Health-focused frames mobilize Americans to action to address LGBTQ inequality

Significance

This work investigates how exposure to different domains of LGBTQ (Lesbian, Gay, Bisexual, Transgender, and Queer) inequality (i.e., health, economic, belonging) influences Americans’ support for action. The findings reveal that emphasizing health-related disparities motivates more support for action meant to mitigate the inequality (e.g., policy support, attending a protest) than economic or belonging-based disparities. We also examine two psychological drivers of this effect: perceptions of blame and moral sacredness violations. Finally, we test a method that increases support for action, namely by mentally linking economic disparities to health disparities. In sum, our work identifies the mechanisms underlying support for advocacy and policy efforts that have implications for health equity for LGBTQ populations.

Abstract

“More than 1 in 8 LGBTQ people live in states where doctors can refuse to treat them.” This headline describes a flurry of laws passed in 2025 allowing doctors to refuse treatment of LGBTQ patients based on personal beliefs. This and other laws like it limit healthcare access and exacerbate health inequality for LGBTQ individuals. But health is not the only domain in which LGBTQ individuals face disparities. Long-standing inequalities also exist in other domains of life, for example, gay men earn up to 32% less wages than their heterosexual peers. In the current research, we ask if exposure to LGBTQ inequality in one domain (e.g., health) motivates more support for action than exposure to inequality in other domains (e.g., economics). In six experiments (five preregistered), we test our hypothesis that emphasizing health-based LGBTQ inequality (vs. economics and belonging) spurs engagement with and support for mitigating action (S1-2). We also investigate why people support action (S1a-1b) and what boosts support (S3a-3c). As predicted and preregistered, we find greater support for action to mitigate health-based LGBTQ inequality, compared to economics and belonging. This occurs, in part, because people blame LGBTQ individuals less for health (vs. economic) disparities and perceive health inequalities to violate morally sacred values, which is associated with greater perceived injustice. Finally, we find that connecting economic inequality to health inequality in individuals’ minds can increase support for mitigating economic inequality.

In 2025 alone, 616 bills have been introduced in the United States that aim, among other things, to limit students’ and teachers’ speech and activities, such as prohibiting teachers to educate on gender identity and sexual orientation or barring students from participating in physical education along gender identities [i.e., so-called “sports bans”; (1, 2)]. Many states now allow health care providers to decline care to people based on their religious beliefs (3); for example, medical providers in Ohio can legally deny care to LGBTQ individuals based on religious or moral objections. These proposed and recent changes contribute to long-standing inequalities in LGBTQ and non-LGBTQ individuals’ lived experiences, including in healthcare access and other domains. These kinds of laws can have serious implications; for example, the enactment of laws that restrict the rights of transgender and nonbinary people have been linked with significant increases in suicide attempts by young people in these communities (4).

LGBTQ Americans face persistent disparities in many domains of life, ranging from healthcare access to social acceptance, and economic opportunities (5–8). For example, LGBTQ communities face higher rates of poverty (9), less access to medical care (10–14), and higher rates of certain mental disorders [such as depression or anxiety; (15)]. Taken together, the lives of LGBTQ individuals are, on average, more challenging, harsher, and more unpredictable than the lives of their straight and cisgender peers.

It is estimated that 10 percent of the US population identifies as LGBTQ (16) and at least 40 percent of Americans report having a family member or close friend who identifies as LGBTQ (17). Thus, approximately half of the American population knowingly experiences the often negative consequences of how LGBTQ people are treated in society—either directly, because they are personally affected, or indirectly, because their friend or family member is affected. While personal connection to an issue can be an important precursor to support for taking action (18–21), intergroup contact can also backfire (22, 23). Given this, what other factors influence support for action to mitigate these inequalities?

Motivation to reduce inequality is multiply-determined. Social norms or social acceptance of inequality-mitigating behavior (24, 25) are important factors that determine support for action. In addition, group membership (i.e., belonging to a group that is advantaged or disadvantaged by inequality) can influence action support. People who belong to the advantaged group are, on average, less likely to support inequality-mitigating action than members of disadvantaged groups (e.g., refs. 26 and 27).

How people construe an issue also predicts support for inequality-mitigating action. For example, perceptions of an issue as unfair or unjust predict motivation to mitigate an issue (19, 27, 28). Importantly, because these construals reflect subjective interpretations of the current environment (see ref. 29), this potent psychological mechanism can be shifted via fairly minimal interventions, such as the framing of an issue. Prior work on inequality framing has examined comparative framing (focusing on the implications of inequality for disadvantaged vs. advantaged groups) as well as emphasis framing (varying what information about inequality is emphasized; see ref. 30). This body of work reveals that comparative framing can impact support for action; for example, focusing on disadvantages (vs. advantages) spurs support for action to mitigate economic inequality. However, the impact of comparative framing on different forms of inequality (i.e., gender, race) and on other interventions is more mixed.

