INSTRUMENTAL LGBT+ TOLERANCE: THE HOMONATIONALIST THESIS

Mass support and tolerance toward LGBT+ individuals in democratic nations has increased overtime. While homonegative biases, trans-exclusion, outright homophobia and transphobia remain present in society, transformations in institutional recognition, social tolerance, and public support for LGBT+ individuals and public policies that would seek to expand this group’s welfare, have been steadily increasing across liberal democratic nationsFootnote ⁴, moving sexual and gender minorities from being a sexual out-group to part of the “mainstream” societal in-group. The explanation for this can be found in increased inter-group contact (Ayoub and Garretson Reference Ayoub and Garretson2017; Garretson Reference Garretson2018)Footnote ⁵, generational replacement (Ekstam Reference Ekstam2023), active in-group and elite-level persuasion efforts (Harrison and Michelson Reference Harrison and Michelson2017a), as well as positive policy feedback effects whereby state-sponsored recognition triggers a norm-establishing effect that induces greater social tolerance and results in intolerance becoming more socially unacceptable (Abou-Chadi and Finnigan Reference Abou-Chadi and Finnigan2019).

As detailed by Harrison and Michelson (Reference Harrison and Michelson2017a), the rapid expansion of increased social tolerance toward LGBT+ individuals and support for LGBT+ rights is, in many regards, remarkable. Instead of resulting from the slow effects of cohort replacement, whereby more socially liberal generations emerge to replace those generations with more conservative values, a rich catalog of evidence from panel studies (Flores and Barclay Reference Flores and Barclay2016) and experiments (Ayoub, Page, and Whitt Reference Ayoub, Page and Whitt2021; Brookman and Kalla Reference Brookman and Kalla2016; Harrison and Michelson Reference Harrison and Michelson2017a; Reference Harrison and Michelson2017b) has demonstrated that, in the case of LGBT+ rights, individual-level changes in attitudes play a significant role in explaining these macro-level trends. The liberalization of attitudes toward LGBT+ individuals is normatively desirable as it has concrete benefits for the psychological welfare and well-being of LGBT+ individuals (Boertian and Vignoli Reference Boertian and Vignoli2019) and, as a result, understanding the amenability and conditionality of LGBT+ inclusion is of widespread importance.

In contrast to the more benevolent explanations that help us understand the liberalization of attitudes toward LGBT+ individuals and support for LGBT+ specific policy provisions, a homonationalist interpretation presents a more critical perspective. As detailed by Spierings, Lubbers, and Zaslove (Reference Spierings, Lubbers and Zaslove2017) and Lancaster (Reference Lancaster2020), an increasingly large proportion of nativists and radical right-wing voters—those that, conventionally, exhibited both traditional heteronormative and patriarchal values alongside nativist, anti-liberal, and nationalist values (Kitschelt Reference Kitschelt1997)—are liberalizing their views of and support for LGBT+ individuals. Observing trends in tolerance toward homosexuality among nativist individuals using data from the European Social Survey (ESS), there is strong evidence of increasing tolerance among nativists in Western Europe, where LGBT+ rights are increasingly nationalized, that is absent among nativist peers in Eastern European countries, where LGBT+ rights have not (yet) been equally nationalized (see Supplementary Figure A1).Footnote ⁶ Homonationalism theorizes that such moves in mass opinion, particularly among those we would expect to be most resistant to issues of LGBT+ rights, toward LGBT+ tolerant social liberalism are, in part, the product of instrumental considerations (Berntzen Reference Berntzen2019) as opposed to a more genuine transformation in attitudes. Similar to evidence signaling the amenability and superficiality of support for liberal democracy and tolerance of violations in democratic norms to in-group dynamics (Graham and Svolik Reference Graham and Svolik2020; Simonovits, McCoy, and Littvay Reference Simonovits, McCoy and Littvay2022), the homonationalist thesis posits that support for LGBT+ rights depends on associations between sexual freedoms and the national ethnic in-group. As a result of the long, and often muddied, social road that has been traveled in the “nationalization of liberal values” in many Western countries (Lægaard Reference Lægaard2007), as well as the archetypal dissociation between LGBT+ rights and the Muslim ethnic out-group (Berntzen Reference Berntzen2019; Rahman Reference Rahman2014), the ground is fertile for the instrumental tolerance of LGBT+ citizens to be adopted in order to differentiate the “nation” from the other (Lægaard Reference Lægaard2007).Footnote ⁷

