CONCEPTUALIZATION, THEORY, AND HYPOTHESES

Why do Asian and Latino adults prioritize their racial or national identity more? We first clarify key terms in our theory: race, pan-ethnicity, national origin, and nationality. We define race as the socially constructed designation of the US population based on perceived physical characteristics, ancestry, and/or language (APA Dictionary of Psychology 2024), which we operationalize via the US Census categories, Asian American and Latino (US Census Bureau Reference Bureau2022a).Footnote ³ The Census deems Latino an ethnicity (US Census Bureau Reference Bureau2022b), which entails membership in a cultural group (APA Dictionary of Psychology 2024). But social scientists often treat Latino as a race, a convention we follow here. This equivalence of Asian Americans and Latinos is facilitated by the fact that both are pan-ethnic categories that encapsulate varied national origin groups (e.g., Asian Americans: Chinese, Filipinos, Indians, etc.; Latinos: Mexicans, Puerto Ricans, Cubans, etc.) (Espiritu Reference Espiritu1992; Mora Reference Mora2014). In turn, nationality is an attachment to one’s country of residence—here, the US.Footnote ⁴

Our theory focuses on the racialization and stratification of Asian and Latino individuals. Racialization is a structural process that stigmatizes these groups and subordinates them to whites (Omi and Winant Reference Omi and Howard1986). Stratification is the “sticky” social ordering of these racialized groups relative to whites (Kim Reference Kim2003; Masuoka and Junn Reference Masuoka and Junn2013). Although stratification produces stable social positions between whites and Asians and Latinos, racialization allows this field of relations to evolve—for example, when elites valorize Asians as “model minorities” relative to Black people (Kim Reference Kim2003; Reference Kim2023) or praise Latinos as “hardworking” in implied comparison to Black individuals (Fox Reference Fox2004). These discursive moves reposition, even if temporarily, Asians and Latinos away from Black people and closer to whites, which bolsters white supremacy.

Kim (Reference Kim2003) observes that minoritized groups (i.e., Asians) are often triangulated between whites and other PoC, which produces structural advantages (“Asians as model minorities”) and disadvantages (“Asians as perpetual foreigners”). Recent work innovates this view by illuminating richer variation in the social stations of minoritized groups (Chen and Hosam Reference Chen and Hosam2022; Kim Reference Kim2023; Masuoka and Junn Reference Masuoka and Junn2013). As Zou and Cheryan (Reference Zou and Cheryan2017) explain, PoC in the US are marginalized along two broad axes: how inferior–superior and how foreign–American are they considered?Footnote ⁵

As Figure 1 shows, whites are the most superior and American group. In turn, PoC are divided by their varied stations in this order. While Black people are sometimes considered a more American minority than Latinos and Asians (Carter Reference Carter2019), Black and Latino people are often stereotyped as inferior compared to Asians and whites. Further, although Asians are sometimes deemed superior to Blacks and Latinos, as indicated by the model minority myth (Kim Reference Kim2003), Asians and Latinos are both stereotyped as perpetual foreigners (Lacayo Reference Lacayo2017). This implies that Asians and Latinos are subordinated to Black people. However, this order also has an anti-Black pole on top of its pro-white pillar (Davies Reference Davies2022; Kim Reference Kim2023). Consequently, Asians and Latinos sometimes benefit from being foreign, but not Black (Chen and Hosam Reference Chen and Hosam2022; Kim Reference Kim2023; Pérez, Robertson, and Vicuña Reference Pérez, Robertson and Vicuña2023).

Figure 1. Two Axes of Subordination

Note: Adapted from Zou and Cheryan’s (Reference Zou and Cheryan2017) Racial Position Model (RPM).

The marginalization of Asians and Latinos as foreigners makes their pan-ethnic identities highly accessible because it regularly drives their exclusion (Pérez and Vicuña Reference Pérez, Vicuña, Huddy, Sears, Levy and Jerit2023). Since the 1990s, this marginalization has occurred in tandem with systemic surges in nativism (Abrajano and Hajnal Reference Abrajano and Hajnal2016), sexism (Frasure-Yokley Reference Frasure-Yokley2018), homophobia/transphobia (Michelson and Harrison Reference Michelson and Harrison2020), and Islamophobia (Lajevardi Reference Lajevardi2020). Against these trends, pan-ethnic identities provide Asians and Latinos with a sense of security and belonging as outsiders (Hogg, Hohman, and Rivera Reference Hogg, Hohman and Rivera2008), plus an outlook that enshrines the role of race and racism in their lives (Espiritu Reference Espiritu1992; Mora Reference Mora2014). For many Asians and Latinos, their pan-ethnic identity is a high priority because it is central to their lives (Ethier and Deaux Reference Ethier and Deaux1994). Although many Asians and Latinos prefer to define themselves by national origin (e.g., Chinese, Mexican), rather than pan-ethnicity (i.e., Asian, Latino) (Abrajano and Alvarez Reference Abrajano and Alvarez2010; Schmidt et al. Reference Schmidt, Alex-Assensoh, Aoki and Hero2010; Wong et al. Reference Wong, Ramakrishnan, Lee and Junn2011), we reason that in national politics, Asian and Latino are the relevant categories, with national origin groups cognitively subsumed under pan-ethnic categories (e.g., Asian = {Chinese, Indian…}; Latino = {Mexican, Cuban…}) (Pérez Reference Pérez2021). Since pan-ethnic and national origin identities are believed to converge and align with the Democratic party (Huddy, Mason, and Horwitz Reference Huddy, Mason and Horwitz2016), it is plausible that identity prioritization also extends to national origin attachment (relative to American identity), which we explore in Study 1.

