Violence by Deleted Data
A note on epistemic injustice under Trump's administration
Epistemology is the study of knowledge. Epistemic injustice is the act of suppressing groups of people by disrupting their capacity to know, speak, or be taken seriously. This includes, but is not limited to:
Restricting access to vital information
Failing to believe the accounts of those with direct experiences of an event when belief is warranted
Using complex terms specifically for the purpose of exclusion
Deeming speakers of one language to be more knowledgeable than speakers of another
Systematically withholding educational resources
Purposefully seeding misinformation into the public sphere
Denying individuals authority on the basis of traits such as gender or race
Appropriating material created by an oppressed group without proper citation, attribution, or respect for the original context
Providing a perspective of history that hides the violence perpetuated by dominant groups in public secondary school
Maintaining an economic system that forces many individuals to stay in survival mode to such an extent that they have no time to deeply criticize the economic system in question
All of these things happen. Fricker (2007) identified two types of epistemic injustice: testimonial and hermeneutic. Testimonial injustice occurs when unfair bias reduces the credibility a listener assigns to a speaker. When activists asserted that we should “believe women” during the #MeToo movement, they were trying to counteract testimonial injustice.
Hermeneutic injustice, on the other hand, is characterized by situations in which the standardized conceptual and linguistic resources available are inadequate for understanding or explaining specific experiences of oppression. For example: before the term “sexual harassment” was circulated, it was challenging to explain experiences of sexual harassment. Both types of injustice require reflection to comprehend.
The Trump administration has done something far less covert; at the beginning of his second term, they simply deleted data related to homophobia, transphobia, and gender-based violence. They retracted reports from the CDC website, among other government sites.
In addition, policy analysts Lauren Bouton and Elana Redfield found that a conservative estimate of “360 different federal data collections” had been removed or altered over the first year of Trump’s second presidency. They searched for these altered collections over 2025 and 2026, and expect to find more. Many of these shifts were done under the guise of “non-substantive” alterations, which require no review period. Replacing the word “gender” with “sex,” for example, was deemed a non-substantive alteration.
These acts should have been obvious. Perhaps they were. Nevertheless, I, perennially ignorant, only found out this month. In an attempt to write a research-based article on the intersection between neurodivergence and sexual violence, I noticed the data I wished to pull up wasn’t quite as forthcoming as I expected it to be. The act of searching felt vaguely akin to grasping for your glasses where they usually are on the nightstand and finding a fistful of air. A friend informed me of the CDC take-downs.
A case study in intentionally broken links
Consider the following 2013 article from the National LGBTQ Task Force: “Bisexual Women have Increased Risk of Intimate Partner Violence, New CDC Data Shows.” This piece cites a CDC report for the statistic that 61.1% of bisexual women have experienced rape, partner violence, or stalking. This report is not easily available. The link that is supposed to map to this report redirects to a page that states: “The page you were looking for has moved.” The reader is not informed of any place this page might have moved to. The other link—whose label indicates it regards the 2010 data taken by the National Intimate Partner and Sexual Violence Survey (NISVS)—leads to a page that says: “The page you’re looking for was not found.” In short, both links lead to a virtual ghost.
Most people, I gander, would give up on attempting to verify a source at this point. This archetypical reader, who is horribly busy, would never learn about the methodology that led to the statistic in question. They would instead leave, unsure if the article they just read was backed up by quality research. But I can be obsessive and stubborn. I tried to find this strangely evasive report—and, additionally, where the original data was housed.
A quick Google search reveals a 2011 summary of the 2010 data on the CDC website, but no actual report. I downloaded this summary. It does not mention any data on rates of sexual violence against the LGBTQ+ population.
When I did not search for “CDC” but focused specifically on the term “NISVS,” I was able to find a brief fact sheet summarizing the data from the 2010 survey, as it pertains to the LGBTQ+ population. It includes the 61.1% statistic. But this fact sheet was made available through the National Center on Domestic and Sexual Violence website. It is not available through the CDC. And by itself, it’s inadequate; it does not include analysis or data-collection methodology.