The empirical literature, relying on observational analyses, supports this idea: in those countries where institutional recognition of LGBT+ rights is greatest, the proportion of individuals who simultaneously harbor more tolerant views toward homosexuality alongside more intolerant views toward ethnic out-groups is significantly higher (Kwon, Scarborough, and Taylor Reference Kwon, Scarborough and Taylor2023). These “sexually modern nativists” (Spierings, Lubbers, and Zaslove Reference Spierings, Lubbers and Zaslove2017) are not trivial in number and some evidence signals their numbers are actually ascending (Lancaster Reference Lancaster2020). A simple descriptive visualization of simultaneously held attitudes toward homosexuality and immigration among respondents in the ESS identifies proportions in excess of 30% in many country-years (Supplementary Figure A2) across Europe, but particularly in Western Europe, between 2002 and 2020. The prevalence of this group of citizens reflects, and perhaps stem from, active engagement by supply-side political actors who seek to leverage national-based and ethnic-based social identities to engender an image of the ethnic “other” (specifically, Muslims) as an inimical threat to members of the national in-group. State governments, most frequently that of Israel (Gross Reference Gross2015; Puar Reference Puar2007; Reference Puar2013), the media (Dhoest Reference Dhoest2020), radical right-wing parties (Siegel Reference Siegel2017), and other radical right-wing groups (Foster and Kirke Reference Foster and Kirke2023; Jennings and Ralph-Morrow Reference Jennings and Ralph-Morrow2020) in continental Europe (Akkerman Reference Akkerman2005; Hunklinger and Ajanović Reference Hunklinger and Ajanović2022; Siegel Reference Siegel, Vries-Jordan and AndersonForthcoming; Spierings Reference Spierings2021; Turnbull-Dugarte Reference Turnbull-Dugarte2021; Reference Turnbull-Dugarte2022b), the United Kingdom, (Lockhart Reference Lockhart2022), and the United States (Murib Reference Murib2018) have been actively pursuing homonationalist strategies. Examples of these strategies—as employed by the German far-right party Alternative für Deutschland (AfD), the leader of the Netherlands’ far-right Partij voor de Vrijheid (PVV), Israeli campaigners, and the UK’s former far-right party (UKIP)—are shown in Figure 1 and demonstrate rhetoric similar to that of Trump discussed in this essay’s opening. Far-right parties in countries that boast high levels of public acceptance toward the LGBT+ community, frequently leverage the defense of LGB (but often not T+)Footnote ⁸ citizens and their rights as a motivation to rationalize their opposition to ethnic others who they portray as inimical to so-called Western values including—among wider signals of liberalism such as the inclusion and protection of women as well as other “women’s issues” (de Lange and Mügge Reference de Lange and Mügge2015; Farris Reference Farris2017; Weeks et al. Reference Weeks, Meguid, Kittelson and Coffé2023)—the defense of sexual minorities (Akkerman Reference Akkerman2005; Hunklinger and Ajanović Reference Hunklinger and Ajanović2022; Siegel Reference Siegel2017; Reference Siegel, Vries-Jordan and AndersonForthcoming).

Figure 1. (Comparative) Examples of Supply-Side Homonationalism

Note: German translation (a): My partner and I do not value the acquaintance of Muslim immigrants, for whom our love is a mortal sin. Dutch translation (b): Stop the gay hate. Stop Islam.

The increasing tolerance of LGBT+ issues among nativists and the far-right, be they ordinary citizens or elite-level actors, presents something of a paradox. One empirical puzzle, for example, remains: it is not clear if the movement toward socially liberal positions on LGBT+ issues is the result of a genuine and organic transformation in attitudes toward this out-group, or rather nativist individuals are instrumentally and strategically diluting their opposition to LGBT+ rights issues in order to further socially demarcate themselves from the ethnic out-group. The instrumentally liberal homonationalist thesis posits that it is the latter. Nativists, we argue, see LGBT+ rights as part of the bundle of liberal positions viewed as a cornerstone of Western values—and self-congratulatory sense of exceptionalism (Rahman Reference Rahman2014)—that allows them to draw a line with the ethnic out-group which they caricature as inimically intolerant, misogynistic, and homophobic. Empirically testing these expectations is fraught with limitations given the constraints of observational data analyses. We seek to overcome these limitations by experimentally manipulating group-based opposition to LGBT+ rights.