In terms of emphasis framing, the domain of inequality that is highlighted influences injustice perceptions. Emphasizing health-based (vs. economic or belonging) inequality leads to greater injustice perceptions and greater support for action to reduce racial disparities. In part, this is the case because inequalities in the domain of health are perceived to violate sacred moral values (i.e., values that are imperative to protect) and these moral sacredness perceptions are a precursor for perceptions of unfairness (27). Importantly, this prior research only examined implications of this type of emphasis framing for racial inequality and it remains unclear if the effect applies to other forms of social inequality.

In the present studies, we focus on examining the effects of emphasis framing—focusing on health vs. economics vs. belonging—on support for action to mitigate inequalities affecting the LGBTQ community. We do this for several reasons. First, given that prejudice against LGBTQ individuals is often based in moral judgments (31, 32) and moral judgments were identified as a mechanism linked with support for mitigating race-based inequality, it is plausible that LGBTQ inequalities operate via different psychological mechanisms and thus, health-based LGBTQ disparities may not elicit support for mitigating action. Second, LGBTQ membership is often perceived as controllable (33, 34) and perceptions of controllability are related to stigmatization of health issues (35, 36). The stigma around negative health outcomes for LGBTQ individuals might lead to perceptions that these outcomes are deserved and just, which could suppress support for action.

Third, prejudice against members of the LGBTQ community (e.g., gay men) involves a unique emotional component that is different from prejudice against Black Americans. Prejudice against gay men (and perhaps, other groups in the LGBTQ umbrella) has been posited to derive from perceived health threats [i.e., a fear of the spread of contagious disease; (37, 38)]. In sum, while it is possible that Americans would react similarly to information about race-based and LGBTQ-based inequality (i.e., health disparities may trigger a moral sacredness violation, heightened unfairness perceptions, and facilitate support for action), there is reason to believe that perceptions of LGBTQ-based inequality might differ from perceptions of race-based inequality. Thus, it is important to empirically test, as we do in the current research, whether emphasizing inequality in the domain of health motivates greater support for action relative to other domains of inequalities faced by the LGBTQ community.

The present research also tests an additional psychological mechanism that may undergird the potency of different disparity framings. We aim to test if examples of economic disparities trigger thoughts on meritocracy and whether individuals are to blame for their plight. Blame has been shown to be a powerful predictor of how marginalized groups are viewed and treated. For example, in the case of social class inequality, decades of research on victim blaming (39, 40) and the just-world fallacy (41, 42) give us insights into the antecedents and consequences of people’s tendency to blame poverty on the poor [i.e., putting the responsibility on behaviors of the poor rather than on structural forces; (43, 44)]. Blaming an individual for negative outcomes tends to be associated with perceptions that the outcomes are deserved or just (also see ref. 45). Given the propensity to attribute blame toward individuals for economic outcomes (i.e., toward people who are poor), we hypothesize that people might be more likely to blame individuals for unequal outcomes in the domain of economics (vs. health) which could lead to perceptions of economic disparities as more just. Put differently, inequalities in the domain of health could elicit greater injustice perceptions because people might be less likely to blame individuals for typical health (vs. economic) inequalities. In turn, decreased blame could lead to heightened perceptions of unfairness and subsequent support for mitigating action. Thus, it is crucial to examine these underlying processes of why Americans perceive and support mitigation of LGBTQ-based inequalities in different domains.

Taken together, in six experiments (five preregistered, N_{behavioral experiments} = 2,644, N_{social media experiment} = 33,797), we investigate whether emphasizing LGBTQ inequality in the domain of health vs. other inequality domains (i.e., economics and belonging) spurs engagement with the topic and support for mitigating action (Studies 1a, 1b, 2), whether perceptions of blame and moral sacredness violation explain why people support mitigating action via greater perceptions of injustice in some domains more than others (Studies 1a, 1b), and how to boost mitigation support for domains with weaker support (Studies 3a–3c). To do this, we utilize a variety of methodologies (such as real-world behavior on social media, S2) and we test our hypotheses via multiple operationalizations of mitigating action: policy support (S1a, S3a–c), collective action support (S1b, S3a–c), and social media engagement (S2).

Results

Studies 1a-1b.