After Googling to the best of my ability, I attempted to check the NISVS index on the CDC website. It’s supposed to function like a table of contents. As of May 28th, 2026, it states, verbatim:
Per a court order, HHS is required to restore this website to its version as of 12:00 AM on January 29, 2025. Information on this page may be modified and/or removed in the future subject to the terms of the court’s order and implemented consistent with applicable law. Any information on this page promoting gender ideology is extremely inaccurate and disconnected from truth. The Trump Administration rejects gender ideology due to the harms and divisiveness it causes. This page does not reflect reality and therefore the Administration and this Department reject it.
For context, the CDC is a department within the HHS. After taking down CDC reports, the Trump Administration was sued by Doctors for America, among others. This resulted in a substantial (but incomplete) restoration of removed material. Still, the report I was looking for wasn’t available on the index.
As a last-ditch effort to find this elusive report, I asked Anthropic’s Claude for assistance. I gave it the information I had: the 2011 summary, the broken links, the original article, the context. Despite hallucinating a little bit, it allowed me to parse the data trail. The original data from 2010 is distributed by the Inter-university Consortium for Political and Social Research. Because it contains qualitative reports with information that might enable someone to identify individuals, it’s only available to those with special permissions. The report I had long sought after, which analyzes how the 2010 data pertains to the LGBTQ+ population in particular, was published in 2013.
After asking Claude multiple times, I found that this mythical 2013 report still exists, despite being nearly impossible to find through Google. It rents an apartment in the CDC stacks. The CDC stacks do not include “gender-based violence” as one of the main categories advertised on the home page, but by using keywords from the already-found report, it is discoverable through the CDC stacks search function.
Alas — before searching, I hadn’t been truly privy to the CDC stacks. It had taken multiple hours and an annoying headache to parse approximately what was going on with the help of artificial intelligence. Without it, I don’t think I would have been able to find the report at all. It all feels eerily similar to a classic passage from Douglas Adam’s Hitchhiker’s Guide to the Galaxy:
“But the plans were on display…”
“On display? I eventually had to go down to the cellar to find them.”
“That’s the display department.”
“With a flashlight.”
“Ah, well, the lights had probably gone.”
“So had the stairs.”
“But look, you found the notice, didn’t you?”
“Yes,” said Arthur, “yes I did. It was on display in the bottom of a locked filing cabinet stuck in a disused lavatory with a sign on the door saying ‘Beware of the Leopard.”
Existence and searchability are different things. Old links broke. In this case, it wasn’t an accident. The Trump administration could have re-instated the 2013 report with the same old link, but chose not to. Centralized, authoritative indexes got neutered. Since Google-ability depends on the connections between pages, deleting digital papers and putting them back up under a different label makes a difference. It makes old articles look unjustified to the casual eye.
Analysis: hermeneutic and testimonial injustice in Trump’s actions
Pushback against the Trump Administration has been partially successful. Lovely humans at educational institutions and volunteer networks worked to save data before it was deleted, which underscores the role these institutions play in helping us remember. The result, as far as I can tell, is a kind of scattering. Some bits of information settle into a different place in the brain of the country, like words that almost but do not quite fall off the tip of the tongue. I worry about holistic dementia.
We rely, often without noticing, on the way in which information transfers between people. We rely on other people to count things that need to be counted, to tell the right people when they’ve finished counting via a network, and to frame the result in a useful way. Trump’s administration has managed to manipulate the counting, the network, and the frame.
They’ve changed the way we count by changing the questions that will later be asked. For instance: the National Crime Victimization Survey (NCVS) will no longer inquire into the gender identity of victims and perpetrators. This will make it harder to know about violence committed against trans and non-binary people. They’ve altered the network by breaking links, as discussed. They’ve disrupted the framing by swapping key terms. If the word “gender” is replaced by “sex,” data that was originally intended to be about the identity of study participants may later appear to be about chromosomes, genitals, and secondary sex characteristics.
Whether the reader cares more about what can be observed from the neck down than personal identity is irrelevant. What matters is that this swap both substantially mangles the intended meaning of the original document, and, functionally, delegitimizes a conceptual recourse: the term ‘gender.’
