How is gender bias in science studied? III. Experiments

28 Oct

This is part 3 of my series on gender bias in science. Read Part 1. Read Part 2.

The last method of studying gender bias in science that I would like to highlight is to actually run an experiment. This means that we control the other possible influences, focus on one factor to alter, and see the effect of such change. This also takes away personal opinions on the subject matter, rendering the study more objective.

The following paper did just about that. “Science faculty’s subtle gender biases favor male students” was written by Corinne Moss-Racusina, John Dovidiob, Victoria Brescollc, Mark Grahama,d, and Jo Handelsmana, published in September 2012 in PNAS. I like that the research team consists of researchers from various fields – biology, psychology, management, and psychiatry, which means that appropriate expertise is available for the study (postscript 1).

As you can see from my previous two posts, many studies in this area use existing data or results from surveys and interviews, but they suffer from the shortcomings of personal bias and the potential influences by other factors. Therefore, the paper that we are looking at right now is of tremendous importance. The authors set up an experiment/scenario – the hiring of a laboratory manager position – and asked 127 professors in biology, chemistry, and physics to review the resumes of applicants. The study was double-blinded, meaning that neither those running the experiment, nor the experimental “subjects,” knew which kind of resume was received by a certain subject. The purpose of this is to remove the bias from those in the experiment and those running the experiment. While the same resume was used, a female or a male name was randomly assigned for the resume, making gender the only variable that we are looking at.

Specifically, the present experiment examined whether, given an equally qualified male and female student, science faculty members would show preferential evaluation and treatment of the male student to work in their laboratory.

Even though there has been a lot of focus on eliminating sexism and bias in the workplace, sometimes such values are not exactly conscious decisions – and this is what we call “implicit bias,” which could be found in both males and females. I will come back to this a bit later (postscript 2), but based on the idea of implicit bias, the authors hypothesized that

Science faculty’s perceptions and treatment of students would reveal a gender bias favoring male students in perceptions of competence and hireability, salary conferral, and willingness to mentor (hypothesis A); Faculty gender would not influence this gender bias (hypothesis B); Hiring discrimination against the female student would be mediated (i.e., explained) by faculty perceptions that a female student is less competent than an identical male student (hypothesis C); and Participants’ preexisting subtle bias against women would moderate (i.e., impact) results, such that subtle bias against women would be negatively related to evaluations of the female student, but unrelated to evaluations of the male student (hypothesis D).

Now, compared to bias in other workplaces, looking for gender bias in science is even more conflicting (and to me, this is the reason why a well-design, blinded and controlled experiment is so important):

On the one hand, although considerable research demonstrates gender bias in a variety of other domains, science faculty members may not exhibit this bias because they have been rigorously trained to be objective. On the other hand, research demonstrates that people who value their objectivity and fairness are paradoxically particularly likely to fall prey to biases, in part because they are not on guard against subtle bias.

Anyways, let’s now look at the results. In short (full comic at the end):

2012 Jen Sorensen www.jensorensen.com

2012 Jen Sorensen http://www.jensorensen.com

First bit of bad news – The student’s gender was significant in the 3 measures that the authors look at: competence, hireability, and mentoring, as well as the level of salary offered to the student.

Competence, hireability, and mentoring by student gender condition (collapsed across faculty gender). All student gender differences are significant (P < 0.001). Scales range from 1 to 7, with higher numbers reflecting a greater extent of each variable. Error bars represent SEs. nmale student condition = 63, nfemale student condition = 64.

Competence, hireability, and mentoring by student gender condition (collapsed across faculty gender). All student gender differences are significant (P < 0.001). Scales range from 1 to 7, with higher numbers reflecting a greater extent of each variable. Error bars represent SEs. nmale student condition = 63, nfemale student condition = 64.