In two separate online samples of US citizens (S1a preregistered: N = 767, 87% identified as straight, 47% identified as a woman, 51% identified as a man, 1% identified as nonbinary, and <1% did not specify their gender^*; S1b preregistered: N = 802, 86% identified as straight, 51% identified as a woman, 48% identified as a man, 1% identified as nonbinary, and <1% did not specify their gender), participants viewed a short infographic highlighting either health-based vs. economic-based LGBTQ disparities (S1a also included a third, belonging-based disparities condition; see Fig. 1). Then, participants reported their support of mitigating action. In S1a, participants reported their support for general policies (two items; e.g., “I support policies that aim to reduce these disparities”; r = 0.88) and concrete policies (two items; e.g., “Increase sales taxes by 0.5% for all Americans to reduce these disparities”; r = 0.64), while participants in S1b reported their support for general collective action (one item; “I really want to take action to reduce these disparities”) and concrete collective action (three items; e.g., “How likely are you to engage in the following actions to reduce these disparities?: Attend a protest that aims to bring awareness of these disparities”; α = 0.87).

Fig. 1.

Infographic about LGBTQ inequality with either a focus on health, economics, or belonging presented to participants in S1a (the health and economic infographics were used in S1b).

As preregistered, in S1a, a main effect of disparity condition on general policy support emerged, F(2, 764) = 11.55, P < 0.001, η_p² = 0.029 (Fig. 2). Specifically, participants exposed to health disparities supported disparity-reducing policies more than those exposed to economic disparities (β = 0.39, 95% CI = [0.22, 0.56], t(764) = 4.43, P < 0.001) and belonging disparities (β = 0.34, 95% CI = [0.17, 0.51], t(764) = 3.85, P < 0.001), with no differences between the economic and belonging conditions (β = 0.05, P = 0.580). Unexpectedly, there was only a marginal main effect of the disparity conditions on concrete policies, F(2, 764) = 2.36, P = 0.095, η_p² = 0.006. Planned contrasts reveal that while participants exposed to health disparities supported concrete policies more compared with those exposed to belonging disparities (β = 0.17, 95% CI = [0.02, 0.33], t(764) = 2.16, P = 0.031), the difference between health vs. economic disparities for concrete policies to address these disparities was not significant (albeit in the same direction, β = 0.10, 95% CI = [−0.05, 0.26], t(764) = 1.29, P = 0.196).

Fig. 2.

Raincloud plot with jittered data for participants’ general policy support for different disparity domains (economic, health, and belonging) in Study 1a. Error bars represent 95% CIs around the means (large black dots). Smaller dots indicate individual participant scores. The “cloud” areas reflect the data distributions.

In S1b, as preregistered, participants exposed to LGBTQ health disparities supported disparity-reducing action more than those exposed to LGBTQ economic disparities for both general collective action [t(799) = 3.30, P < 0.001, d = 0.24, 95% CI = [0.10, 0.37]] and concrete collective action [t(800) = 2.10, P = 0.034, d = 0.15, 95% CI = [0.01, 0.29]; see SI for descriptive statistics for S1a-1b].

Integrative data analysis (IDA).

We next tested the robustness and homogeneity of these effects in a preregistered IDA, combining Study 1a and 1b (and two pilot tests see SI Appendix).^† As preregistered, we find robust and homogenous effects across studies, such that participants show heightened support for all action (i.e., concrete action and general action, as well as policy action and collective action) to mitigate health (vs. economic) LGBTQ disparities [β = 0.24, 95% CI = [0.15, 0.34], t(1745) = 5.14, P < 0.001].

Mediation analysis.

We also assessed two different possible mediators that could impact perceptions of injustice and subsequently support for action across studies: Participants reported if health (vs. economic) LGBTQ disparities were more unjust (two items, rs > 0.74), a greater moral sacredness violation (three items, αs > 0.90), and that LGBTQ individuals were less to blame for these disparities (four items, αs > 0.92). Specifically, participants reading about health disparities reported that the disparities were more unjust [β = 0.23, 95% CI = [0.14, 0.33], t(1745) = 4.94, P < 0.001], a greater moral sacredness violation [β = 0.16, 95% CI = [0.07, 0.25], t(1744) = 3.39, P < 0.001), and that LGBTQ individuals were less to blame [β = −0.19, 95% CI = [−0.29, −0.10], t(1745) = −4.09, P < 0.001], compared to participants reading about economic disparities (adjusting for study).

We predicted that moral violation and blame would affect a sense of injustice which would contribute to explaining why people support action to address health versus economic disparities. To test this model, we conducted a serial mediation analysis (with 1,000 bootstrap samples, utilizing the lavaan package in R, adjusting for study). We find that highlighting health (vs. economic) disparities elicited support for action due to perceptions that these disparities are a greater moral sacred violation and LGBTQ individuals are less to blame for these disparities, which were both associated with greater perceived injustice (Fig. 3).

Fig. 3.

Serial process model of the indirect effect of disparity condition (health vs. economic) on action support through moral sacredness violation and blame affecting perceived injustice (with 1,000 bootstrap samples and unstandardized coefficients). *P < 0.05; **P < 0.01; ***P < 0.001.