Still, not all Asians and Latinos prioritize their pan-ethnic identities (Hickel et al. Reference Hickel, Alamillo, Oskooii and Collingwood2020). This can be partly explained by demographics within these groups, where rootedness in the US (e.g., nativity, immigrant generation, age) predisposes some individuals to value their American identity more (e.g., Abrajano and Alvarez Reference Abrajano and Alvarez2010; Wong et al. Reference Wong, Ramakrishnan, Lee and Junn2011). But there are also major systemic forces that encourage some Asians and Latinos to prioritize their American identity. As Sidanius and Pratto (Reference Sidanius and Pratto2001) observe, racial hierarchies are sustained by a consensus between the oppressors and some of the oppressed. This arises for two reasons. First, despite the benefits of identifying with a minoritized ingroup, it also poses risks, such as a chronic lack of social respect, which negatively affects some people’s well-being (e.g., low self-esteem) (Huo and Binning Reference Huo and Binning2008). Thus, some minoritized individuals look for paths out of this marginalization, such as identifying with a higher-status group like Americans. Footnote ⁶

For instance, Pérez, Robertson, and Vicuña (Reference Pérez, Robertson and Vicuña2023) establish that some Latinos are motivated to prove their worth as new Americans, leading them to express racism toward Black people despite their shared station as inferior groups (Figure 1). Indeed, inclusion in a higher-status group (American), no matter how tenuous, often drives individuals to sharpen the boundaries between “us” and “them” (Pickett and Brewer Reference Pickett, Brewer, Dominic, Michael A. and José M.2005). This implies that some Asians and Latinos will prioritize their American identity to counter their racial subordination, but at the expense of bolstering the larger system of oppression against non-Americans (Jost Reference Jost2019), which includes their pan-ethnic groups (Hickel et al. Reference Hickel, Alamillo, Oskooii and Collingwood2020). Although American identity provides these individuals with greater prestige and a sense of belonging, it also desensitizes them to the welfare of non-Americans, including their own pan-ethnic groups (Dovidio, Gaertner, and Saguy Reference Dovidio, Gaertner and Saguy2015). We therefore hypothesize that the racial stratification of Asians and Latinos produces significant variation in their prioritization of their racial or American identity (H1). Given the heterogeneity between Asians and Latinos in terms of their arrival to the US, their geographic spread, cultural nuances (e.g., common Latino language; no common Asian language), and their distinct treatment by US institutions (Schmidt et al. Reference Schmidt, Alex-Assensoh, Aoki and Hero2010), we expect some variation in the size of this relationship.

But how does this identity preference relate to Asians’ and Latinos’ partisan allegiance? Democrats and Republicans have become more racially and ideologically homogenous than before (Levendusky Reference Levendusky2009; Mason Reference Mason2018). This process yields two implications we exploit. First, polarization has yielded clearer stereotypes about who comprises each party and what it stands for (Ahler and Sood Reference Ahler and Sood2018; Westwood and Peterson Reference Westwood and Peterson2020). People have racialized images about Democrats and Republicans (Zhirkov and Valentino Reference Zhirkov and Valentino2022), with the former perceived as intertwined with PoC and the latter with whites, the “real” Americans (Danbold and Huo Reference Danbold and Huo2015). Alas, both whites and PoC associate the category, American, with whites (Devos and Banaji Reference Devos and Banaji2005).

These mental schemas indicate clear mental links between parties and racial and national symbols. Democrats privilege the identities of racially minoritized supporters, while Republicans enshrine American identity among their rank and file. Each party also has different visions about race in US society. Democrats widely recognize the continued influence of prejudice in US politics and broadly advocate for a racially inclusive society, while Republicans downplay racism’s prevalence and advocate for a polity that excludes sundry outgroups, including racial ones (Abrajano and Hajnal Reference Abrajano and Hajnal2016). This suggests two party alternatives for Asians and Latinos: Democrats who privilege racial liberalism or Republicans who espouse racial conservativism. Thus, we predict that Asians’ and Latinos’ partisan allegiance is influenced by how much weight they place on their racial or national identity (H2). Asians and Latinos who privilege their racial (national) identity will report stronger allegiance to Democrats (Republicans).

Finally, we explain when identity prioritization is activated. Many Asians and Latinos trend toward Democrats because their racial identity is central to them (Abrajano and Alvarez Reference Abrajano and Alvarez2010; Wong et al. Reference Wong, Ramakrishnan, Lee and Junn2011). Yet it is also true that many Asians and Latinos deem their American identity just as important, if not more (Silber Mohamed Reference Silber Mohamed2017). This means each party’s image as the home of PoC (Democrats) or Americans (Republicans) risks being misaligned with Asians’ and Latinos’ own identity priorities. This misalignment is called categorization threat (Branscombe et al. Reference Branscombe, Ellemers, Spears, Doosje, Ellemers, Spears and Doosje1999), and it reflects the cognitive discomfort one feels when classified into a group against one’s preferences. For example, Latinos who prioritize their American identity will experience categorization threat if they feel social pressure to align with Democrats, like most of their co-ethnics, prompting them to express Republican allegiance. Thus, categorization threat should activate one’s prioritized identity (racial, national), leading Asian and Latino adults to adjust their partisanship in line with it (H3).

Table 1 displays our hypotheses. Study 1 evaluates (H1-H2) with analyses of surveys on Asian and Latino adults since 2006, including a meta-analysis. Study 2 uses a pair of IATs to further assess (H2) by revealing the automaticity of identity prioritization. Studies 3–4 use two experiments testing for identity prioritization’s moderating influence on Asian and Latino partisanship (H3).

Table 1. Hypotheses About Identity Prioritization and Asian and Latino Partisanship

Studies 1 and 2: Identity Prioritization and Its Association with Asian and Latino Partisan Allegiance

To test H1 and H2, Study 1 assembled major publicly available population surveys of Asian American and Latino politics since 2006. Our efforts yielded N = 20,237. The inclusion criteria for this analysis were two-fold. First, surveys had to include large numbers of Asian and Latino adults to analyze (N ≥ 500 per group). Second, surveys had to contain items to operationalize our main variables, identity prioritization and partisan self-identification. This yielded (1) the 2006 Latino National Survey (LNS) (N = 8,634); (2) the 2008 National Asian American Survey (NAAS) (N = 5,159); (3) the 2012 American National Election Study (ANES) (N = 1,005 Latinos); (4) the 2016 Collaborative Multi-Racial Post-Election Survey (CMPS) (N = 3,055 Asians and N = 3,002 Latinos); and (5) the 2020 Collaborative Multi-Racial Post-Election Survey (N = 3,956 Asians and N = 4,585 Latinos).