Fig 2 from the article. Salary conferral by student gender condition (collapsed across faculty gender). The student gender difference is significant (P < 0.01). The scale ranges from $15,000 to $50,000. Error bars represent SEs. nmale student condition = 63, nfemale student condition = 64.

Fig 2 from the article. Salary conferral by student gender condition (collapsed across faculty gender). The student gender difference is significant (P < 0.01). The scale ranges from $15,000 to $50,000. Error bars represent SEs. nmale student condition = 63, nfemale student condition = 64.

Second bit of bad news – the faculty members’ genders is not a factor; female faculty members did not consider female students more competent or more hireable, and are not more willing to offer mentorship, in comparison to their male colleagues. They also did not offer more salary to female students than the male faculty members did – and this is particularly telling.

From the article: Means for student competence, hireability, mentoring and salary conferral by student gender condition and faculty gender

From the article: Means for student competence, hireability, mentoring and salary conferral by student gender condition and faculty gender

Next, the authors developed a composite competence variable, which allowed a broader understanding of “competence” and its perception by male and female faculty members in this study. (To be super honest, I am not really familiar with this – I would think that this is something done more often in studies in psychology. If you are interested in reading more about this, check out the additional analyses section provided by the authors.)

Via the additional mediation and moderation analyses, the authors found that female students were less likely to be hired because they were viewed as less competent. The authors also asked the faculty participants to complete a questionnaire called the Modern Sexism Scale (additional info on it, supplement info from the article), commonly used to detect the underlying negative attitudes toward women.

Results revealed that the more preexisting subtle bias participants exhibited against women, the less composite competence (β = −0.36, P < 0.01) and hireability (β = −0.39, P < 0.01) they perceived in the female student, and the less mentoring (β = −0.53, P < 0.001) they were willing to offer her.

Last but not the least, the authors examine how much the faculty participants liked the student, and found that while faculty members liked the female student more than the male student, this did not reflect positively toward the perception of the female student’s competence, job offer, salary, or mentoring.

These findings underscore the point that faculty participants did not exhibit outright hostility or dislike toward female students, but were instead affected by pervasive gender stereotypes, unintentionally downgrading the competence, hireability, salary, and mentoring of a female student compared with an identical male.

Overall, this study tells us that while the female students could be well-liked, and that there might not be obvious “hostility” (intentional bias) toward the female student, she would still be considered less competent, less hireable, and would be offered less salary and mentoring in comparison to male students of the same qualification and experience. While the gap might not be huge, the fact that this study is done with the focus on the early career stage means that students who just graduated from science, hoping to stay in science, might be affected. This can trickle up to affect the number of female researchers available for higher level faculty jobs or prominent laboratory positions in the long run. The result from this study also echoes that of another study published in 1999 by Steinpreis, Aders, and Ritzke, looking at how altering the gender of the name on a CV (an academic long-form resume) could affect the perception of qualification of a candidate for a job in psychology. In addition, as the authors suggested:

Because most students depend on feedback from their environments to calibrate their own worth, faculty’s assessments of students’ competence likely contribute to students’ self-efficacy and goal setting as scientists, which may influence decisions much later in their careers.

If this is the case, what can we do? The authors suggested that having better academic policies and mentoring programs, and establishing transparent hiring and admission standards in academia, can help guard against the underlying, unintentional bias against female students. Educating faculty and students about this bias might also help, and should be studied to see if it can be an effective way of eliminating gender bias in science.

I think this is the first paper with a clear indication of gender bias in science that I have seen. Not a lot of good news, indeed, but it is because of research like this that we are better able to look at the real cause of the problem, and find effective means to target gender bias and hopefully kick it out of academia in the future.

Universal Laws of Ladies in Science

In my next post, I hope to look into the suggestions and policies changes that have been implemented, and how we can collectively make a change.

