In Executive Order 14168, President Trump announced a ban on the use of federal funds to promote “gender ideology,” an ill-defined term which roughly corresponds to acknowledging the existence of transgender people. As part of that prohibition, agencies were expected to assess grant conditions and ensure that grant funding does not promote gender ideology. This executive order has been the subject of significant litigation, including an omnibus lawsuit by the American Public Health Association and the United Auto Workers challenging cancelled National Institutes of Health grants. Many of these challenges focus on the termination of existing grants or changes to grant evaluation, correctly pointing out that these actions will upend careers, violate long established practices, and harm the public.

But terminated grants are just the tip of the iceberg. Conditions on federal funding like the executive orders banning the promotion of “DEI” or gender ideology, when applied to ongoing research, are especially insidious. In effect, these restrictions constrain not just the topics of research but the acceptable outcomes of the scientific process. Such constraints will also skew the scientific literature, and in doing so distort our collective knowledge about the world.

This incompatibility with the scientific process makes these types of viewpoint-based, ideological conditions impossible to square with promoting the program of science (the purpose of the National Science Foundation) or seeking fundamental knowledge about the nature and behavior of living systems (the purpose of the National Institutes of Health). These restrictions are thus unconstitutional when applied to scientific grant funding, even when the government may in other circumstances fund programs based on its own preferences.

The Gender Ideology EO and Scientific Studies

Under the current executive order banning the promotion of gender ideology, scientists may find it difficult to report their findings accurately, and even harder to make appropriate recommendations based on their findings. The first public example of how the EO might do this comes from Schiff v. U.S. Office of Personnel Management, a case involving the federal government website Patient Safety Net (PSN), which hosted medical case reports. PSN’s editors interpreted the EO to mean that hosting articles mentioning transgender people counted as promotion of gender ideology. Likewise, using the words “male,” “man,” or “boy” to refer to people who are not biologically “male” (which the Trump administration defined as “the sex that produces the small reproductive cell”) also was impermissible promotion of gender ideology. Several articles on PSN did not comply with these rules, and the editors offered their authors an ultimatum—remove the offending language or have the articles taken down. At least two authors chose the latter and then filed suit. (A federal judge recently ordered the restoration of all articles that were taken down in response to the EO.)

As to how this skews the scientific literature and constrains the publication of accurate results, let’s take an example. It is a fact about the world that transmasculine people who were assigned female at birth (AFAB) and undergo masculinizing hormone therapy typically have higher hemoglobin and hematocrit levels than other AFAB people because testosterone administration increases hemoglobin and hematocrit.

Imagine an NIH-funded project—one that is subject to a restriction that federal funds cannot be used to promote gender ideology—is updating reference ranges for hemoglobin and hematocrit for determining blood cancer risk. (This example is taken from Ada S Cheung et al., Approach to Interpreting Common Laboratory Pathology Tests in Transgender Individuals, 106 J. Clin. Endocrinol. Metab. 893 (2020).) The study does not explicitly focus on transgender people, so even after keyword-based purges, it is still funded. After recruiting a study population, the authors find that there are transgender people who participated. They analyze their data and prepare to publish their results.

Our study authors now have a question of how to organize their data. Including AFAB study participants on testosterone in the male category is prohibited, because that’s gender ideology. However, including AFAB participants who have undergone hormone therapy in the category of ‘women’ or ‘female’ would increase the average hemoglobin/hematocrit levels in the AFAB group. This, in turn, could impact the recommended reference ranges for screening and could lead to worse assessment of cancer risk. The binary labels of sex that are required under the executive order run headlong into the messy reality of sex characteristics in practice, and our authors are now caught in a double bind. Publishing their results puts their grant funding at risk, but not publishing makes it difficult to show the kind of track record that would get funding in the future. (Of course, if they had found the opposite result—that AFAB participants on testosterone were best categorized with AFAB people who have not taken testosterone, they might only encounter issues of wording.)

The ban on gender ideology prohibits the publication of results that support sex/gender distinctions, or even those that just accurately report the reality that transgender, intersex, and gender-diverse people exist. Functionally, as in the example, it serves as a form of prior restraint based on ideological compliance at the cost of scientific accuracy. Many scientists have already written about the harms that publication bias towards positive results have caused to various fields. Such literature skews also have significant negative effects on the accuracy of meta-analyses, which count studies to determine what the agreed-upon consensus is in certain areas. The ban on gender ideology thus creates systemic risks to the accuracy of federally funded scientific research, and thus, the literature as a whole.

Ideological Compliance for Science Cannot Be Constitutional

The Supreme Court has recognized that the government can spend money promoting particular beliefs without running afoul of the First Amendment. In Rust v. Sullivan, the Supreme Court considered whether restrictions on the use of Title X funding that prohibited counseling, referral, and the provision of information regarding abortion as a method of family planning were unconstitutional, finding that they were not. However, the decision in Rust notes that the restrictions contemplated did not prevent a Title X project from referring a pregnant person to abortion-related services if necessary for medical reasons. In finding the restrictions on the use of funds to be narrowly tailored, the Court looked to Congress’s intent to prevent the use of abortion for family planning and separated it from being not able to refer patients to abortion care if medically indicated.

Rust, and similar analysis in subsequent cases including Legal Services Corp. v. Velazquez and United States Agency for International Development v. Alliance for Open Society International, Inc., evaluates the relationship between funding restrictions and the programmatic aims of the funded programs, finding that they must be connected. In the gender ideology EO context, a recent opinion from the Northern District of California found that a group of healthcare providers was likely to succeed on the merits of a First Amendment challenge, as the EO’s requirement banning the promotion of gender ideology was not directed towards advancing any legitimate objectives of the provision of healthcare.

Although there are many similarities between the application of the gender ideology EO to scientific grant funding and the context in the Northern California case, advocates for science should make a stronger categorical argument than the litigants in that case. Ideological restrictions, like those imposed by the gender ideology EO or by banning the promotion of DEI, are not only programmatically unrelated to the funding of scientific research, they are fundamentally incompatible with the scientific process. The purpose of scientific research is to find out facts about the world, and scientific progress depends upon the ability to publish one’s results and to trust the accuracy of prior work. Such a process may return results that do not match the ideological priors of its funders, or their preferred beliefs about the world. But, of course, as we are often reminded, facts don’t care about one’s feelings.