We also conducted exploratory moderation analyses to assess whether participant sexuality (identifying as straight vs. nonstraight), gender identity (identifying as a woman, man, or nonbinary), and political orientation (1 = very liberal, 7 = very conservative) influenced the effects of inequality domain on action support. We did not find evidence of moderation for the outcome measure (all Ps > 0.368). However, we do find that political orientation moderates inequality domain effects on a subset of mediators, namely moral sacredness violation and injustice perceptions. Specifically, participants who identify as more conservative (vs. as more liberal) show a larger difference in the extent to which health (vs. economic) disparities are seen as violating sacred moral values and unjust (see SI Appendix, Fig. S3 for further discussion). Finally, we tested a possible alternative hypothesis, that people support health disparities over other disparities due to greater perceived frequency of health disparities; however, we find no evidence that the results are driven by perceived frequency of the disparities (see SI Appendix for details).

Study 2.

In S1, participants were given examples of disparities in different domains to strengthen the manipulation. S2 sought to provide a more controlled test and only included definitions of disparities alone (without specific content) in the experimental stimuli. In an ecologically valid context and utilizing stimuli that only differed by three words, S2 sought to assess if health disparities spark more engagement on social media compared to economic disparities. We created two versions of a sponsored advertisement on Facebook (see refs. 27, 28, and 46 for similar methodology): Each ad presented a headline about LGBTQ inequality, one focusing on health disparities and the other focusing on economic disparities (SI Appendix).

In a split test (randomized experiment) over 24 h, the health ad “won.” As preregistered, the health ad was more engaged with, reached more users, and was more cost-effective compared to the economic ad. The health ad reached 441 (2.6%) more people than the economic ad and had a 95% chance of outperforming the economic ad in a replication. This finding reveals that, as with racial health disparities (27), LGBTQ health disparities reflecting only the definition of disparities—vs. examples used in S1—elicit greater engagement on social media than LGBTQ economic disparities.

Studies 3a–3c.

Across S1–2, with varying methodologies, people reported greater support to address and greater online engagement with health-based LGBTQ disparities compared to economic-based LGBTQ disparities (and belonging-based LGBTQ disparities in S1a). Thus far, we have purposefully exposed participants to one inequality domain at a time. While keeping inequality domains strictly separated from each other bolsters the internal validity of these experiments, in real life, disparities in one domain often cause or coincide with disparities in other domains. For example, higher rates of poverty among LGBTQ individuals often lead to less health insurance coverage and subsequent poorer health outcomes (e.g., ref. 47). Thus, connecting LGBTQ nonhealth disparities (e.g., economic disparities) with health disparities, especially emphasizing how economic disparities may cause health disparities, might lead people to perceive that it is important to address health and economic disparities together and thus boost mitigating-support for economic disparities. In the final set of experiments (S3a, S3b preregistered, and S3c preregistered) we sought to tie together health and economic disparities to assess if recognizing the connected nature of both economic and health disparities might increase support to address economic disparities.

In three separate online samples of US citizens (S3a: N = 195, 82% identified as straight, 38% identified as a woman, 59% identified as a man, 3% identified as nonbinary; S3b: N = 386, 84% identified as straight, 47% identified as a woman, 52% identified as a man, 1% identified as nonbinary; S3c: N = 494, 83% identified as straight, 42% identified as a woman, 56% identified as a man, 2% identified as nonbinary, <1% NA sexuality), participants reported their general support for reducing LGBTQ-based disparities in both health and economics (two-items for health; rs > 0.83; and two items for economics; rs > 0.84). Following this initial measure of support, participants were exposed to an infographic emphasizing either that economic disparities cause health disparities (S3a: n = 99; S3b: n = 191; S3c: n = 243) or that there are economic disparities in addition to health disparities (S3a: n = 96; S3b: n = 195; S3c: n = 251; in both conditions, we provided the same examples of economic- and health-based LGBTQ disparities). Afterward, participants completed the same two general action support items (as before the manipulation) for both health (rs > 0.84) and economics (rs > 0.84).

We preregistered the same two hypotheses for S3b and S3c. We predicted that support for action to reduce economic-based LGBTQ disparities will increase after considering how economic disparities cause health-based disparities (H1). We also predicted that the condition highlighting causation between the two domains (vs. the condition highlighting both domains in tandem) would be more effective at increasing support for action (H2). We predicted H2 because we assumed that the “cause” condition (compared to the “additive” condition) would more readily link economic inequality to health inequality in participants’ minds, thus highlighting the interrelated nature of inequality and the importance to address both—health and economic inequality—simultaneously.

