Today is the second International Day of LGBTQ People in STEM, and I’m reminded of a story a colleague told me after I took my first faculty job seven years ago. Before deciding whether to interview me, “we made some calls,” the colleague said, “and we found out you were gay.” Because I was gay, the search committee thought there’d be no way I’d ever come, and so they almost passed on interviewing me, my colleague explained. It was an excellent institution, but they knew it lacked diversity and was far from urban life. “But we took a shot, and you’re here!”

At first blush, it may seem appropriate for departments to consider that candidates from underrepresented groups may find it challenging to thrive in places lacking diversity. After all, hiring decisions often involve lifelong faculty appointments, and a candidate’s fit can have decades of implications or waste a department’s precious funds and time. But it’s a problem when departments use perceived “fit” as veiled discrimination—whether intended or not.

In many STEM departments today, it would likely cause outrage to pass on a female or racial minority candidate simply because recruiting and retaining someone of their gender or race might be challenging (and it would also be illegal). But still today, departments might pass on an LGBTQ candidate—not explicitly because they’re LGBTQ—but because stereotypical assumptions about being LGBTQ (e.g., only living in urban locations) raise questions about a candidate’s “fit” for the departmental, institutional or local culture. And perceived “fit” is likely even worse for those multiply underrepresented by gender, race and LGBTQ identity.

These barriers to entry for LGBTQ people aren’t limited to the faculty-search level, either. Students, trainees and faculty members regularly encounter implicit biases that go under the radar, often from scientists with the best of intentions. When a visibly out undergraduate is deemed too “unprofessional” to be admitted to a PhD program, that’s bias. When a postdoc’s gender transition is taken to mean she’s “indecisive” and not yet ready for the tenure track, that’s bias.

When a trainee gets nine faculty-job interviews, briefly references being a lesbian in her job talk and doesn’t receive a single offer, that’s likely bias too. Besides barriers to opportunity, there are many other unintended ways LGBTQ people can be made to feel unwelcome in STEM. When I was starting out, for example, a colleague took the time to pull aside a postdoc candidate for my lab to let him know that I was gay, in case that might be a problem.

Although few studies on LGBTQ people in STEM exist, the growing evidence is sadly unsurprising. Estimates suggest LGBTQ people are roughly 20 percent less represented in STEM fields than expected. When LGBTQ people persist in STEM, they report more negative workplace experiences than their non-LGBTQ counterparts, and roughly 70 percent of out STEM faculty members report feeling uncomfortable in their department.

The leaks and cracks in the STEM pipeline start quite early for LGBTQ people. Sexual-minority undergraduates are more likely than their heterosexual peers to leave STEM majors, even when accounting for other demographic and educational factors. And it’s not that they’re uninterested; if anything, they show greater signs of interest in STEM than their heterosexual peers.

It’s hardly news that implicit bias exists in the scientific community; research on other underrepresented groups is clear. STEM fields in which scientists harbor stronger stereotypes about gender and racial differences in scientific ability have lower representation of women and black people. In applying for grant funding from the National Institutes of Health (NIH), black scientists are less successful than white scientists, even after accounting for academic pedigree and publication record. So too for women relative to men, but only at later career stages. And white men continue to be disproportionately invited onto panels at scientific conferences, recently leading the NIH director to boycott “manels” (all-male panels). None of these biases are necessarily intentional (although they can be), but their effects are insidious.

Most important is that we only know all this, and only have a means to evaluate our progress forward, because STEM institutions—government, universities, funding agencies and scientific societies—track gender and race. But while all the emerging evidence suggests LGBTQ people are encountering similar biases as other underrepresented groups in STEM, we lack the official data and means to hold STEM institutions accountable for LGBTQ biases in a similar way.

When the 1980 Science and Engineering Equal Opportunities Act was signed, it charged the National Science Foundation (NSF) with ensuring the participation of underrepresented groups in U.S. STEM fields and helping protect against the biases that hinder it. To track progress, NSF has collected nationwide data using several surveys that serve as a kind of “STEM census.”

The data feed into detailed reports used by funding agencies and institutions across the country, and diversity efforts and policies in higher education depend on them. They’re the only way we know, for example, that over the past 20 years the percentage of female and black undergraduates has actually decreased in the fields of math and statistics; that although in most other STEM fields black people are increasingly represented, they still experience pay inequality; or that computer science suffers from the worst underrepresentation of women.

NSF’s STEM census collects information on countless measures—gender, race, ethnicity, disability, income, education, employment satisfaction—but sexual orientation and gender identity are missing. Some might think that such measures are too personal or sensitive to include. But numerous government surveys on the U.S. population already collect sexual orientation and gender identity data—some even since the 1980s. They’ve all found that such measures don’t cause any survey issues. Moreover, there’s no personal mandate to respond; options such as “I don’t know” or “I don’t wish to respond” are always available. And for those who do wish to respond, federal law protects the confidentiality of their data.

My colleagues and I, along with 17 scientific organizations, have encouraged NSF to include sexual orientation and gender identity measures in the national STEM census. NSF has expressed interest and is in a piloting stage, but there are no promises for inclusion. Yet real progress in eliminating the career and educational barriers LGBTQ people face requires it. Universities, funding agencies and scientific societies could follow suit, and scientists should consider encouraging their own institutions to track sexual orientation and gender identity data in demographic surveys.

As with advocacy for inclusion of these measures in the 2020 U.S. Census and other major federal data collections, policy experts agree that LGBTQ people must be counted to effectively inform policy and prevent employment, housing and other forms of LGBTQ discrimination. The same should go for institutions in STEM.

The benefits of all this would be immense. Finally, potential disparities of LGBTQ people across STEM fields, career stage, location, and other factors (including the unique intersections of gender, race, disability or economic background) could be carefully tracked, and researchers and policy makers could develop strategies and systems of accountability to address them.

Including sexual orientation and gender identity measures in surveys and official reports of the national STEM census, and in reports by universities, funding agencies and scientific societies, could also even have the potential to change hearts and minds, leading scientists to question their own biases or assumptions. Take same-sex marriage. Implicit levels of antigay bias in a given U.S. state dropped markedly after each state legalized same-sex marriage. Official signs of acceptance and inclusion matter.

There are certainly steps we can take now to make things better. Visibility initiatives like 500 Queer Scientists are needed and impactful, as is fostering supportive and inclusive environments. Role models are critical, and starting LGBTQ networking events at conferences is a great way to foster community. But these alone aren’t enough. Universities, funding agencies and scientific societies all need a mechanism to track the biases LGBTQ people may encounter if they are to ever prevent those biases—or stop them when already underway.

Nearly 40 years ago when the Science and Engineering Equal Opportunities Act was signed, Congress declared that it is in “the national interest to promote the full use of human resources in science and engineering.” Our world today faces perhaps even more complex and urgent scientific challenges than it did then, and if for that sake alone it’s imperative that all individuals wishing to contribute to science not be hindered in their opportunity.

LGBTQ people have already contributed enormously to science. Carolyn Bertozzi invented groundbreaking “bioorthogonal” chemical reactions leading to new treatments of disease. Lynn Conway created dynamic instruction scheduling that revolutionized CPU processing power. Nergis Mavalvala helped the world first observe gravitational waves. Alan Turing’s work inspired the field of artificial intelligence.

But without a means to track progress and hold STEM institutions accountable, LGBTQ people’s ability to contribute to science will continue to be vulnerable.