New label, different identity? Three experiments on the uniqueness of Latinx

ABSTRACT

Groups use labels to define what communities stand for. Yet sometimes multiple labels refer to the same group (e.g., Hispanic, Latino). Do different labels generate distinct political opinions? Some work suggests that assorted labels evoke substantively similar views, since the attributes that define group members are highly correlated across categories. Other work, though, implies that varied labels can alter the configuration of group attributes in a way that elicits unique attitudes. We use these insights to evaluate Latinx: a new pan-ethnic label said to imply more gender-inclusive views. In three experiments, we randomly allocated Latino adults to report attributes that make them unique individuals (control) versus Latinx, Latino, or Hispanic. Assignment to the Latinx condition consistently increased participants’ support for pro-LGBTQ policies, an effect that was most precisely estimated in a meta-analysis of all three experiments. These results suggest that Latinx yields meaningful shifts in gender-inclusive opinions, consistent with claims about this label’s nature. We discuss our results’ implications for ongoing debates about Latinos’ self-designations.

KEYWORDS:

• Latinx
• Latino
• Hispanic
• identity
• experiments
• meta-analysis

Acknowledgements

We thank Tiffany Brannon, Jason Chin, Pete Fisher, Yuen Huo, David Sears and the rest of our colleagues in UCLA’s Intergroup Relations Lab (IRL) for their incredibly useful advice on this paper.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 Some scholars also suggest that Latinx is used by individuals to classify their own gender in non-binary fashion (Vidal-Ortíz and Martínez 2018). We do not empirically address this claim.

2 The US Department of Education defines HSIs as institutions where undergraduate enrollment is at least 25% Hispanic. We also considered running experiments that sampled on the basis of non-binary gender and sexual orientation, but our data-collection platforms did not have the capacity to yield large enough numbers of these individuals for statistical analysis.

3 Since randomization ensures participants in our conditions will be similar on all (un-)observed characteristics, we limit the number of measured pre-treatment covariates.

4 It is plausible that our Latinx condition induces categorization threat, which occurs when people feel an identity is mis-applied to them. If categorization threat is prevalent in our studies, our Latinx treatment should yield negative effects on our outcome. Our inspection of open-ended replies identifies only four Ps across our studies (N = 1204) that expressly rejected the Latinx label (e.g., “I do not consider myself Latinx”). Dropping them from our analyses leaves our inferences unchanged (see section A.5).

5 We designed these items tap a broad domain (pro-LGBTQ policy), while minimizing the degree of error that inheres in each single item. Our scale’s alpha is consistent with this goal, but future work should develop additional items to improve appraisals of opinion in this realm.

6 To gauge spillover effects beyond pro-LGBTQ policy, Ps in each study also expressed their feelings toward LGBTQs and ethnoracial groups. We detail these exploratory analyses in A.3.