Our preregistered analyses of these surveys involved linearly predicting expressed allegiance to the Democratic (Republican) party based on one’s identity prioritization, holding constant differences in perceived discrimination, ideology, church attendance, education, nativity, gender, and age (Supplementary SI.1). We also undertake a preregistered meta-analysis of these surveys to estimate a common trend among them (Supplementary SI.1). We measured partisanship with a 7-point scale arranging individuals in terms of strength of identification as Democrats (Republicans), with those indicating “independence” arrayed as partisan leaners upon further prompting (Supplementary SI.2). Substantial shares of respondents indicated no partisan affiliation in the 2006 LNS. They are excluded from our analyses because their replies were not further probed for partisan leanings, leaving us with N = 5,802. Throughout, we rescale partisanship to a 0–1 range, where higher values indicate greater Democratic self-identification.

Our key predictor is identity prioritization. Except for the 2008 NAAS, we measured it by differencing pairs of pan-ethnic and American identity items (Supplementary SI.2).Footnote ⁷ For example, in the 2012 ANES, Latinos were asked, separately, “How important is being [an American/Hispanic] to you personally?”, with replies arrayed from 1 (extremely important) to 5 (not at all important). We recoded each variable so that higher values indicate more importance. We then subtracted one’s importance score for American identity from one’s importance score for racial identity. Thus, Identity prioritization = Racial identity – American identity. Positive scores reflect prioritization of one’s racial identity as Asian or Latino, while negative values indicate prioritization of one’s American identity. We rescale this variable to a 0–1 range. (Supplementary SI.3) describes the coding of our covariates, also on a 0–1 range. Our coefficients reflect percentage-point shifts. In all our studies, the reported p-values are two-tailed. We advise readers to interpret these analyses as correlational and suggestive of our proposed mechanism, rather than causal and strictly unidirectional.

STUDY 1 RESULTS

Table 2 displays mean identity prioritization scores in each survey (in raw form), along with their standard deviations. Across all samples, reliable variation exists in the degree of identity prioritization among Asians and Latinos. Here, a value of zero indicates no prioritization of either pan-ethnic or national identity. We generally observe a modest preference for one identity over the other, with variation around each mean. Indeed, the average prioritization of pan-ethnic identity over American identity varies by year, survey, and sampled population. This aligns with the situational salience of identities (Turner et al. Reference Turner, Hogg, Oakes, Reicher and Wetherell1987), while highlighting possible historical, cultural, and demographic nuances between Asians and Latinos (Schmidt et al. Reference Schmidt, Alex-Assensoh, Aoki and Hero2010).

Table 2. Significant Heterogeneity in Asian and Latino Identity Prioritization

Note: Positive values in identity prioritization indicate greater weight on one’s racial versus American identity. All significance tests are two-tailed.

These patterns align with prior analyses of identity prioritization among Latinos (Hickel et al. Reference Hickel, Alamillo, Oskooii and Collingwood2020), but here, we broaden this evidence to multiple survey samples, time periods, and populations (i.e., Asian Americans). The last column in Table 2 formally supports H1. The t-tests indicate we can confidently reject the null that no systematic differences in identity prioritization exist among Asian and Latino adults. Consistent with our theoretical discussion, older, US-born Asians and Latinos are less likely to prioritize their pan-ethnic identity over their American identity (see results in Supplementary SI.4).

Our next hypothesis (H2) is that these individual differences in identity prioritization are meaningfully associated with Asian and Latino adults’ expressed allegiance to Democrats (Republicans). Table 3 supports this claim, with complete results in Supplementary SI.5. Across samples, we find a consistently positive, substantively meaningful, and generally reliable association between identity prioritization and partisan allegiance among Asian American and Latino adults. Supplementary SI.6 reports provisional results showing similar patterns when identity prioritization is measured with items on national origin identity (e.g., Mexican) rather than pan-ethnic identity (i.e., Latino).

Table 3. Prioritization of Pan-Ethnic ID (over American ID) Predicts Democratic Partisanship Among Asians and Latinos

Note: **p < 0.010 or better, two-tailed. All variables run along a 0–1 interval. Full results in Supplementary SI.5.

Table 3 also reveals some hints that the associations between identity prioritization and partisan allegiance have increased over the period under analysis and that the observed associations are generally stronger for Latinos than Asian Americans. The latter might stem from population differences (e.g., more foreign-born Asian Americans than Latinos) and/or contextual variation (e.g., more consistent partisan outreach during recent presidential elections among Latinos, than Asians) (Schmidt et al. Reference Schmidt, Alex-Assensoh, Aoki and Hero2010).

Our analyses reveal associations ranging from a low of about 7 percentage points to a high of 25 percentage points. These patterns are independent of our covariates, which means that having knowledge of Asian and Latino adults’ identity prioritization helps to reliably predict the strength of partisan allegiance among them.Footnote ⁸

Table 3’s last two rows report diagnostics from sensitivity analyses for observational analyses like these (Cinelli and Hazlett Reference Cinelli and Hazlett2020). Each regression’s robustness value (RV) indicates the percentage of residual variance of identity prioritization and partisanship that would bring our estimated effect to zero. In turn, the robustness value for the t-value of each point estimate (RV_{α = 0.05}) indicates the minimum strength of association (in terms of partial R²) that would bring our point estimates into a range where they are no longer statistically distinguishable from zero. Looking at the leftmost entry for Asians in Table 3 (row RV), we see that an unobserved confounder would need to explain at least 4.99% of the residual variance of our identity prioritization predictor and our partisanship outcome to reduce that estimated effect to zero. The entry below that (under RV_{α = 0.05}) indicates that unobserved confounders would need to explain at least 1.77% of the residual variance in identity prioritization and partisanship for us to be unable to reject the null (at p < 0.05) that the effect of identity prioritization is equal to zero. These diagnostics suggest the estimated effects of identity prioritization are reasonably robust to confounding. Nevertheless, they do not provide evidence of a causal and/or a unidirectional relationship in these cross-sectional data.