Should more amicable positions toward the sexual out-group be truly independent of views toward the ethnic out-group, experimentally manipulating the ethnic out-group status should not influence preferences for policies that benefit the sexual out-group (organic liberalism thesis). Should, however, more positive dispositions toward the sexual out-group emerge when those hostile to the ethnic out-group are randomly exposed to anti-LGBT+ positions by the same out-group, this would indicate that sexually liberalism is the result of instrumental preference formation (instrumental liberalism thesis). Before laying out our empirical test of these opposing theoretical expectations, we present a theoretical model that lays out an intuitive road map of how homonationalism can engender psychological processes that result in instrumental tolerance to LGBT+ rights expansion.

A PSYCHOLOGICAL THEORY OF HOMONATIONALISM

We theorize that support for LGBT+ rights can be driven by a desire to create social disidentification with those perceived as opposed to LGBT+ rights. Disidentification with a certain group, as argued by Elsbach and Bhattacharya (Reference Elsbach and Bhattacharya2001), can be expressed by one seeking to actively define themselves as not harboring the attributes, features, and values that the individual believes to define that same group. Social identity theory (Tafjel Reference Tafjel1974; Tafjel and Turner Reference Tafjel, Turner, Austin and Worchel1979) establishes that individuals are incentivized to engage in activity that enhances one’s positive self-perceptions and, as a result, where associations with a certain group run the risk of undermining one’s perceived social value, disidentification and a desire to alter one’s identity to augment the marked social distinctiveness from the group is high (Elsbach and Bhattacharya Reference Elsbach and Bhattacharya2001; Turner Reference Turner1975). In short: when individuals are presented with a threat to the integrity of their identity and sense of self resulting in shared attributes or values with other identity groups that are perceived negatively, individuals—out of fear of being “tarred by the same brush”—engage in disidentification strategies in order to enlarge the space between their own identity and that from which they wish to present themselves as distinct (Brown and Williams Reference Brown and Williams1984).

We expect these strategies to play a role in shaping concrete policy preferences. Just as primes of an in-group identity that cue in-group-based sympathy for a policy can engender increased support (Harrison and Michelson Reference Harrison and Michelson2017a; Ostrom et al. Reference Ostrom, Carpenter, Sedikides and Li1993), a simple corollary of this phenomenon would be that primes of out-group identities that cue out-group-based opposition for a position can also engender increased support for the same position. Put simply, instead of adopting political preferences that are shaped and defined by who we are, we theorize that associations between concrete issue positions and salient out-groups can result in adopting political preferences that are defined by who we are not.

Our theoretical proposition is founded in Heider’s (Reference Heider1958) balance theory. Balance theory assumes that individuals, seeking to maintain their idea of self cognitively balanced, are incentivized to identify, establish, and maintain relationships that provide balance between their personal preferences and their personal affect toward groups and individuals. In other words, individuals seek to maintain a positive balance between those issues they endorse and those individuals they identify as their friends, alongside a symmetrical balance between those issues they reject and those individuals they identify as their enemies. If an individual discovers one of their friends disagrees with them on a salient political issue, then they are out of balance. Symmetrically, an individual is also out of balance should they discover that one of their enemies agrees with them on a salient issue. In these scenarios, the individual is presented with two alternatives: either disidentify with their issue position and maintain their kinship group, or disidentify from their kinship group and maintain their issue position. We know from a rich body of work assessing the efficacy of identity primes—including, in the specific case of LGBT+ rights, Harrison and Michelson’s (Reference Harrison and Michelson2017a) work on dissonant identity priming—that individuals are likely to opt for the latter option. Essentially in-group priming effects are an effective means of persuasion given individuals are more inclined to be receptive to cues and, by extension, prone to update their preferences to match those of respected members of the same identity-based in-group and reject those of out-groups (Ostrom et al. Reference Ostrom, Carpenter, Sedikides and Li1993).

A corollary of both balance theory and identity priming is that primes from social identities that simultaneously signal out-group membership and out-group opposition to an orthogonal out-group, then in-group members will be inclined to close the gap between one of these out-groups. In short, the premises of balance theory lead us to expect a pattern captured by the “enemy of my enemy is my friend” proverb outlined above.