To test the consistency and homogeneity of the effects across the pilot (S3a) and preregistered studies (S3b-c), we conducted a preregistered IDA, combining Studies 3a-3c,^‡ (see Table 1 for all results).^§ Aligned with Studies 1-2, before exposure to the examples of health and economic disparities, we found that participants reported greater support to address health than economic disparities, t(1074) = 8.53, P < 0.001, d = 0.26. Crucially, we find an interaction between time and disparity domain, F(1, 1071) = 11.82, P < 0.001, η_p² = 0.011. Specifically, people report that they want to address health disparities more than economic disparities both before and after exposure to the causal and additive infographics, but this effect is significantly smaller after exposure to these manipulations [comparing pre–post difference scores for economic and health: t(1074) = 4.00, P < 0.001, d = 0.12].

Table 1.

Means, SDs, difference scores, and F-ratios for Study 3 (combined N = 1,075) adjusting for study

	Cause manipulation			Additive manipulation
	Health	Econ		Health	Econ
	M (SD)	M (SD)	Difference (health-econ)	M (SD)	M (SD)	Difference (health-econ)
Support Premanipulation	5.00a (1.77)	4.81b (1.82)	0.19*	5.13a (1.70)	4.93b (1.77)	0.20**
Support Postmanipulation	5.07c (1.77)	4.96d (1.88)	0.11*	5.20c (1.75)	5.08d (1.79)	0.12**
Difference (post-pre)	0.07*	0.15*		0.07**	0.15**
F-ratios
Main effect of condition (cause vs. and)				F(1, 1071) = 1.40, P = 0.237, ηp2 = 0.001
Main effect of domain (health vs. econ)				F(1, 1071) = 68.44, P < 0.001, ηp2 = 0.060
Main effect of time (pre vs. post)				F(1, 1071) = 38.28, P < 0.001, ηp2 = 0.035
Main effect of study				F(2, 1071) = 0.54, P = 0.583, ηp2 = 0.001
Interaction between condition and domain				F(1, 1071) = 0.02, P = 0.892, ηp2 < 0.001
Interaction between condition and time				F(1, 1071) < 0.00, P = 0.982, ηp2 < 0.001
Interaction between domain and time				F(1, 1071) = 11.82, P < 0.001, ηp2 = 0.011
Interaction between study and domain				F(1, 1071) = 1.02, P = 0.360, ηp2 = 0.002
Interaction between study and time				F(2, 1071) = 0.94, P = 0.389, ηp2 = 0.002
Interaction between study, time, and domain				F(2, 1071) = 0.56, P = 0.574, ηp2 = 0.001
Interaction between condition, time, and domain				F(1, 1071) = 0.06, P = 0.806, ηp2 < 0.001

Notes. Means with different subscripts indicate significant differences at P < 0.05. We computed difference scores by subtracting the mean for health minus the mean for economics for pre- and postratings and difference scores by subtracting the preratings and postratings for health and economics. Difference scores with * reflect significant (P < 0.05) and ** reflect significant (P < 0.01) contrasts comparing difference scores.

Another way to decompose the interaction is to assess how much people’s support changed for health and economic disparities after exposure to the infographics; indeed, after reading the infographics (compared to before exposure) participants significantly increased their support for both economic disparities, t(1074) = 7.54, P < 0.001, d = 0.23, and health disparities, t(1074) = 3.18, P < 0.001, d = 0.10, with a significantly larger increase in support for economic disparities, t(1074) = 4.00, P < 0.001, d = 0.12.

These results from Studies 3a–3c—for each individual study and for the IDA—are aligned with one preregistered hypothesis (H1) and not in line with another preregistered hypothesis (H2). Specifically, we found that support for action to reduce economic-based LGBTQ disparities increased after considering how economic disparities cause health-based disparities, but we also found support for action to reduce economic-based LGBTQ disparities increased after considering how economic disparities exist in addition to health-based disparities.

Fortunately, we also asked participants an exploratory open-ended question at the end of the survey (in S3c) in order to probe the degree to which participants connected economic to health inequality. Specifically, participants were asked “How do you think economic and health disparities are related? Please answer this question with 1-2 sentences.” We coded each text response using ChatGPT-4o (paid version) to assess if participants connected health inequality with economic inequality (on a scale of 0 = Did not at all connect economic and health domains, 1 = Did somewhat connect economic and health domains, 2 = Did connect economic and health domains, 3 = Very much connected economic and health domains, see SI Appendix for the exact prompt). ChatGPT-4o ratings reveal that participants connected the two disparities in both conditions to equivalent extents (cause: M = 2.00, SD = 0.71; additive: M = 1.91, SD = 0.74; P = 0.176). We verified the ChatGPT-4o coding with two human independent raters, and the results remain substantively unchanged if instead using the human raters’ coding (see SI Appendix for human raters’ results). Indeed, the ChatGPT-4o coding correlated r = 0.77 (P < 0.001) with the human raters’ coding (see SI Appendix for the correlation table).