Given the variability around identity prioritization’s effects, we undertake a preregistered meta-analysis, which estimates the average association, net of differences in survey samples under analysis (Goh, Hall, and Rosenthal Reference Goh, Hall and Rosenthal2016) (https://osf.io/6tfh5/; Supplementary SI.1). This approach utilizes a precision-weighted estimate of the association between identity prioritization and expressed partisanship across survey samples. Figure 2 displays these associations, with a meta-analyzed estimate of 0.149, SE =0.008, p < 0.001. We convert this into a d-value, which is in standard deviation units. Our meta-analyzed d-value of 0.29 suggests that a unit increase in the prioritization of one’s racial identity (versus American identity) heightens one’s expressed allegiance to the Democratic party by about one-third of a standard deviation, which we interpret as substantial, since partisan identities are characterized as being difficult (though not impossible) to shift, given their stability (Egan Reference Egan2020; Hopkins et al. Reference Hopkins, Kaiser, Pérez, Hagá, Ramos and Zárate2020). These layers of evidence support (H2): identity prioritization is meaningfully associated with Asian and Latino partisanship.

Figure 2. Percentage Point Increase in Democratic Allegiance by Greater Prioritization Racial ID Over American ID (95% Confidence Bands)

Still, our analyses here rest on self-reported data, which means the correspondence between identity prioritization and partisan allegiance might be affected by social desirability pressures to express one’s allegiance with a certain party (e.g., “since most Latinos are Democrats, then I, as a Latino, should also say I’m a Democrat”). Study 2 throws light on this possibility by digging deeper into the automaticity of identity prioritization among Asians and Latinos.

STUDY 2: IMPLICIT ASSOCIATION TESTS (IATs)—PAN-ETHNICS|DEMOCRATS VERSUS AMERICANS|REPUBLICANS

We further evaluate (H2) with a pair of IATs that use reaction times (in milliseconds) to gauge how automatically people associate pairs of categories in memory (Theodoridis Reference Theodoridis2017) (preregistration in Supplementary SI.7). A virtue of IATs is that they are less prone to social desirability bias than self-reports since one’s ability to control responses is short-circuited through this rapid, timed sorting task (Pérez Reference Pérez2013).

We conducted our IATs among Asian and Latino undergraduates in a laboratory at a major public university on the West Coast during May 2023. Participants were recruited from political science classes and provided with two extra credit points in their course in exchange for their participation. After consenting to participate via an information sheet emailed to them, participants chose a convenient time to visit our lab to complete an IAT. These IATs estimate the degree to which Asians (N = 39) automatically associate (1) members of their pan-ethnic group (e.g., Chinese, Filipinos, Koreans, etc.) with Democrats or Republicans; and (2) Americans with Democrats and Republicans (Supplementary SI.7 contains stimuli). The same logic was adapted to Latinos (N = 32) (e.g., Mexicans, Puerto Ricans, etc.). While sample sizes for these IATs are small, IAT effects are often large (d > 0.80), so they require fewer observations to detect (Pérez Reference Pérez2013). Even with our sample sizes, our power to detect a large and reliable effect (d > 0.80) is 0.83.Footnote ⁹

Since the order of the IATs’ classification pairs used by participants are randomly assigned, each IAT is a within-subjects experiment, with the estimated quantity being a D-value ranging from −2 to 2. Positive values indicate a stronger association between the hypothesized classification pair (i.e., Latinos|Democrats versus Americans|Republicans) in comparison to its opposite pair (i.e., Americans|Democrats versus Latinos|Republicans). In our IATs, participants sat behind laptop computers and sorted stimuli reflecting each category in these classification pairs (e.g., Mexican flag, Democrat logo, Republican logo, Chinese flag, American flag). Using one classification pair, each participant made 40 classifications of these stimuli. They repeated the same sorting exercise another 40 times with the opposite classification pair. For Asians, a positive D-value reflects the degree to which Asians and Democrats, and Americans and Republicans, are associated in their memory. For Latinos, a positive D-value indicates the extent to which Latinos and Democrats, and Americans and Republicans, are associated in their memory.

Figure 3 depicts our D-values, which are positive and range from moderately strong (Asians D: 0.164, SD = 0.459) to very strong (Latinos D: 0.472, SD = 0.317). If we divide each D-score by its standard deviation, we yield a Cohen’s d, or standardized mean difference. Cohen’s d values around 0.20, 0.50, or 0.80 or greater are considered small, medium, and large, respectively (Asian Americans d = 0.357; Latinos d = 1.489). The stronger IAT effect for Latinos is reliably different at the 1% level (M_Latinos - M_Asians = −0.307, SE = 0.096; t = −3.210, df = 69, p < 0.002). These results suggest the associations between Democrats and Asians and Latinos—and between Republicans and Americans—are automatic for these individuals, thus extending Study 1’s observational analyses.

Figure 3. D-values from Implicit Association Tests with Asian and Latino Participants

Note: For Asians, a positive D-value reflects the degree to which Asians and Democrats, and Americans and Republicans, are associated in memory. For Latinos, a positive D-value indicates the extent to which Latinos and Democrats, and Americans and Republicans, are associated in memory.