Empirical support for this corollary of balance theory has been found in both experimental and ethnographic applications of the theory. Borgeson and Valeri (Reference Borgeson and Valeri2007), for example, explore how shared out-group dislike for members of the Jewish faith explains the establishment of formal alliances and collaborations between members of the Aryan Nation and followers of the Islamic Jihad. Similarly, Whigham (Reference Whigham2014) explores how out-group animosity toward the English among Scottish football fans drives the latter to support any rival team, or “Anyone but England!” (Abell Reference Abell2011; Whigham Reference Whigham2014), in competitive football matches. In a laboratory experiment where subjects’ sense of affinity (friendship) and animosity (enmity) toward concrete individuals was experimentally induced, Aronson and Cope (Reference Aronson and Cope1968) show that subjects were significantly more inclined to reward independent third-party individuals who exercised animosity toward those with whom the subjects also held animosity. In other words, the enemy of a subject’s enemy, was rewarded like a friend.

Now imagine the following scenario. A nativist individual who holds negative affect toward Muslims (Out-Group 1) as well as the LGBT+ population (Out-Group 2) is presented with an identity prime that signals the identity-based association between the rejection of the LGBT+ population with the ethnic out-group. This information induces cognitive discomfort, and signals a lack of balance between their group-based affect: if Group 2 is the enemy of my enemy (Group 1), should I update my views on the sexual out-group or do I accept that I am, in fact, not dissimilar to the ethnic out-group?

This imbalance incentivizes the individual—assuming they are unwilling to accept this psychological discomfort of this dilemma—to make one of two choices to re-establish cognitive balance. This dilemma, and the resulting alternatives detailed below, can be summarized succinctly as a decision between defining oneself as we who are (e.g., anti-LGBT) or who we are not (e.g., Muslim).

Option A—the individual retains their negative affect toward the LGBT+ population and accepts the cognitive cost of reducing the identity-based distinctiveness between their national in-group identity and that of the ethnic out-group.

Option B—the individual updates their view toward the LGBT+ population in order to maintain the identity-based distinctiveness between their national in-group identity and the ethnic out-group, rejecting the cognitive costs of reduced inter-group distinctiveness and association with the out-group.

STUDY 1: U.K. EXPERIMENT

To test these expectations, we rely first on an original survey experiment fielded in the United Kingdom to a representative sample of online panel respondents in March 2021 ( $ N=1,200 $ ). Our experimental research design asks: is support for LGBT+ rights driven by out-group disidentification?

To measure support for LGBT+ rights, we analyze support for LGBT+ inclusive education in schools. LGBT+ education is of recent salience in the United Kingdom, where the experiment is fielded, as well as elsewhere, and represents a contemporary LGBT+ policy issue that remains contentious even in those countries where equal marriage has been legalized for some time. In the United States, for example, recent state-level actions have outlawed the discussion of LGBT+ issues in schools such as that evinced by the so-called Don’t Say Gay! bill in Florida (Goldstein Reference Goldstein2022; Philips Reference Philips2022). Similar moves toward state-sponsored prohibitions on LGBT+ inclusive education have been observed in the comparative context including in countries with precedents of low-level (e.g., Hungary; [Rankin Reference Rankin2021]) and high-level (e.g., Spain; [Mateo Reference Mateo2021]) tolerance of LGBT+ individuals. Given attitudes toward LGBT+ education is an outcome that captures specific policy support toward a concrete, and politicized, LGBT+ policy concern, we argue it serves as an ideal variable for testing the presence of homonationalist instrumental liberalism. Theoretically, however, we would expect similar empirical relationships to be observed with other viable outcomes measuring specific support for LGBT+ policies.

The dependent variable is recorded via the following survey question: Some people think that schools should discuss lesbian, gay, bisexual, and transgender (LGBT+) issues in the classroom and others think that schools should not discuss these matters. Where would you place yourself on this scale? Responses were measured on an 11-point scale ranging from “Discuss LGBT+ issues in school” (0) to “Do not discuss LGBT+ issues in school” (scale order randomized). For parsimony, the results reported below rely on dichotomized support for the policy (0–1). Estimations using the full continuous scale are reported in the Supplementary Material (see Supplementary Table A8 and Supplementary Figure A4). The conclusions remain constant across operationalizations.