Thus, regardless of whether health and economic disparities were explicitly causally connected or simply co-occurred in the experimental materials, participants connected LGBTQ economic disparities and health disparities to similar degrees. Taken together, these results reveal that one way to increase support for economic disparities is to connect them to health disparities, either by mentioning how economic disparities cause health disparities or mentioning examples of the two domains in tandem.

Discussion

Six studies using diverse methodologies and open science practices such as preregistration and replication, document greater engagement with and greater support for policy action and collective action to mitigate LGBTQ inequality in the domain of health, compared to the domains of economics (S1–S3) and belonging (S1a). This reveals generalizability of previous work on racial inequality, as participants perceived LGBTQ-based health (vs. economic) inequality to violate morally sacred values, which was associated with greater perceived injustice, and in turn, more support for action. Unique to the present work, we also uncovered perceptions of blame as a potential mechanism undergirding support or opposition to mitigating action for different domains of inequality. Specifically, in the current work, Americans are less likely to blame individuals for health than economic inequality, and this is associated with greater injustice perceptions and subsequent support for action to reduce LGBTQ health inequalities. Finally, we documented a simple and easily implementable intervention—emphasizing the interconnected nature of health and economic inequality—that increased support for economic inequality mitigation.

While it was unclear (based on past research and theoretical assumptions) whether perceptions of race-based inequality would generalize to perceptions of LGBTQ-based inequality, ultimately, the empirical evidence strongly supports generalization across these two forms of inequality. However, we would like to note that the mechanism of blame (S1a and 1b) as well as the intervention (S3a-3c) were only tested in regard to LGBTQ inequalities. Thus, it remains untested if the findings presented in this manuscript generalize to other forms of inequality, including racial inequality. More generally, whether health inequality (e.g., compared to economic inequality) sparks more support for action when it comes to other forms of inequality (gender, social class, disabilities) remains an empirical question that should be addressed in future research.

The present results suggest that health (vs. economic) disparities can elicit moral sacredness violations and reduced blame, to trigger a sense of injustice and support for reducing these disparities. However, this is not to say that every example of health disparities is likely to elicit these mechanisms similarly. Indeed, there may be specific examples within each larger category of disparities that impacts these mechanisms differently. For example, exposure to sexually transmitted infections (STIs), an area of health, is especially stigmatized due to the perception that those affected are to blame and can control their exposure (35). Thus, if the salient example of health disparities between LGBTQ and non-LGBTQ Americans involves HIV/AIDS, support for mitigating action may be reduced, due to greater perceptions of blame (see ref. 48). While the present research identifies that Americans’ understanding of health disparities broadly elicits less blame than their default conceptualization of economic disparities, future work can detangle what examples within each domain may trigger more or less support for action.

Interestingly, supplementary analyses of the present data reveal that Americans do not perceive health disparities as the most frequent form of inequality faced by the LGBTQ community. Instead, belonging-based disparities are perceived as the most prevalent (additional results from S1a, see SI Appendix). As such, it is possible that the relative novelty of the information as well as bringing health inequality “top of mind” then triggers a cascade of mechanisms. In turn, these mechanisms are associated with a strong drive to want to address and resolve health disparities. Future research could test if surfacing information that is novel or unexpected plays a role in producing the observed effects.

That awareness of LGBTQ health-based inequalities is relatively lower than other forms of inequalities is not necessarily surprising. Issues such as same-sex marriage or “Don’t Ask, Don’t Tell” are frequently discussed in the public discourse. Health disparities receive far less attention, despite that health disparities can be extremely costly for societies. For example, in 2018, racial health disparities cost the US economy 451 billion [up 41 percent from previous estimate in 2014; (49)]. Among other reasons, this is due to the fact that health disparities can have persistent effects on various other life domains, including one’s ability to work and socialize. As such, addressing health-based inequality is a crucial step to ensure a flourishing and functioning society.

Importantly, while the moderation analyses in the present work reveal that greater support for addressing health-based inequality (relative to economic inequality) occurs across political leanings, individuals’ views of what constitutes a health violation may be more complex. For example, denying access to gender-affirming care to minors is hotly debated, perhaps because those opposed perceive the denial of hormones as harmful to a child’s mental health, while those in favor perceive hormone treatments as harmful to a child’s physical health. As such, while both camps may aim to support health, their proposed solutions are irreconcilable. Thus, it is crucial to also determine what ways people support addressing health problems and how the same goal (e.g., protecting children’s health) could produce opposing solutions.