While the evidence across Studies 1–2 supports our first two hypotheses, what remains unclear are the mechanisms behind this pattern—that is, under what conditions does identity prioritization matter for Asian and Latino partisanship? Answering this question is partly complicated by some evidence suggesting that partisan identities can influence racial identities (Agadjanian and Lacy Reference Agadjanian and Lacy2021; Hopkins, Kaiser, and Pérez Reference Hopkins, Kaiser and Pérez2022). Thus, to validate this aspect of our thinking, we undertook a pair of experiments, which allowed us to control the causal order between identity prioritization, partisan allegiance, and our proposed mechanism connecting both.

STUDY 3: CATEGORIZATION THREAT AND IDENTITY PRIORITIZATION AMONG LATINOS

Per our categorization threat hypothesis (H3), the internal diversity of Asian Americans and Latinos lends itself to depicting them as principally concerned about their status as pan-ethnics or as Americans (Branscombe et al. Reference Branscombe, Ellemers, Spears, Doosje, Ellemers, Spears and Doosje1999). That is, members of these pan-ethnic groups possess, to varying degrees, attributes reflecting their nominal membership in two major categories with high partisan relevance: Asian/Latino and American. This implies that characterizations of each pan-ethnic population’s central tendency and variance are somewhat malleable (Turner et al. Reference Turner, Hogg, Oakes, Reicher and Wetherell1987). Thus, Asians and Latinos should be sensitive to whether they are depicted as a primarily pan-ethnic or American group. This raises the possibility that their identity priorities will be misaligned with their ingroup’s characterization. If this reasoning is valid, then depicting Asians and Latinos as generally pan-ethnic (American) communities should activate their identity prioritization (H3), with identity prioritization moderating the impact that being categorized as Asian/Latino or American has on partisanship.

Our initial test of (H3) uses a preregistered experiment with Latino adults (N = 1,572) (Supplementary SI.8). Study 3 occurred online on Dynata’s survey platform, which compensates participants via an internal reward system akin to frequent flier miles. Latino adults from Dynata’s respondent panel were invited via email to participate in our study. After providing informed consent, Study 3 collected demographic data and measured our moderator, identity prioritization. In the middle of this pretreatment module, participants completed identity centrality items gauging the importance of four identities for Latinos. In randomized order, these were one’s identity as American, as Latino, as a family member, and as a professional (Leach et al. Reference Leach, Van Zomeren, Zebel, Michael, Pennekamp, Doosje and Spears2008). Here, American and Latino identities were gauged with two items each, answered on a 7-point scale, with a neutral option: (1) “Being [American/Latino] is personally important to me;” and (2) My [American/Latino] identity is central to who I am.” We use these pairs of items to gauge each of these identities to increase our measure’s reliability (α_American = 0.833, α_Latino = 0.840). We build our identity prioritization measure by taking participants’ average Latino identity centrality and subtracting from it participants’ average American identity centrality. This produces an identity prioritization measure ranging from +6 (prioritize Latino identity) to −6 (prioritize American identity), where higher values indicate that Latinos prioritize their pan-ethnic identity more than their American identity. We leave this variable in its raw metric (M = 0.328, SD = 1.696).

Following an attention check, participants were randomly assigned to a control group or one of two treatments. All three conditions exposed participants to a mock news article from the Associated Press, which presented an alternative perspective to another point of view on a matter. In the control, participants read a news brief about how, contra the views of some people, the number of giant tortoises around the globe is declining. In turn, our American categorization treatment discussed how, despite being characterized as a largely pan-ethnic population, many Latinos place greater weight on their American identity. In contrast, our Latino identity treatment explained that, despite their similarity to Americans in terms of some traits (e.g., optimism, hardworking), many Latinos prioritize their pan-ethnic identity.Footnote ¹⁰

Post-treatment, participants answered a conventional 7-point scale of partisan identification. This measure probes for partisan leaners, who are ultimately included along this spectrum; and it aligns with the same outcome analyzed in our observational analyses. We leave this partisanship measure, Democratic allegiance, in its raw metric and code it so that higher values indicate greater reported attachment to the Democratic party (i.e., 1 to 7 range). Below, we report fully interactive models where the impact of identity prioritization on partisan allegiance is moderated by exposure to our Latino and American categorization treatments. Footnote ¹¹

STUDY 3 RESULTS

The statistical key to (H3) is a set of reliable interactions indicating that the impact of identity prioritization on Latinos’ Democratic allegiance is moderated by our Latino categorization and American categorization treatments. Given Equation 1 below, we expect [β₄] and [β₅] to be reliably different from zero, which would suggest that the relationship between partisanship and identity prioritization is dependent on each of our treatments.

(1) $$ {\mathrm{Y}}_{\mathrm{Democrat}}={\displaystyle \begin{array}{l}{\unicode{x03B2}}_{1,\mathrm{Latino}\ \mathrm{categorization}}+{\unicode{x03B2}}_{2,\mathrm{American}\ \mathrm{categorization}}\\ {}+\hskip2px {\unicode{x03B2}}_{3,\mathrm{ID}\;\mathrm{prioritization}}+{\unicode{x03B2}}_{4,\mathrm{Latino}\;\mathrm{x}\;\mathrm{prioritization}}\\ {}+\hskip2px {\unicode{x03B2}}_{5,\mathrm{American}\;\mathrm{x}\;\mathrm{prioritization}}+\unicode{x025B} \end{array}} $$

Table 4 provides raw evidence supporting (H3). First, notice that in our control group, participants who prioritize their Latino identity are more likely to express Democratic allegiance by about one-third of a point on our 7-point scale of partisanship (0.302, SE = 0.045, p < 0.001), which is a substantively meaningful effect. Second, both interaction terms are highly reliable, suggesting the influence of identity prioritization on expressions of Democratic allegiance depends on how Latinos are categorized.

Table 4. Categorization as Latino or American Activates Identity Prioritization Among Latinos

Note: Entries are Ordinary Least Squares (OLS) coefficients with robust standard errors in parentheses. ***p < 0.001, **p < 0.050.