To test if individuals’ views toward LGBT+ policy concerns are shaped by strategically instrumental liberalism, we leveraged an experimental research design that randomly assigned individuals to one of two treatment conditions: in-group condition (control) and out-group condition (treatment). Immediately before being asked to express their views on LGBT+ education, both groups were exposed to a short news article about opposition to the policy. Those in the control condition were informed of opposition to the policy among non-denominational families and three named protesters with conventional white-British names (Daniel Smith, James Wilson, and Karen Jones). Those in the treatment condition were informed of opposition to the policy explicitly among Muslim families and three named protesters with conventional Muslim names (Shakeel Afsar, Amir Ahmed, and Farah Begum). Both treatment texts (visualized in Figure 2) were also accompanied by images: the image accompanying the treatment group text displays individuals in traditional Muslim dress (Niqab) holding protest placards, whereas the image accompanying the control group text displayed the placards without revealing those who held them. Importantly, the texts presented in both conditions were presented as independent of any political affiliation and were not, as a result, attached to any elite-level actor that may be actively engaged in leveraging homonationalist talking points. Of note is that the treatment scenarios reflect anti-LGBT+ protests that have actively taken place. These include the salient, if somewhat exceptional, Muslim protestsFootnote ⁹ against LGBT+ education in the United Kingdom (for a discussion of these protests and the subsequent judicial action that resulted from these, see Vincent Reference Vincent2017), as well as protests in Spain, organized by the Islamic community and other far-right actors, in favor of a parental veto [PIN parental] on LGBT+ content in the classroom (Moreno Reference Moreno2022).

Figure 2. Experimental Design (UK)

Our theoretical argument posits that those individuals who harbor negative predispositions toward immigration and are subsequently exposed to the treatment condition will display more socially liberal preferences toward LGBT+ education than individuals with comparable anti-immigrant sentiment in the control condition. The empirical test of this expectation is summarized in Model 1 in which $ {\delta}_1Treatmen{t}_i $ is dichotomous indicator of exposure to the out-group (Muslim) treatment (1) or control (0), and $ {\beta}_1ImmigrationAttitude{s}_i $ is the moderating variable indicating attitudes (recorded pre-treatment) toward immigration. Attitudes toward immigration are recorded on an 11-point scale (0–10) with higher values indicating more positive (liberal) positions toward immigration.Footnote ¹⁰ The estimand of interest is the conditional average treatment effect (CATE)—that is, the effect of exposure to treatment conditioned by pre-treatment attitudes toward immigration. Empirically, we estimate the CATE via both a linear estimation of the conditionality of moderator values and a dichotomous indicator based on below- and above-mean attitudes. Power calculations for the estimation of the CATE are reported in Supplementary Figure A8.

(1) $$ {Y}_i={\displaystyle \begin{array}{l}\alpha +{\delta}_1Treatmen{t}_i+{\beta}_1ImmigrationAttitude{s}_i\\ {}+\hskip2px {\delta}_1Treatmen{t}_i*{\beta}_1ImmigrationAttitude{s}_i+{\epsilon}_i.\end{array}} $$

The empirical results from Study 1 are summarized in Figures 3 and 4. The full regression output, including the overall average treatment effect (ATE) in the full sample, is reported in Supplementary Tables A7 and A8.Footnote ¹¹ Figure 3 visualizes the effect of random exposure to the treatment (ethnic out-group) prompts vis-à-vis control across pre-treatment levels of support for immigration where higher values on the x-axis indicate a more liberal immigration position. The upper panel visualizes the predicted probability of supporting LGBT+ education in schools across observed values of the moderator, and the lower panel reports the CATE across these same values.

Figure 3. Treatment Effect across Distribution of Immigration Preferences (UK)

Note: Full regression output in Supplementary Table A7.

Figure 4. Treatment Effect among Those with Dichotomised Immigration Preferences (UK)

Note: Full regression output in Supplementary Table A7. Treatment group outcome statistically distinct at p<0.1(*), p<0.05(**), and p<0.01(***).

As one might expect, support for LGBT+ inclusive education and liberal attitudes toward immigration are correlated, signaling their coalesced nature as socially progressive issue positions: the more one is supportive of immigration, the more likely it is one is supportive of progressive LGBT+ policies (upper panel of Figure 3). Should individuals’ support for LGBT+ education be genuine and independent of the out-group identity of those opposed to the policy, we would expect the slopes of those in the control and treatment groups to consistently overlap across different values of the moderator. They do not. Those exposed to treatment and with conservative (illiberal) views on immigration deviate significantly from the baseline counterfactual trend observed by those in the control group with similar views on immigration.

