Gender Parity in NSF Astronomy Research Programs

 

During my first year as a Program Officer in NSF’s Astronomy Division, I was able to compile data on the success rates of different opportunities in the Individual Investigator Programs. As chair of CSWA, one of my top priorities was to look for gender differences.

 

The results* for Astronomy are summarized in the table. The top line is for the Astronomy and Astrophysics Research Grants (AAG), our main grant program. PIs are mainly (but not exclusively) senior and mid-career scientists. There is both good news and bad news here. The good news is that the percentage of female awards (19 +/- 2%) is equal to the percentage of females in the pool (19 +/- 1%). The bad news is that the percentage of women in astronomy at the senior and mid-career levels is still so low. The next line in the table is for Faculty Early Career Development, the CAREER program, which generally makes less than 10 awards per year. The following line is for the NSF Astronomy and Astrophysics Postdoctoral Fellowships (AAPF), which currently makes 9 awards per year. Although the numbers are smaller and the uncertainties are larger, the results for both of these programs agree with those of the AAG. If we do weighted averages of all three programs, we again find that the percentage of female awards (22 +/- 2%) is equal to the percentage of females in the pool (22 +/- 1%).

The raw numbers agree with demographic trends where the fraction of female astronomers at the postdoc and early career levels is significantly higher than the fraction of female senior and mid-career astronomers [1]:

 

-Graduate enrollment in US astronomy departments has risen from 25% in 1997 to 30% in 2006 (NSF-NIH Survey of Grad Students and Postdocs in Science and Engineering).
-The percentage of Astronomy PhDs earned by women in the US has increased steadily from less than 20% in 1997 to almost 30% in 2006 (NSF Survey of Earned doctorates).
-The percentage of women faculty at stand-alone astronomy departments in 2006 was 28% for assistant professors, 24 % for associate professors, and only 11% for full professors (Astro2010 Demographics study group).

 

The good news is that the graduate student-postdoc joint of the leaky pipeline does not appear to be leaking, but the not-so-good news is that the associate professor-full professor joint of the pipeline is not just leaking but gushing. Unfortunately, it is not simply a matter of waiting for these younger scientists to propagate up the ranks. Analysis shows that 30% of prize fellowships in astronomy have gone to women for over 30 years! So women are dropping out of astronomy at significantly higher rates than men, especially at the senior levels. There are several possible reasons for this, including the scarcity of parental leave, limited access to quality and affordable childcare, and a desire for a better work-life balance. There is also, however, the possibility of the toxic work environment. See, for example, the article in the Washington Post: Uncivil work environment pushing women out of the engineering field.

 

The table shows the same numbers* for the NSF Physics division. Although Physics does not have a postdoc program (the equivalent of the AAPF), their main grants program and CAREER programs are roughly equivalent to those in Astronomy. Once again we find that for each program and for the weighted means, the percentage of female awards is equal to the percentage of females in the pool.

 

Why are these NSF results so different from the gender statistics for the Hubble Space Telescope (HST) Telescope Allocation Committee (TAC) reported by Neill Reid? I agree with Neill’s assessment that there is something fairly complex going on with the TAC results. I was especially intrigued by the results that showed that senior women in astronomy have an especially difficult time competing successfully for HST time. This result agrees with those from the groundbreaking unconscious bias study [2], where teams of male and female university psychology professors (search committees) evaluate candidates for an open position (assistant professor of psychology). The application packages for “Karen” and “Brian” are identical except for the name, but the search committees preferred 2:1 to hire “Brian” over “Karen.” Then, when evaluating a more experienced record, say when “Karen” and “Brian” come up for tenure, reservations are expressed 4 times more often for “Karen” than for “Brian.” What would happen when “Karen” and “Brian” were competing for department chair, dean, or provost? These results agree with my own personal experience. I’ve often used the analogy that being a woman in astronomy is like swimming against the current; you can get to a particular destination, but it requires more work. For me, that current has gotten stronger since I entered the senior ranks. There appears to be an extra added bias when considering women as the chair/boss/supervisor/director.

 

The astronomy community is just beginning to learn about unconscious bias. NSF program officers have been giving review panelists an introduction to unconscious bias for less than 5 years. It is difficult, however, to accept and internalize these biases, especially when hearing about them for the first time. Since individuals do not really believe these biases applies to them, I have started asking my panelists to take the Implicit Bias test before they come to the panel itself. Not many do it, but some do, and they are often astonished by their own results. It also helps when I am not the only one in the room talking about bias.

 

Here is my personal pitch: I’m the chair of CSWA; I spend many hours per week on gender-related issues in astronomy. I am unconsciously biased against women! If I can do it, so can you. In fact we all do it, and we do it all the time.

 

Some reviewers are now more aware of their own unconscious biases. In the CAREER and Postdoc programs, many of the PIs are not personally known by the panel members. In order to minimize the effects of their own unconscious biases, reviewers often specifically don’t look for or don’t remember the gender of the PI.

 

One of the main innovations for AAPF is that no letters of recommendation are required (or allowed). Letters of recommendation have been shown to suffer from and impart unconscious bias [3,4]. In one well-known study, the letters for men were longer and had more references to the candidate’s CV and publications. Letters for women, however, were shorter and had more references to personal life. The women’s letters also contained more “doubt raisers” (hedges, faint praise & irrelevancies) such as:

 

-“It’s amazing how much she’s accomplished.”
-“It appears her health is stable.”
-“She is close to my wife.”

 

There is a distinct difference between compliments like, “He is an established leader,” and “She might make an excellent leader.”

 

In addition, NSF programs, especially those for postdocs and CAREER, require that strong “Broader Impact” projects be included as a major component of the proposal. Many of the PIs have significant education and outreach experience. This experience may lessen factors like Imposter Syndrome [5,6] and Stereotype Threat [7,8] since there is no existing stereotype that men are better than women at education and outreach like there is for math and science.

 

[1] Schmelz et al. (2009), Proceedings from the Women in Astronomy III conference, University of Maryland, Oct 21 23, p. 234
[2] Steinpreis, Anders, & Ritzke (1999), Sex Roles, 41, 509
[3] Trix & Psenka (2003) Discourse & Society, Vol 14(2): 191
[4] Martin, Hebl & Madera (2009) Jour of Applied Psych, Vol. 94, No. 6, 1591
[5] Clance & Imes (1978): Psychotherapy: Theory, Research and Practice, 241
[6] Ivie & Ephraim (2011) STATUS, Jan 2011, 4
[7] Steele & Aronson (1995) Jour of Personality & Social Psych, 69(5), 797
[8] Good et al. (2003) Applied Developmental Psychology, 24, 645

 

*The NSF data base was mined for results for 10 years (FY04 to FY13) by Paul Spyropoulos of the NSF Mathematics and Physical Sciences Directorate. The data were separated into three categories: (1) no gender information provided by the PI; (2) PIs self-identified as female; and (3) PIs self-identified as male. Data from category (1) were not used in this analysis.