To better interpret the interactive effects from our model, we depict the change in pro-Democrat allegiance in light of a ± 1 SD in Latinos’ identity prioritization when exposed to Latino categorization or American categorization. Given identity prioritization’s coding, a positive SD indicates a shift toward prioritizing one’s Latino identity, whereas a negative SD decrease reflects a shift toward prioritizing one’s American identity. The depicted quantities are raw estimates drawn from these two subsets of data.

Panel A in Figure 4 reveals that when Latinos are characterized as a largely ethnic population, those who prioritize their Latino identity report stronger allegiance to Democrats (0.674, SE = 0.158, p < 0.001), while those who prioritize their American identity express reliably weaker allegiance to Democrats (−0.351, SE = 0.125, p < 0.005). The difference between both marginal effects is highly significant (F_{(1, 1566)} = 44.26, Prob > F = 0.000).

Figure 4. Latinos’ Categorization as Pan-Ethnics Activates Their Identity Prioritization

Note: The marginal effects depicted here are from the model reported in Table 4.

Substantively, this pattern implies that when Latinos are characterized as a largely pan-ethnic population, those who prioritize their Latino identity express greater allegiance to Democrats by two-thirds of a point (0.674) on our partisanship 7-point scale. However, the same treatment propels Latinos who prioritize their American identity to express lower allegiance to Democrats by about one-third (−0.351) of a point on the same partisanship scale. This trend suggests that Latinos’ reaction to their group’s depiction as pan-ethnics affirms allegiance to Democrats among those who prioritize their Latino identity but undermines allegiance to the same party among those who privilege their American identity. This finding suggests that public characterizations of their population as pan-ethnics stand to bring some Latinos closer to the Democratic party, but at the expense of losing some adherents—all because of how (mis)matched their prioritized identity is to public characterizations of their pan-ethnic population.

Panel B further shows that when Latinos are categorized as Americans, those who prioritize their ethnic identity express significantly weaker allegiance to Democrats, a decrease of nearly half a point on our partisanship scale (−0.459, SE = 0.149, p < 0.002). However, among those who prioritize their American identity, the categorization of Latinos as new Americans propels them to express greater allegiance to Democrats, an effect of about two-thirds of a point on our partisanship scale (0.625, SE = 0.192, p < 0.001). This shift in partisan allegiance is statistically reliable (F_{(1, 1566)} = 20.61, Prob > F = 0.000), indicating that categorizing Latinos as Americans increases Democratic allegiance among those who prioritize their American identity, but decreases it among those who prioritize their pan-ethnic identity. Our next study evaluates whether these patterns also characterize identity prioritization’s influence on Asians’ partisanship.

STUDY 4: CATEGORIZATION THREAT AND IDENTITY PRIORITIZATION AMONG ASIAN ADULTS

Study 4 is a preregistered experiment with Asian adults aiming to conceptually replicate Study 3’s findings (S.9). Study 4 followed the same structure as Study 3 and also occurred online via Dynata’s survey platform. Asian adults from Dynata’s respondent panel were invited via email to participate in our study. After providing their informed consent, we collected demographic data and our measure of identity prioritization, which captures the degree to which Asian participants place greater importance on their pan-ethnic versus American identity. The primary difference between Studies 3 and 4 arises from the specific content of our manipulations, given that Asian Americans are considered a foreign, but superior minoritized outgroup and Latinos are considered a foreign and inferior outgroup (Figure 1).

After collecting demographic data, our measures of identity prioritization (α_American = 0.843, α_Asian = 0.879), and an attention check, participants were randomly assigned to a control group or one of two treatments. All three conditions, again, exposed participants to a mock news article (attributed to the Associated Press) that presented an alternative perspective to another point of view on a matter. In the control, participants read a news brief about how, contra the views of some people, the number of giant tortoises around the globe is declining. In turn, our American categorization treatment discussed how, despite being characterized as a largely pan-ethnic population, many Asian individuals place greater weight on their American identity, as evidenced by wealth-induced increases in consumer purchases (which aligns with the construal of Asian individuals as model minorities, relative to Latino and Black people) (Kim Reference Kim2003). In contrast, our Asian identity treatment explained that, despite their similarity to Americans in terms of some traits (e.g., optimism, hardworking), most Asians prioritize their racial identity, as evidenced by increasing hate crimes against this population.

Post-treatment, participants answered a conventional 7-point scale of partisan identification. We again leave this partisanship measure, Democratic allegiance, in its raw metric and code it so that higher values indicate greater reported attachment to the Democratic party (1–7 range). Below, we report fully interactive models where the impact of identity prioritization on partisan allegiance is moderated by exposure to our racial and American categorization treatments.

STUDY 4 RESULTS

Similar to Study 3, we are looking for evidence that identity prioritization is activated in light of Asians’ categorization as a primarily pan-ethnic or American population. This entails reliable and substantively meaningful interactions between identity prioritization and assignment to each of our manipulations. Table 5 provides our raw results, with some evidence reaffirming (H3). In our control group, participants who prioritize their Asian identity over their American identity are more likely to express Democratic allegiance by about one-fourth of a point on our 7-point scale of partisanship (0.267, SE = 0.051, p < 0.001), which is a substantively meaningful effect that is comparable to the one observed among Latinos in Study 3. Given that our constant (4.582) indicates a pro-Democrat outlook, our average effect here suggests an intensification of this allegiance based on one’s identity prioritization.

Table 5. Categorization as Asian or American Activates Identity Prioritization Among Asians

Note: Entries are OLS coefficients with robust standard errors in parentheses. ***p < .001, **p < .050.