Consistent with our theoretical expectations, the results provide causal evidence that individuals who harbor anti-immigrant preferences become significantly more inclined to adopt more amicable preferences toward LGBT+ inclusive education when exposed to news of opposition to the same by the ethnic out-group. Among those positively predisposed toward immigration, treatment exhibits no significant ( $ p<0.05 $ ) effect if, however, the CATE is negatively signed. The same is not true of those with negative views toward immigration. An individual with views on immigration 4 on an 11-point scale, becomes 7.7 percentage points ( $ p<0.05 $ ) more likely to support LGBT+ education when exposed to ethnic out-group opposition to the policy. The conditional ATE is greatest among those with the most negative dispositions toward immigration who become 14.7 percentage points ( $ p<0.05 $ ) more likely to express socially liberal preferences toward LGBT+ issues than their equally anti-immigrant peers in the control group. Importantly, these effects are not conditioned by linearity assumptions of the multiplicative interaction (Hainmueller, Mummolo, and Yiqing Reference Hainmueller, Mummolo and Yiqing2019) as demonstrated in Supplementary Figures A6 and A7.

Figure 4 visualizes a parsimonious test of the instrumentally liberal thesis by dichotomously stratifying the sample in relation to their views on immigration relative the sample mean. The left-hand panel reports the effect of the out-group treatment message among those with positive dispositions toward immigration (those equal to or greater than the sample mean). The right-hand panel reports conditional treatment effect among those who have a more negative immigration stance (a position below the sample mean). As in the case of the continuously operationalized linear measure (Figure 3), and in line with the expectation of our instrumental liberalism thesis, the results demonstrate that treatment exposure to out-group opposition to the LGBT+ policy only influences support for the policy among those with negative predispositions toward the ethnic out-group. Respondents with below-average dispositions toward immigration become 9.7 percentage points ( $ p<0.05 $ ) more inclined to favor LGBT+ inclusive education vis-à-vis those with symmetrical anti-immigrant attitudes in the control condition. Those with illiberal immigration views have a baseline (control group) probability of supporting LGBT+ inclusive education in schools of 0.45. A treatment-engendered effect of 9.7 percentage points is consequently equatable to a 22% increase against the baseline.

A rich catalog of sensitivity tests are reported in the Supplementary Material, including a comprehensive multiverse analysis (Simonsohn, Simmons, and Nelson Reference Simonsohn, Simmons and Nelson2020) with variation in model specifications and covariate inclusion (see Supplementary Figures A10–A13), which demonstrates the robustness of these findings. In some specifications, the CATE among respondents with positive dispositions toward immigration is significantly negative. These indicative findings suggest that—at least in the British case—among those with strong pro-immigrant preferences, attitudes on LGBT+ issues may actually become less positive. This would imply that non-nativists update views on the sexual out-group negatively as a result of increased sympathies with the ethnic out-group. Theoretically, these results could be the result of those with positive affect toward migrants seeking to identify with the preferences of ethnic out-groups. We do, however, exercise caution in our interpretation of the CATE among pro-immigrant respondents given the significance of the results is not robust to sensitivity tests and, as we reconsider in Study 2, this finding does not replicate outside of the British context.

The results of our first study provide robust empirical support for our primary thesis. Consistent with a homonationalist interpretation of the nationalization of LGBT+ tolerant values and the corollaries of balance theory, we find that exposure to ethnic out-group opposition to LGBT+ rights causes a significant and substantive uptick in support for LGBT+ rights among those with nativist attitudes in the United Kingdom. We now turn to answer two additional subsequent questions that emerge from these findings.

STUDY 2: SPAIN EXPERIMENT

In Study 2, we replicate the test of our theoretical argument via an additional experiment fielded on an online crowd-sourced (Prolific) sample of survey respondents in Spain ( $ N=1,200 $ ) in March 2022. The sample is weighted to approximate population parameters based on gender, age, education, and geographical region. The value added of Study 2 is twofold. First, we sought to assess the external validity of the primary findings of Study 1, asking: are these results externally valid and replicable beyond the concrete British case? Second, we seek to expand upon our findings by introducing additional components in our design that facilitate supplementary tests which assess treatment’s effect on ancillary outcomes that may give insight into the underlying mechanisms behind the instrumental liberalism we observe, asking: does ethnic out-group opposition result in increased national pride in liberal “Western” values?