Moss-Racusin C.A., Dovidio J.F., Brescoll V.L., Graham M.J. & Handelsman J. (2012). Science faculty’s subtle gender biases favor male students, Proceedings of the National Academy of Sciences, 109 (41) 16474-16479. DOI:

***

Postscript 1: In my opinion, a study on bias, which is a psychological concept, should be done as a collaboration between physical scientists (because that’s where the problem exists) and psychologists (because of their expertise in studying bias) . The lack of collaboration can result in pre-determined ideas that will affect the results of the study, as well as improper set up of experimental methods.

Postscript 2: Implicit bias is pretty nasty. A group of Harvard researchers have done extensive work on implicit bias. I would recommend that people try this out. Read about implicit bias on their site, and then go to the Test Page, click on “I wish to proceed” and then select the Gender – Career test. Give it a try.

Also take some time to read about how implicit bias can affect women in science in the “Why So Few” overview published by the American Association of University Women (Chapter 8).

Updated Nov 7:  Thanks Artem Kaznatcheev for recommending this post covering the evidence for implicit bias. Implicit Biases & Evaluating Job Candidates (updated) by duffymeg.

Updated Nov 15: An additional resource for implicit bias: CSWA Resource – Unconscious Bias

Postscript 3: There was another experiment that I was going to cover, but I wanted to keep the focus of this post on the PNAS study. Therefore, I took that article out of the post. With that being said, I think it is still worth reading. Its implication would be on how the gender of the researcher and the corresponding subject matter of study can affect one’s perception of the quality of the research.

The Matilda Effect in Science Communication: An Experiment on Gender Bias in Publication Quality Perceptions and Collaboration Interest (subscription required, unfortunately) was written by Knobloch-Westerwick, Glynn, and Huge, published in Science Communication in January.

and reviews done by others:

Advertisements

6 Responses to “How is gender bias in science studied? III. Experiments”

  1. Artem Kaznatcheev October 28, 2013 at 12:13 pm #

    Seems like the experiment was missing an obvious control: CVs with name removed or written in such a way as to obscure gender “J. Blow”.

    A long term intervention based experiment I would like to see done: Select some number of research centres/departments/etc that are relatively equal in demographics, funding, and research output. Select half of them at random to proceed with operations as usual, and for the other half implement the obvious intervention that the study you reviewed suggests: remove names and gender identifiers from all submitted CVs (although, don’t tell applicants about this, just do it in HR). Five years later evaluate the centres at their demographic trends and research output. If there is implicit bias then the controls will have fewer female employees, if that bias is unjustified then the controls will also have worser academic productivity. This sort of experiment would also do more than just observe a problem, but also suggest a potential intervention. Similar studies have been done for blind auditions for orchestras, and have proven to be effective.

    • Terrific T October 28, 2013 at 1:56 pm #

      Good point. I think it would be difficult to have the name removed (if we are looking at gender as a variable), but perhaps a gender neutral name could be use to obscure the gender, as a control. Agreed – would love to see a long term intervention experiment designed and implemented!!

Trackbacks/Pingbacks

  1. How is gender bias in science studied? II. Learning from existing data | Science, I Choose You! - November 17, 2013

    […] This is part 2 of my 4-part series about studying gender bias in science (See part 1, part 3). […]

  2. How is gender bias in science studied? I. Surveys and interviews | Science, I Choose You! - November 17, 2013

    […] is part 1 of my 4-part series about gender bias in science (part 2, part 3). It is not a surprise that I am interested in gender issues in science. As one who has gone […]

  3. How is gender bias in science studied? IV. The future | Science, I Choose You! - December 10, 2013

    […] has been studied: through surveys and interviews (Part 1), through existing data (Part 2), and through experimentation (Part 3). What we have learned is that there is evidence to support gender bias in science, most objectively […]

  4. Women in STEM – 2014 in Review | Science, I Choose You! - December 31, 2014

    […] many of the studies relied on anecdotal data, we do have experimental data demonstrating that junior scientists with female names are less likely to be hired into intro science careers. Some additional digging shows that the research work by the team could be […]

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: