Now Reading
Does Coming Out in a STEM Workplace Have an Imprint on the Scientific Literature?

Does Coming Out in a STEM Workplace Have an Imprint on the Scientific Literature?

Photo: NCI/Unsplash


  • A new study has reported that scientists who don’t disclose their queer/transgender identities have their names on fewer peer-reviewed publications than those who do.
  • The authors extended this argument to highlight the need for a queer- and trans-affirming workplace environment in the sciences.
  • The study may account for lived experiences of some queer and/or transgender persons in the science ecosystem, several experts told The Wire Science.
  • However, they added that there were several methodological flaws with the paper that undermine its findings.

Hyderabad: For the most part of her B. Tech. degree, Geetanjali Aich lived a double life. As a queer woman, she did not feel safe enough at her institution to assert her queer identity. When she did “come out” – that is, disclosed her queer identity – she was met with harassment and abuse from her peers, which she claims led to her losing valuable time that could have been dedicated to research.

Aich also came out to her parents during her B. Tech. studies. Fortunately, this interaction turned out to be positive and affirming for her, after some initial hurdles. Soon after, Aich got her degree and joined the Indian Statistical Institute (ISI), Kolkata, as a research trainee, where she came out to a circle of close friends.

With their support, she has communicated several research articles in conferences and journals in just a year of her joining the institute, and her productivity is at an all-time high, she told The Wire Science.

A study published recently seems to agree with Aich’s experience. It documented results from two surveys undertaken by researchers Joey Nelson, Allison Mettheis and Jeremy Yoder. Nelson is based at Stanford University, Mettheis is at California State University, Los Angeles, and Yoder is at the California State University Northridge.

In their paper, the team claimed that scientists who don’t disclose their queer/transgender identities have their names on fewer peer-reviewed publications than those who do. The authors extended this argument to highlight the need for a queer- and trans-affirming workplace environment in the sciences.

The study may account for lived experiences of some queer and/or transgender persons in the science ecosystem, several experts told The Wire Science. However, they added that there were several methodological flaws with the paper that undermine its findings.

The surveys and an assumption

The study began in 2013, when the authors floated an online survey to collect data about the number of peer-reviewed papers published by STEM professionals who identified as lesbian, gay, bisexual, transgender, queer or asexual (LGBTA). The survey also sought their gender identities, their job and career positions, and whether they had disclosed their queer identities at their workplace.

Around 600 people responded. Yoder told The Wire Science that one of their most important findings was that “over 40% of people surveyed said that they weren’t ‘out’ at work – that is, their LGBTQIA+ identities weren’t known to more than half of the people they worked with.”

The survey also said respondents who hadn’t disclosed their queer identities at work had significantly fewer peer-reviewed publications than their colleagues who had disclosed their identities.

In this exercise, the authors used the number of peer-reviewed publications by an individual as a measure of their productivity. This is a very narrow definition of the term. Yoder agreed.

“Publication counts are just one measure of ‘productivity’ in research, but they’re an easy one to collect, and for better or for worse they have an impact on career advancement, promotion and research funding decisions,” he said.

A queer physicist in an Indian university, who spoke to The Wire Science on condition of anonymity (for fear of repercussions), said “productivity” in an academic setting must account for teaching and mentorship duties as well.

Anasuith P. Pridhvish, a Young India Fellow at Ashoka University, Haryana, who identifies as a transgender woman, also said that “for a transgender person, dealing with [gender] dysphoria is also being productive.”

Gender dysphoria is a state of unease that several transgender persons face as a result of the incongruence between the sex they have been assigned at birth, the social demands of performing as an individual of that particular sex and their gender identity.

Put another way, by limiting itself to the number of peer-reviewed papers1, the study ignored a large part of the benefits that studying in an academic centre should confer – and thus limits the usefulness of its results as well.

The findings

The team’s findings suggest that there could be what the authors call a “productivity cost” associated with LGBTQA scientists not disclosing (or not being able to disclose, as the case may be) their queer identities at their workplace.

The authors also found a relationship between respondents’ productivity and their gender and academic seniority. That is, while the number of peer-reviewed publications differed significantly between disclosure v. non-disclosure for queer men, this didn’t hold for queer women and non-binary persons.

They also found that people who were higher up the seniority ladder had more peer-reviewed publications and were more likely to disclose their queer identities. The team attributes this to two possible reasons:

  1. That non-disclosure has an effect on productivity, as a result of which non-disclosing respondents may not be able to publish sufficiently to reach senior positions
  2. That more senior and productive scientists might feel emboldened to disclose their queer identities

The authors carried their work forward with a second survey, launched in 2016. This time, along with the previous parameters, they also collected disaggregated data for disclosure of both sexual orientation and gender identities.

They also asked participants the number of years that had elapsed since their first peer-reviewed publication. Finally, they sought responses from cisgender- and straight-identifying scientists as a control.

This time, more than a thousand LGBTQA-identifying scientists and around 600 cisgender- and straight-identifying scientists responded. The authors found that among all gender identities, people who had not disclosed their sexual orientation had fewer peer-reviewed publications than colleagues who had disclosed their sexual orientation at the workplace.

They also found that cisgender men had the highest number of peer-reviewed publications to their credit, while the number of peer-reviewed publications by straight women did not differ significantly from non-binary and queer persons who had disclosed their sexual orientation.

The researchers also found that non-binary and transgender participants reported fewer peer-reviewed publications than cisgender colleagues – although there was no significant difference between participants who had disclosed their non-binary or transgender identities and those who hadn’t.

Now, they wished to know whether disclosure improved productivity by alleviating stress and improving job satisfaction – or if productivity facilitated disclosure by empowering individuals. For this, they hypothesised that the latter would lead to scientists who had disclosed their queer identities to have more peer-reviewed publications than their straight colleagues.

Further, queer individuals disclosing their identities would be expected to show faster academic growth than their straight colleagues.

However, the team didn’t find evidence for either hypothesis – supporting the idea that workplace disclosure of being queer may facilitate productivity rather than the other way.

Problems with the study

The study, a synthesis of the results of the two surveys, adds to a growing body of work that highlights the unwelcoming nature of STEM disciplines and institutions to people of marginalised genders and sexualities.

Several individuals told The Wire Science that the key result of this study – that there is a productivity cost associated with non-disclosure of an individual’s queer and/or transgender identities – is in line with their lived experience.

Along with Aich’s experience, Pridhvish told The Wire Science that disclosing one’s queer or transgender identities could make functioning in the workplace easier by making the individual feel more at peace with themselves and their environment.

The queer physicist said disclosing one’s queer identity could help their colleagues understand the circumstances and the commitments specific to queer individuals better.

This said, Aich highlighted several nuances that the paper doesn’t address. First, she said, different STEM fields have different publication rates. For example, she said it could be easier to publish on, say, computer science and mathematics, than in biology.

Second, she said a queer and/or transgender person may come out not only in their workplace but also to their family and friends. According to her, the paper should have taken this into consideration because the family’s response to one’s queer or transgender identity also affects their mental health, and because the idea of a workplace has shifted during the pandemic.

Third, she said the paper doesn’t report the country or countries from which the survey’s respondents hailed. Revealing this is important, Aich added, because publication rates vary significantly from one country to another.

Finally, she asked about the institution’s response to someone disclosing their queer or transgender identity. For example, while coming out at ISI helped her, doing so at the institute where she pursued her B. Tech. led to bullying and abuse.

Taken together, these concerns discourage interpreting the conclusions of this paper to mean disclosing one’s queer and/or transgender identity should be encouraged in an academic setting.

Statistical assumptions

There may also be quantitative issues. A. Mani, a lesbian scientist at ISI, told The Wire Science that “the paper is replete with unjustified statistical assumptions and computations”.

According to her, the researchers assumed that their data had certain patterns at the outset. This is important.

There are two ways to track data about events. One is to count them in absolute terms: Event Green occurred once; Event Black occurred four times; Event Yellow occurred two times. The other is in terms of their relative frequency – which is the number of times an event occurred divided by the total number of events. So: Event Green has a relative frequency of 1/7; Event Black, 4/7; Event Yellow, 2/7.

When you plot the relative frequencies on a graph, you get the frequency distribution.

Frequency distributions often have characteristic shapes. For example, when events occur at random, the resulting distribution is called a normal distribution. Here, the data near the centre of the distribution occurs more frequently than the data farther from the centre, resulting in a bell shape (see below).

The bell curve of a normal distribution. n denotes the number of events. Illustration: Stpasha/Wikimedia Commons, public domain

There are standard statistical tools to gather insights from normal distributions.

But not all datasets follow these patterns. It’s possible that those in the paper are some such. Yet the researchers have used the same tools to analyse them. The paper has not provided any explanation for why they assume their data to be normally distributed, and that in turn renders their conclusions suspect.

For example, the authors have used a t-test to claim that there is a significant difference between the number of publications by disclosing and non-disclosing queer scientists. A t-test works only within the assumption that the data is normally distributed. Contrary to the norm, the authors don’t clarify why they have assumed it is.

Similarly, they have used a tool called ANOVA to check whether the gender identity of LGBQA scientists confounds the relationship between disclosure and productivity. ANOVA also presumes normal distribution of the data. Without clarification for why this assumption holds true, the finding that the effect of disclosure is significant for queer men but not so for queer women or non-binary persons has to be taken with a pinch of salt.

Mani also said that “the p-values are all unreliable”.

The p-value is a number used to test whether the observed difference between two events is just by random chance, and thus not important. The higher the p-value, the higher the chance that such a difference can be ignored.

For example, the researchers can calculate a p-value to understand the difference between the number of peer-reviewed publications by disclosing and non-disclosing LGBTQA scientists. The lower the p-value of the observed difference, the more important the difference is.

However, many scientists in the international scientific community have criticised the widespread misuse of p-values. (Three journals – Basic and Applied Psychology, Epidemiology and Political Analysis – have even banned their use.) p-values by themselves don’t mean much and may not be required at all, Mani told The Wire Science.

“They have a central role in the replication crisis in psychology and other fields as they can be ‘fixed’ to support a conclusion.”

Mani also said the researchers used “snowball sampling”, in which existing participants of a survey identify more participants. According to her, snowball sampling has demographic bias, and such methods are “not particularly scientific”.

While the researchers wanted to investigate the number of peer-reviewed publications authored by LGBTQA and cisgender straight people, their sample was biased towards the former, she said.

Mani also said there could be finer bias within the LGBTQA community itself, where lesbian women may know more lesbian women, gay men may know more gay men, and so forth. These biases could further confound the study’s results.

She suggested that the researchers redo their analysis using “descriptive” statistics, which are used to investigate – rather than presume – the nature of a dataset, and that they don’t make “unjustified assumptions about a particular kind of distribution of the data”.

Sayantan Datta (they/them) are a queer-trans science writer, communicator and journalist. They currently work with the feminist multimedia science collective TheLifeofScience.com, and tweet at @queersprings.


  1. Thus also overlooking preprint papers

Scroll To Top