At the same time, there is an urgency to the present work. The risk of suicide among LGBTQ teens may rise due to recent changes in policy and law (4), as the federal government has shut down the LGBTQ youth specialized services program of the national hotline for suicide and crisis prevention [988 Suicide and Crisis Hotline; (50)]. We aim to shed light on the psychology behind responses to LGBTQ inequality and to provide initial insights into when and why Americans support inequality-reducing or inequality-enhancing action. The present research reveals that highlighting the health consequences (vs. economic and belonging consequences) of LGBTQ disparities can mobilize support to address these disparities. This is, in part, because health disparities are seen as less blameworthy and more morally sacred, which is associated with perceptions of health disparities as more unfair. We also show that mentally connecting health disparities to other domains of inequality increases support for disparities that are perceived as less unjust (i.e., economic disparities). While political party leaders fiercely disagree on most issues and policies around LGBTQ rights, we believe in the importance of understanding Americans’ perceptions of this issue in order to start building bridges across political schisms.

Materials and Methods

Studies 1a, 1b, 2, 3b, and 3c were preregistered (S1a: https://aspredicted.org/7t25-2kvj.pdf; S1b: https://aspredicted.org/mwqz-8ptm.pdf; S2: https://aspredicted.org/7mdv-r69g.pdf; S3b: https://aspredicted.org/wn74-vx5f.pdf; S3c: https://aspredicted.org/ggby-wzfn.pdf). The Institutional Review Board of University of California Irvine (IRB-1700), New York University (IRB- FY2018-2070), and The Ohio State University (IRB-2024E1341) approved all consent procedures and protocols. Data and analysis scripts can be accessed at OSF (https://osf.io/kx27a/). For more details and all study materials, see SI Appendix.

Studies 1a and 1b.

Two samples (S1a: N = 820; S1b: N = 825) of US citizens were recruited from MTurk.com in March and April 2024. After removing people who are not US citizens, who fail the Qualtrics bot check (reCAPTCHA score < 0.5), and who fail our attention check (who choose the wrong answer to the close-ended multiple-choice question “What did you read about in the beginning of the study?”), we had a total of 767 participants^¶ (53 participants excluded) in S1a [age range^#: 19–80, M_age = 42.98, SD_age = 12.50; 73 African American, Black, African, or Caribbean; 73 Asian American, Asian, or Pacific Islander; 600 European American, White, or Anglo; 41 Hispanic American, Latino(a), or Chicano(a); 5 Middle Eastern or Arab American, 18 Native American or American Indian; 4 Another not mentioned above] and 802 participants (23 participants excluded) in Study 1b [age range: 18 to 74, M_age = 37.12, SD_age = 13.67; 100 African American, Black, African, or Caribbean; 62 Asian American, Asian, or Pacific Islander; 605 European American, White, or Anglo; 57 Hispanic American, Latino(a), or Chicano(a); 5 Middle Eastern or Arab American, 14 Native American or American Indian; 5 Another not mentioned above]. For S1a, we sought to recruit a minimum of 729 participants based on a power analysis for an ANOVA to detect f = 0.10 for 80% power. For S1b, based on a power analysis (80% power) using the results from S1a, we sought to recruit a minimum of 729 participants for an ANOVA to detect d = 0.2. These samples have 80% power to detect small-sized effects (S1a: f = 0.11; S1b: f = 0.10).

In S1a, after informed consent, participants viewed a short infographic highlighting either health-based, economic-based, or belonging-based LGBTQ disparities (Fig. 1). Following the disparities manipulation, participants reported their support for general policies (2 items; r = 0.88; 1 = strongly disagree, 7 = strongly agree) and concrete policies (2 items; r = 0.64; 1 = strongly oppose to 7 = strongly support). Participants then indicated how much they think there are disparities between LGBTQ and non-LGBTQ Americans in the following domains: health, economics, and belonging in the community (with additional filler options: politics, college admission, and dating). All materials are available in the SI Appendix.

In Study 1b, after informed consent, participants were either exposed to the health or economic LGBTQ disparity infographic (Fig. 1). After reading the infographic, participants reported their support for general collective action (1 item; 1 = strongly disagree, 7 = strongly agree) and concrete collective action items to reduce these disparities (3 items; α = 0.87; 1 = very unlikely, 7 = very likely).

In both studies, participants also reported their agreement (1 = strongly disagree, 7 = strongly agree) if LGBTQ disparities were a moral sacredness violation (3 items, e.g., “These disparities involve issues or values which should never be violated”; S1a: αs = 0.92; S1b: αs = 0.93), were unjust (2 items, e.g., “These disparities are unjust”; S1a: r = 0.74; S1b: r = 0.76), and whether LGBTQ individuals were to blame for these disparities (4 items, e.g., “The way that LGBTQ individuals behave causes these [health-based/economic-based] disparities”; S1a: αs = 0.93; S1b: αs = 0.93). Finally, participants completed demographics and were debriefed (for similar methodology see ref. 46).