What happens when we examine the activation of identity prioritization in light of Asian adults’ categorization as a racial community? Very little, actually. In both substantive and statistical terms, the interaction between identity prioritization and Asian categorization is essentially zero (−0.021, SE = 0.072, p < 0.769). Panel A in Figure 5 confirms this null pattern, where the difference between both depicted marginal effects is, statistically, zero (β_{1, racial categorization} + β_{4, racial x prioritize Asian ID} = −0.118, SE = 0.118, p < 0.321 versus β_{1, racial categorization} + β_{5, racial x prioritize American ID =} − 0.061, SE = 0.176, p < 0.728).

Figure 5. Asians’ Categorization as Asians Activates Their Identity Prioritization

Note: The marginal effects depicted here are from the model reported in Table 5.

Interestingly, however, when Asian individuals are categorized as new Americans, their identity prioritization levels behave similarly to what we observed among Latinos under the same type of categorization in Study 3 (Figure 4, panel B). The interaction between identity prioritization and our American categorization treatment is highly reliable and meaningful (−0.215, SE = 0.080, p < 0.001). Panel B in Figure 5 depicts this significant pattern. When Asian adults who prioritize their Asian identity are categorized as a primarily American population, they express weaker allegiance to Democrats (−0.179, SE = 0.121, p < 0.139). In turn, when Asian adults who prioritize their American identity are categorized as Americans, they report reliably stronger allegiance to the Democratic party (0.391, SE = 0.188, p < 0.038), with this difference in marginal effects being significantly different from zero (F_{(1, 1662)} = 7.30, Prob > F = 0.007). Similar to Latinos, then, Asians who value their American identity are more inclined toward the Democratic party when their ingroup is characterized as being composed of new Americans—an effect that comes at the expense of weaker Democratic allegiance among those Asian individuals who prioritize the pan-ethnic identity.

MINI META-ANALYSIS OF STUDIES 3–4

Here, we formally probe for any summary trends across our experiments by undertaking another pre-registered mini meta-analysis, using Goh, Hall, and Rosenthal’s (Reference Goh, Hall and Rosenthal2016) template (Supplementary SI.9). Table 6 reports a model based on a dataset that pools Studies 3–4. This model estimates the same interactive relationship in Tables 4 and 5, but tests whether meaningful trends emerge across our pooled sample.

Table 6. Does Racial (American) Categorization Activate Identity Prioritization Across Latinos and Asians?

Note: N = 3,240. Entries are GLS coefficients from a random effects model with robust standard errors in parentheses. ***p < 0.001, **p < 0.050.

Readers will recall that Latinos’ identity prioritization was activated in the wake of their ingroup’s racial categorization, but the same pattern was absent among Asians under this same circumstance (see Figure 4A versus Figure 5A). Consistent with that mixed pattern, our meta-analysis here shows that the impact of identity prioritization on Democratic allegiance among Latino and Asian adults (0.290, SE = 0.017, p < 0.001) is statistically similar across our control group and racial categorization condition (−0.161, SE = 0.109, ns). We cannot reject the null hypothesis that the effect of identity prioritization on Democratic allegiance in the control group versus the racial categorization is zero (Χ²₍₁₎ = 2.19, Prob > Χ² = 0.139). This means identity prioritization’s effects in light of racial categorization are specific to Latinos and do not apply to Asians in our experiments.

In contrast, the reader will recall that in light of American categorization, we observed parallel patterns in identity prioritization among Latinos and Asians (see Figures 4B and 5B). In the wake of American categorization, Latinos and Asians who prioritize their pan-ethnic identity expressed weaker allegiance to Democrats, while those who prioritize their American identity reported significantly stronger allegiance to Democrats. That pattern is borne out by our meta-analysis: the effect of identity prioritization on Latinos’ and Asians’ Democratic allegiance is reliable in the control group (0.290, SE = 0.017, p < 0.001) and the American categorization condition (−0.277, SE = 0.050, p < 0.001), with this difference in effects being significantly different from zero (Χ²₍₁₎ = 30.52, Prob > Χ² = 0.001). This summary pattern is depicted in Figure 6, indicating that exposure to American categorization steers Latino and Asian adults who prioritize their pan-ethnic identity away from Democrats, but propels those who privilege their American identity toward Democrats.

Figure 6. Mini Meta-Analysis: Identity Prioritization Activated in Light of American Categorization Among Latino and Asian Adults

Note: The marginal effects depicted here are from the model reported in Table 6.

CONCLUSION AND IMPLICATIONS

We theorized and found that Asian and Latino adults vary significantly in their identity prioritization (H1). We also hypothesized and discovered that variation in identity prioritization substantively predicts the intensity of Asians’ and Latinos’ partisan allegiance, with some evidence underlining its automatic nature (H2). Finally, we clarified when identity prioritization should affect Asian and Latino partisanship. Using two experiments that induced categorization threat, we found that Asians and (especially) Latinos react to this sense of jeopardy by reaffirming their allegiance to Democrats or Republicans.

We think a major implication of our work is the underappreciated sensitivity that members of these pan-ethnic populations have to how their respective racial ingroups are characterized. It is a truism among social scientists that Asians and Latinos are internally diverse populations. And while that assertion is, on the surface, correct, less understood is what this internal diversity implies for the political incorporation of these groups. Our findings underline one way this internal heterogeneity can matter politically—the partisan allegiance that Asian and Latino individuals feel is shaped by whether a party reflects the importance these individuals attach to their American or pan-ethnic identity. If we make the reasonable assumption that most Asian and Latino individuals trend toward the Democratic party (Abrajano and Alvarez Reference Abrajano and Alvarez2010; Wong et al. Reference Wong, Ramakrishnan, Lee and Junn2011), our evidence implies that this steady drift toward the major parties can be slowed or even reversed depending on whether Asian and Latino individuals believe their more prioritized identity is overlooked. Of special relevance here is the centrality of American identity among significant shares of Asian and Latino individuals. When these people believe their prioritized identity is overlooked, they become less enthusiastic supporters of the Democratic party, the “natural” home of their respective pan-ethnic population. Against a backdrop of measurable gains by Republicans among segments of these communities (Fraga, Velez, and West Reference Fraga, R. Velez and West2024), our evidence points to the role that identity prioritization might play in these evolving dynamics.

