The Hidden Costs of Anti-Bias Education

For decades, institutions have poured billions into programs designed to fight racism, antisemitism, Islamophobia, caste discrimination, and other forms of bias. The premise has been simple: if people better understand systemic injustice, they will become more empathetic and less prejudiced. But widely used materials are rarely subjected to scientific testing. The Network Contagion Research Institute (NCRI sought to understand the effects of the way we teach about identity-based harm.

Research by the Rutgers-University-based (NCRI, a leading think-tank on cyber-threats and information disorder, suggests that some of the most widely adopted anti-oppression approaches may unintentionally produce the very hostility they aim to eliminate.

Source: Created using Ai/Pamela Paresky

Interventions examined were based in “anti-oppressive” frameworks popularized by contemporary DEI pedagogy. In such programs, moral life is defined as a struggle between oppressor groups and oppressed groups, emphasizing structural harm and collective responsibility.

The study focused on diversity training interventions that emphasize awareness of and opposition to “systemic oppression,” a trend fueled by the 2020 Black Lives Matter movement and popularized by texts such as Ibram X. Kendi’s How to Be an Antiracist. While not representative of all DEI pedagogy, “anti-racism” and “anti-oppression” pedagogy and intervention materials have seen widespread adoption across sectors like higher education and healthcare. Although designed to increase awareness, the studies reveal a more troubling psychological effect.

Across thousands of participants, multiple identities, and varied contexts, the research uncovers a consistent psychological pattern: Identity-based, accusatory messaging that negatively focuses on inherited characteristics like race and gender (for example, “We’re putting the bigots and antisemites on notice: The law is coming after you,” and “White identity is inherently racist”) can trigger defensiveness, suspicion, and even overt prejudice. In other words, in studies undertaken to understand the effects of the way we teach about identity-based harm, the NCRI concluded that the way we teach about identity-based harm can sometimes cause harm.

Hostile Attribution Bias

In one intervention, 423 research participants recruited from the Rutgers campus read DEI-style excerpts on racism (from Ibram X. Kendi and Robin DiAngelo), Islamophobia (from materials produced by the Institute for Social Policy and Understanding), or caste oppression (from Equality Labs) and were then given neutral scenarios containing no evidence of bias.

The results were striking:

• Participants exposed to anti-racist texts perceived 21% more racial discrimination in a race-unspecified admissions scenario, despite the absence of any racial cues.

• Those who read anti-Islamophobia materials judged an identical criminal trial significantly less fair when the defendant had a Muslim-coded name.

• Participants who read caste-oppression narratives perceived 32.5% more microaggressions, 15.6% more harm, and 11% more bias in a caste-neutral admissions scenario.

This is classic hostile attribution bias—a pattern in which people begin to see prejudice even in neutral scenarios.

Increased Punitiveness

The same participants were also significantly more willing to punish the fictional individuals involved even though the scenarios gave no evidence of wrongdoing:

• Support for suspending or retraining the race-scenario admissions officer rose by 12-16%.

• Punitive attitudes toward the fictional caste-scenario interviewer increased by 19%.

• Exposure to caste-related DEI material intended to reduce bias against the lower castes increased (by 27-35%) participants’ endorsement of statements adapted from demonizing historical propaganda portraying Brahmins (the highest caste) as “parasites,” “viruses,” or “the devil personified.”

What Happens When Anti-Bias Education Targets Antisemitism?

If identity-based messaging about racism, Islamophobia, and caste can trigger hostile attribution bias, the antisemitism experiment with 3,355 participants shows a similar pattern. The study tested how different types of antisemitism-related messages—including two anti-oppressive essays derived from the Anti-Defamation League (ADL)— a universal-values alternative, and neutral controls affected participants’ defensiveness, perceptions of Jews, and the likelihood of expressing antisemitic or identity-based responses.

Bias Essential Reads

Is It Actually Possible to Overcome Our Prejudices?

When the Brain Shapes Belief

Participants who were randomly assigned anti-oppressive antisemitism-education materials currently available for use on the ADL website reported more than a 170 percent increase in defensiveness, describing the messages as making them feel attacked, accused, or manipulated. They also showed large increases in irritation and perceived hostility, interpreting the messages themselves as adversarial. In their open-ended written responses, these same conditions produced more than a 15-fold increase in antisemitic comments and roughly double the amount of toxic language, defined as statements rude or discouraging enough to make others want to leave a discussion.

The aim of these materials was to reduce hostility toward Jews. The study shows that, for some participants, the effect was the opposite. (When presented with the results of the NCRI study, the ADL responded that they no longer use those materials. However, as of this writing, they are still live on their website.)

Why Identity-Based Framing Backfires

These studies converge on a psychological explanation: Identity-group framing shifts moral focus from shared values to group suspicion. When trainings define morality in terms of group oppression:

1. People fear being cast as members of a harmful group.

2. People become hypervigilant for prejudice, both in others and in themselves.

3. Moral life becomes adversarial.

4. Defensiveness fuels counterprejudice.

5. Authoritarian impulses rise.

Perhaps most concerning, efforts to reduce bias can backfire. The backfire effect occurs "when a correction leads to an individual increasing their belief in the very misconception the correction is aiming to rectify.” In this case, when moral responsibility is framed as a group contest, people become more suspicious of others and more willing to police them.

The Alternative: Shared Humanity Reduces Hostility

Source: Created using Ai/Pamela Paresky

If identity-based approaches trigger defensiveness, what works better?

The NCRI also tested a message emphasizing shared moral values—integrity, fairness, conscience, responsibility—rooted in both civic ideals and Abrahamic teachings.

This values-based message produced:

• No defensiveness

• No antisemitism

• Lowest perceptions of exaggeration or bias

• Highest ratings of meaning and importance

• Increases in universalistic, human-connecting language

• A 12% reduction in the misperceptions that Jews as a group are racist.

What This Means for Anti-Bias Efforts

When people are addressed as individuals capable of conscience and agency, they are more likely to respond with openness than with the defensiveness that is activated when they are addressed as members of identity groups.

This offers a roadmap for improving educational and institutional messaging:

1. Avoid framing morality as group competition.

2. Don’t assume that merely exposing hidden prejudice reduces it.

3. Use values, not identities, as the foundation of moral education.

4. Treat prejudice reduction like a medical intervention with possible side effects.

The combined research reveals a consistent psychological truth: People respond better to being invited into moral responsibility than being accused or shamed, or being persuaded that they are oppressed. Anti-oppressive frameworks often assume that confronting people with structural harm will open their hearts. But for many, it either closes them or encourages them to look for someone to blame. Defensiveness, suspicion, and backlash are predictable responses to feeling blamed or threatened.

The alternative is shared moral purpose. Using shared values messaging increases openness, meaning, and constructive engagement. When we appeal to what is highest in people—not what is worst or most vulnerable about their groups—we create the conditions for compassion, agency, and genuine reduction in prejudice.