Britain is arguably a “most likely” case for us to observe instrumental homonationalist liberalism. First, Britain is a country where opposition to immigration and out-group animosity toward ethnic out-groups, including Muslims (Donnaloja Reference Donnaloja2022; Ford Reference Ford2011), has been comparatively high compared to that observed among the country’s European peers (Dennison and Geddes Reference Dennison and Geddes2019). Second, Britain has, until very recently, played host to a politically influential radical right-wing party (UKIP), and other radical right-wing groups (Jennings and Ralph-Morrow Reference Jennings and Ralph-Morrow2020), that have actively pursued homonationalist strategies (Foster and Kirke Reference Foster and Kirke2023; Lockhart Reference Lockhart2022).

Spain deviates from the British case on a number of significant variables, not least including in the country’s electoral and party systems. First, and of theoretical significance for our argument, the salience of concerns related to immigration and ethnic out-group (Muslim) rejection is, unlike the United Kingdom, marginal and significantly below the average observed in other West European nations (Dennison and Geddes Reference Dennison and Geddes2019; Encarnación Reference Encarnación2004). In a low-salience context and where aggregate public opinion on non-European migration is consistently net positive (Dennison, Kustov, and Geddes Reference Dennison, Kustov and Geddes2023), the incentives to demarcate oneself by in-group and out-group status are reduced and, consequently, may have less of a moderating role on responses to out-group positions. Second, public support for LGBT+ rights issues, as cross-national observational data on LGB and T+ inclusive education in Europe demonstrates (Supplementary Figure A3), is particularly high in Spain and the country, under the leadership of the then-governing Socialist party (PSOE), was one of the pioneering “early movers” on equal marriage legislation (Calvo Reference Calvo2007; Kollman Reference Kollman2007). Spain legalized equal marriage in 2005, some 8 years before similar legislation was introduced in the United Kingdom. Third, while Spain does have an electorally successful far-right party in the form of VOX, the party, rather than homonationalist, is more explicitly and outspokenly opposed to LGBT+ rights (Rama et al. Reference Rama, Zanotti, Turnbull-Dugarte and Santana2021).Footnote ¹² Recent anti-LGBT+ policies pursued by VOX, which is currently Spain’s third largest national party in terms of both votes and parliamentary representation, include efforts to introduce a parental veto on LGBT+ inclusive education, prohibiting the presence of the rainbow flag on public buildings, the repeal of prohibitions on conversion therapy, and the derogation of LGBT+ anti-discrimination laws. Given these distinct factors, we consider Spain a “less likely” case and, as a result, a replication of our original experiment in this country represents a tough test of our theoretical argument that would provide notable value-added in terms of broader external validity.

The core dependent variable (support for LGBT+ education) and the moderator (pre-treatment attitudes toward immigration) are measured and operationalized symmetrically in Studies 1 and 2. The treatment text and accompanying images in Study 2, as shown in Figure 5, do vary from those in the original experiment. The treatment message in Study 2 in Spain focuses on protests specifically around the use of LGBT+ inclusive textbooks. As in the case of Study 1, out-group (Muslim) identities were primed using both visual cues based on individuals in traditional Muslim dress, as well as named organizations (e.g., Muslim Parents for Freedom) and typical Muslim names (Farah Begum, Fatima Bennani, and Mohammed El Idrissi).

Figure 5. Experimental Design (Spain)

Note: English translation of original Spanish treatment text reported in Section E of the Supplementary Material.

In Study 2, we build on our first experimental study by introducing a new survey item in order to test ancillary outcomes that may indicate the mechanisms at play. After recording the core outcome measure, we included a question regarding the extent to which individuals felt national pride in a battery of outcomes. Of interest to our theory was the extent to which individuals felt pride in liberal “western” values. We did, however, also solicit levels of pride in several placebo items which, theoretically should not be influenced by the potential for out-group threat primed in treatment, including: EU norms, green policies, domestic violence laws, the Spanish flag, and the army. Should part of the support for LGBT+ rights engendered by treatment allocation be the result of out-group opposition priming individuals regarding the nationalization of liberal values (Lægaard Reference Lægaard2007) among the “enlightened West” (Siegel Reference Siegel2017), we would expect to observe a simultaneously induced uptick in pride in these values to result from treatment assignment.

