black and Native American physicists and astronomers drops between undergrad and grad; it holds steady for Asian Americans and Hispanic Americans in physics, but drops in astronomy (especially for Hispanic students). Retention rates fall again at the postdoc level. There are very few persons of color in faculty jobs, and especially few women of color. In both fields, the participation of women is low. There are, percentage-wise, more women in astronomy than in physics, but the gains are largely seen for white women. While there have been gains for women faculty (again, mostly for white women), the participation of women in undergraduate physics and astronomy programs has dropped since I graduated from college in 2002. It is clear (to me) that without sustained effort, we are not going to see improvement in demographics at any level, and that instead we are likely to see backsliding. We are already seeing it at the undergraduate level.
In future posts, I’ll focus on other aspects of representation. In the near term, expect a study of big undergraduate programs at public universities, as this topic has consumed much of my time since August.
Other data:
Federal STEM workforce in 2011: Pay particular attention to how the demographics shift with seniority
(will add to list with time)