Study 2.

Our audience for Study 2 (that is, individuals who would be randomly assigned to the ad manipulation) consisted of adult Facebook users (aged 18+), located in the United States, who accessed the newsfeed feature on their desktops. An individual user could see only one of the ad versions and, to count as a unique view, each user could see it only once. A total of 33,797 users were presented with the ads between 26 and 27 December 2023 for 24 h.

Consistent with past research (28), Study 2 used Facebook’s ad manager to run the two different advertisements in a split test, such that ads are tested against each other to determine which ad performs best in terms of the marketing objective (i.e., the “reach” as proxy for measuring engagement). Facebook determines the winning ad by calculating the cost for each ad to reach 1,000 people (Facebook can also declare no winner if the ads perform equally well). In addition, Facebook reports the absolute number of people reached by each ad. As an ad format, we used a single-image advertisement, which included a headline, descriptive text, an image, and a website URL (SI Appendix). Each ad was labeled as sponsored content. We used the same free stock image and NYU Qualtrics URL for each ad. If users clicked on the ad, they were forwarded to a form, which included the contact information of the researchers and debriefing information (consent was waived by the IRB). Each ad presented a headline about LGBTQ inequality: one focusing on health and the other focusing on economic disparities (for similar methodology see ref. 46).

Studies 3a–3c.

Three samples (S3a: N = 200; S3b: N = 400; S3c: N = 525) of US citizens were recruited from MTurk.com in January, February, and March 2025. After removing people who are not US citizens, who fail the Qualtrics bot check, and who fail our attention check from S1, we had a total of 195 participants (five participants excluded) in Study 3a [age range: 19 to 67, M_age = 37.43, SD_age = 10.15; 24 African American, Black, African, or Caribbean; 32 Asian American, Asian, or Pacific Islander; 131 European American, White, or Anglo; 23 Hispanic American, Latino(a), or Chicano(a); 1 Middle Eastern or Arab American, 2 Native American or American Indian; 1 Another not mentioned above] 386 participants (14 participants excluded) in Study 3b [age range: 18 to 77, M_age = 39.38, SD_age = 11.95; 46 African American, Black, African, or Caribbean; 30 Asian American, Asian, or Pacific Islander; 292 European American, White, or Anglo; 29 Hispanic American, Latino(a), or Chicano(a); 5 Middle Eastern or Arab American, 5 Native American or American Indian; 3 Another not mentioned above] and 494 participants (31 participants excluded) in Study 3c [age range: 18 to 82, M_age = 38.13, SD_age = 11.66; 71 African American, Black, African, or Caribbean; 50 Asian American, Asian, or Pacific Islander; 348 European American, White, or Anglo; 41 Hispanic American, Latino(a), or Chicano(a); 8 Middle Eastern or Arab American, 10 Native American or American Indian; 2 Another not mentioned above]. For S3a, we sought to recruit around 200 participants as a pilot. For S3b, based on a power analysis (80% power) using the results from S3a, we sought to recruit a minimum of 336 participants for a mixed ANOVA to detect d = 0.14; and for S3c we sought to recruit a minimum of 500 participants for a mixed ANOVA to detect d = 0.14 at 90% power. These samples have 80% power to detect small-sized effects used in the present study (S3a: f = 0.10; S3b: f = 0.07; S3c: f = 0.06).

In Studies 3a–3c, after informed consent, participants read a short blurb about LGBTQ inequality emphasizing that disparities exist between LGBTQ and non-LGBTQ Americans, which affect many areas of life, such as people’ s health and their economic situation. Participants then reported their general support for reducing LGBTQ-based disparities in health and economics in a pretest questionnaire (“I support policies that aim to reduce [health-based/economic-based] disparities between LGBTQ and non-LGBTQ Americans”; “I really want to take action to reduce [health-based/economics-based] disparities between LGBTQ and non-LGBTQ Americans”; 2-items for health; rs > 0.83; and 2 items for economics; rs > 0.84). Participants were then exposed to an infographic emphasizing either that economic disparities cause health disparities (S3a: n = 99; S3b: n = 191; S3c: n = 243) or that there are economic disparities in addition to health disparities (S3a: n = 96; S3b: n = 195; S3c: n = 251; in both example conditions we provided the same examples of economic- and health-based LGBTQ disparities; see SI Appendix for materials). Afterward, in a posttest questionnaire, participants completed the same general action support items (as before the manipulation) for both health (rs > 0.84) and economics (rs > 0.84). In Study 3c, participants were asked to report whether they think health and economic disparities are related (open-ended), as a manipulation check. Finally, participants completed demographics and were debriefed.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

Anonymized quantitative data have been deposited in Open Science Framework (https://osf.io/kx27a/) (51). All other data are included in the manuscript and/or SI Appendix.