The primary results of our experimental manipulation in Study 2 are visualized in Figures 6 and 7. The upper panel of Figures 6 visualizes the predicted probability of supporting LGBT+ education among those randomly exposed to ethnic out-group opposition (treatment) and ethnic in-group opposition (control) to LGBT+ inclusive education in Spanish schools across observed values of attitudes toward immigration. The lower panel reports the CATE across these same pre-treatment immigration attitudes. As in the case of Britain, Spanish respondents with more liberal attitudes toward immigration are more inclined to support LGBT+ education. In other words, nativism is, on average, a strong predictor of opposition toward more liberal and inclusive LGBT+ policies. Experimentally exposing Spanish respondents to ethnic out-group opposition to LGBT+ rights, however, significantly increases support for the liberal LGBT+ policy vis-à-vis those randomly exposed to ethnic in-group opposition and this effect is significantly larger among those with nativist preferences.

Figure 6. Treatment Effect across Distribution of Immigration Preferences (Spain)

Note: Full regression output in Supplementary Table A9.

Figure 7. Treatment Effect among Those with Dichotomised Immigration Preferences (Spain)

Note: Full regression output in Supplementary Table A9. Treatment group outcome statistically distinct at p<0.1, p<0.05(**), and p<0.01(***).

As visualized in the lower panel of Figure 6, the treatment-induced uptick in support for LGBT+ education vis-à-vis the control group counterfactual is present across all values of moderator. In other words, although treatment induces a significantly larger increase in pro-LGBT+ policy preferences among those with more nativist attitudes, treatment also exhibits a small increase in LGBT+ policy support among those without nativist views. This latter finding is distinct from that of Study 1 where indicative evidence pointed toward opposite effects among individuals based on their position related to immigration. This is more explicitly apparent when the sample is stratified based on respondents’ immigration position relative the sample mean (Figure 7). Treated respondents in Spain, regardless of whether or not they held nativist views, reported support for LGBT+ education that was 10 or 11 percentage points higher, respectively, than their peers with similar (non)-nativist views. Given that each group has a different baseline, this does indicate that the magnitude of the effect among anti-immigrant respondents is notably larger. While for pro-immigrant respondents exposure to treatment resulted in a 14% change in the outcome vis-à-vis the control group-based counterfactual, the same treatment message resulted in a 21% change among anti-immigrant respondents.

Ancillary Evidence: Pride in “Western Freedoms”

In addition to the core dependent variable (support for LGBT+ education), in Study 2, we further explore the homonationalist mechanism with an additional indicator (pride for the freedoms inherent to Western lifestyle). Does exposure to ethnic out-groups promoting positions at odds with LGBT+ tolerance result in a shift in citizens’ pride in the liberties of the West?

The results visualized in Figure 8 demonstrate that, consistent with our theory of instrumental liberalism, nativist individuals exposed to ethnic out-group opposition to LGBT+ rights, in addition to becoming more supportive of LGBT+ right themselves, also report significantly higher feelings of national pride in the liberties and freedoms afforded by their country’s “Western lifestyle.” This positive effect is not observed among those with more amicable views on immigration where the observed effect is the reverse (negative) if, however, notably smaller than that observed among nativists. In other words, while treatment induced a positive effect on LGBT+ policy support among all respondents, the same is not true when it comes to pride in western values where positive treatment effects are unique to nativists. Estimating the conditional effect of treatment on national pride on a number of placebo concerns including “green” policies, domestic violence protections, the flag, and the military, provides null results (see Supplementary Table A11) demonstrating that the incremental effect among nativists of exposure to ethnic out-group opposition to LGBT+ rights is unique to a sense of pride in Western liberal values. This provides evidence that nativists not only become more positively inclined toward pro-LGBT+ education as a result of negative affect against the ethnic out-group, but that their instrumental pro-LGBT+ positioning responds to a binary worldview in which sexual freedoms are associated with a “Western” nationalization (Lægaard Reference Lægaard2007) of pro-LGBT values that situates “the West” above the ethnic out-group.

Figure 8. CATE on Pride in Western Liberties

Note: Full regression output in Supplementary Table A11.