Our results also suggest the value of sifting through available models of partisanship formation in order to make new connections. Various scholars of race and politics have correctly noted that parental socialization cannot fully explain partisanship among populations with deep immigrant roots. Yet we think that in dismissing this element of that model, Race, Ethnicity, and Politics (REP) scholars are also, incorrectly, too quick to dispense with all elements in this framework. In particular, the role of social groups, parties, and the mental associations between them offer a potential link to how some PoC and white individuals learn to be Democrats or Republicans. As Green, Palmquist, and Schickler (Reference Green, Palmquist and Schickler2002) and others have taught us (Zhirkov and Valentino Reference Zhirkov and Valentino2022; Westwood and Peterson Reference Westwood and Peterson2020), individuals are attracted to specific parties based on various social identities that are “owned” by each party. The stability of partisanship, in turn, stems from the strength and consistency of these associations over time. Our studies indicate that many PoC adhere to this cognitive process, too, where they associate their pan-ethnic group with Democrats and Americans with Republicans. Given the robust sense of partisanship among PoC and the rich variation in it, our research also suggests these trends are steady and long enough for Asians and Latinos to have sorted themselves comfortably into respective Democratic and Republican camps. As our nation continues to racially diversify, political scientists will need to wrestle further with whether standing models of political behavior should be discarded in favor of new frameworks that reflect the reality of specific communities of color or whether revised, but more generic models can effectively accommodate the broad realities of a multiracial mass public.

One fruitful direction here is to consider the extent to which identity prioritization is also influenced by party competition and other dynamics. Consider Margolis (Reference Margolis2018), Egan (Reference Egan2020), and others (Agadjanian and Lacy Reference Agadjanian and Lacy2021; Hopkins, Kaiser, and Pérez Reference Hopkins, Kaiser and Pérez2022), who all demonstrate that under some conditions, partisan identities can shape the strength and direction of one’s racial or ethnic identity. This aligns with social psychological research establishing the role of identity as both a cause and an effect of intergroup dynamics (Doosje, Spears, and Ellemers Reference Doosje, Spears and Ellemers2002). Consequently, while our evidence supports the viability of identity prioritization as an influence on partisan allegiance, more work is needed on evaluating the path of influence from partisanship to identity prioritization.

Another promising direction involves the tension between partisanship’s stability among Asian Americans and Latinos over time (Hopkins, Kaiser, and Pérez Reference Hopkins, Kaiser and Pérez2022; McCann and Jones-Correa Reference McCann and Jones-Correa2020) and the shifts in partisan identity observed here. Some of this disjuncture is simply due to levels of analysis, where differences between individuals can coexist with aggregate stability, akin to models of macro- and micro-level politics (Erikson, Mackuen, and Stimson Reference Erikson, Mackuen and Stimson2002). Yet some of this gap is likely theoretically meaningful, which should encourage scholars to explore when and why individual-level shifts in partisan identity among Asians and Latinos translate into more durable, longer-run shifts in the distribution of partisanship in these publics.

While our observational and experimental research designs generally supported our hypotheses, this evidence places a stronger emphasis on average correlations or effects. This evidence should not be construed as implying that there are no meaningful differences detected between Asian Americans and Latinos, which are to be expected given the heterogeneous composition of each population (Schmidt et al. Reference Schmidt, Alex-Assensoh, Aoki and Hero2010) and the varying efforts of Democrats and Republicans in mobilizing these voters during presidential elections (Abrajano and Alvarez Reference Abrajano and Alvarez2010; Chan, Kim, and Leung Reference Chan, Kim and Leung2022; Fraga Reference Fraga2018; Fraga, Velez, and West Reference Fraga, R. Velez and West2024; Kuo, Malhotra, and Mo Reference Kuo, Malhotra and Mo2017; Wong et al. Reference Wong, Ramakrishnan, Lee and Junn2011). Indeed, in some instances, we find that some of our results are relatively stronger for Latinos than Asian Americans, which raises interesting questions on their own. Although outside of the scope of this paper, we think that, theoretically, these nuances reflect, to some degree, variation in the intensity and consistency of mobilization efforts aimed at these populations by partisan and non-partisan operatives, as well as the relatively unique historical experiences of these two groups in the US (Fraga Reference Fraga2018; Fraga, Velez, and West Reference Fraga, R. Velez and West2024; Ramírez Reference Ramírez2015; Schmidt et al. Reference Schmidt, Alex-Assensoh, Aoki and Hero2010; Wong Reference Wong2005). We encourage researchers to further leverage this heterogeneity to yield more insight into partisanship’s origins among Asians and Latinos.

Finally, we call attention to what our findings imply about the heterogeneity of partisanship among Asian Americans and Latinos. A panoramic view of partisan identity within these two pan-ethnic populations indicates a robust central tendency in favor of Democrats, with 66%–75% expressing allegiance to this party and acting on it during elections. But this consistent focus on averages glosses over the nontrivial share of Asian and Latino individuals who identify, enthusiastically, with the Republican party. For those scholars who may wonder why the share of Asian and Latino Republicans is not lower, our findings point to identity prioritization’s role. It suggests that the strong allegiance that some Asians and Latinos express toward the Grand Old Party (GOP) is less likely a function of false consciousness, lack of knowledge, or misinformation, and more likely rooted (at least partly) in the fact that they perceive Republicans as the guardians and stewards of Americans—a social identity that substantial shares of Asians and Latinos deem central to them (Silber Mohamed Reference Silber Mohamed2017). Without further attention to this dynamic, scholars risk being surprised, repeatedly, about the lack of enthusiasm for Democrats in some elections among these populations or by the commitment of some Asians and Latinos to a Republican party that is outright hostile to racial diversity. Therefore, we invite political scientists to continue probing the roots and nature of partisan identity among members of these two critical pan-ethnic populations.