It used to be that the term ‘sex’ fuzzily encompassed biological sex, social role, and gender identity. When groups of people began to repurpose old terms in a manner that allowed speakers to more clearly differentiate the concepts BIOLOGICAL SEX and GENDER IDENTITY (with the terms ‘sex’ and ‘gender identity,’ respectively), a new conceptual resource was made available; the shift in term-use made it easier for people to explain experiences they’d previously lacked easy language for.
Thus, by changing the word “gender” to “sex,” I believe the Trump administration is engaging in testimonial and hermeneutic injustice, simultaneously, in one suave swap:
Testimonial injustice: warped the testimonies of the subjects studied by changing the meaning of the text to something the original authors did not intend
Hermeneutic injustice: reduced the institutional backing of a conceptual resource needed to explain an important human experience
All of this comes under the cover of a “non-substantive” alteration. In comparison to more colorful forms of violence, I can see how it would seem non-substantive—just a little thing, snuck in.
References
Fricker, M. (2007). Epistemic injustice: Power and the ethics of knowing. Oxford University Press.
National LGBTQ Task Force. (2013). Bisexual women have increased risk of intimate partner violence, new CDC data shows. https://www.thetaskforce.org/news/bisexual-women-have-increased-risk-of-intimate-partner-violence-new-cdc-data-shows/
Centers for Disease Control and Prevention. (n.d.). The page you were looking for has moved [Redirected page]. Retrieved May 28, 2026, from https://www.cdc.gov/ViolencePrevention/pdf/NISVS_Report2010-a.pdf
Bouton, L., & Redfield, E. (2026). Removal of sexual orientation and gender identity data from federal data collections. Williams Institute, UCLA School of Law. https://williamsinstitute.law.ucla.edu/publications/sogi-data-collection-removal/
Black, M. C., Basile, K. C., Breiding, M. J., Smith, S. G., Walters, M. L., Merrick, M. T., Chen, J., & Stevens, M. R. (2011). The National Intimate Partner and Sexual Violence Survey: 2010 summary report. Centers for Disease Control and Prevention. https://stacks.cdc.gov/view/cdc/11637
Centers for Disease Control and Prevention. (2025). NISVS documentation [Page containing court-ordered restoration notice]. U.S. Department of Health and Human Services. Retrieved May 28, 2026, from https://www.cdc.gov/nisvs/documentation/index.html
Inter-university Consortium for Political and Social Research. (n.d.). National Intimate Partner and Sexual Violence Survey, 2010 [Restricted-access dataset]. National Archive of Criminal Justice Data. Retrieved May 28, 2026, from https://www.icpsr.umich.edu/web/NACJD/studies/34305
Walters, M. L., Chen, J., & Breiding, M. J. (2013). The National Intimate Partner and Sexual Violence Survey: 2010 findings on victimization by sexual orientation. National Center for Injury Prevention and Control, Centers for Disease Control and Prevention. https://stacks.cdc.gov/view/cdc/12362
Adams, D. (1979). The Hitchhiker's Guide to the Galaxy. Pan Books.
Nietzel, M. T. (2025, February 9). Harvard, others saving data as Trump's team scrubs federal webpages. Forbes. https://www.forbes.com/sites/michaeltnietzel/2025/02/09/harvard-others-saving-data-as-trumps-team-scrubs-federal-webpages/
Quinn, A. (2026, May 7). 'Things were going dark left and right': The race to save US government datasets before they're deleted. The Guardian. https://www.theguardian.com/us-news/2026/may/07/save-government-datasets-deleted
Meyer, I. H., Redfield, E., Bouton, L. J. A., & Flores, A. R. (2026). Erasure of anti-trans violence data in the United States. Williams Institute, UCLA School of Law. https://williamsinstitute.law.ucla.edu/publications/ncvs-sogi-data-removal/
Scott, J. W. (1986). Gender: A useful category of historical analysis. The American Historical Review, 91(5), 1053–1075. https://www.jstor.org/stable/1864376