However, a recent preprint by Leonardo Carella and Francesco Raffaelli (2026) shows that it is not only people's self-identification with certain social groups that influences their political attitudes, but that this pathway can also work the other way around, under conducive cues. Focusing on the case of immigration preferences, the authors show that people with anti-immigration attitudes become more likely to identify as Christian or working-class after being exposed to the narrative that these groups are under threat from immigration.
What was the experiment?
To investigate the attitude-identity spillover, Carella and Raffaelli ran two original survey experiments that estimate the effect of priming with exposure to group-based anti-immigration appeals and how this effect varies by pre-treatment immigration attitudes, using an interaction model. Both experiments were chosen to represent traditional social cleavages that are historically anchored and often linked to immigration in public debate, yet flexible enough to allow for variation in self-identification: class in Britain and Christianity in Italy.
In the first stage of the surveys, respondents were asked about their opinions on immigration as well as “objective” markers of the respective identity (i.e., social grade position and church attendance). In the second stage, respondents were randomly allocated to a treatment or a control group, in both of which they were exposed to real, unattributed quotes by national politicians: the control condition primed a threat to the working class (UK) or Christianity (Italy) without mentioning immigration; the treatment made that threat explicitly about immigration. Finally, both groups answered multiple identity measures and several class- or religion-linked policy items, like minimum wage increases or taxation of religious buildings.
How do identity spillovers work?
Carella and Raffaelli base their theory on the effect of ‘new issues’ like immigration on traditional cleavages, like class and religion, on three novel strands of research. Firstly, they draw on the argument that people match their identities to their politics, for example, in the case of Democrats being more likely to claim Latino identities. Secondly, the authors make use of existing research on the spillover from immigration attitudes onto other inter-group political issues, like LGBTQ+ rights, which shows that anti-immigration individuals are more likely to support the latter when exposed to the narrative that migrants are socially conservative on this matter. Lastly, previous research on group-based appeals has shown that narratives which position two groups as mutually exclusive (i.e., working-class Britons and immigrants) serve to redefine group boundaries.
Combining these approaches on cross-dimensional spillovers (Identity -> Identity, Attitudes -> Attitudes), the authors theorize that spillover from attitudes, i.e., anti-immigration, to identity, i.e., social class / religion, also occurs. Their argument is that Identification with a group in (apparent) opposition to immigrants is beneficial to those with anti-immigration views, as it justifies their attitudes. More specifically, they argue that the spillover is motivated by “instrumental belief justification”, increasing self-certainty and legitimizing their beliefs about immigration without changing their attitudes on other policy areas affecting the respective group.
What does the study find?
The authors’ hypotheses are confirmed by their findings: The treatment with anti-immigration group-appeals increased identification among anti-immigration respondents, in particular for those with few “objective” markers of group belonging. In the British case, Carella and Raffaelli find that the treatment worked symmetrically: pro-immigration respondents also disidentified from the working class after being exposed to the quotes. A potential explanation is that anti-prejudice norms among some British respondents might be sufficiently strong to cause some pro-immigration individuals to want to distance themselves from a social group presented as having “a good reason” to take anti-immigration stances.
While the Italian case also confirms the general expectations, i.e., a positive effect of the treatment on anti-immigration respondents without any “objective” markers of group belonging (church attendance in the past two weeks), the findings are not significant across the board. Significance holds for two of the three post-treatment identity measures, but the effect on “closeness to Christians” was only insignificantly positive. Interestingly, contrary to the British case, they also find a small, insignificant positive effect of exposure to religion-based anti-immigration statements on the Christian self-identification of pro-immigration individuals. This might be due to the fact that some pro-immigration respondents may nonetheless harbor some prejudice against Muslim immigrants (who are portrayed in opposition to native Christian in the experimental vignettes) specifically.
Further, the data support the authors’ assumption that the identity spillover is instrumental: they do not find evidence of increased support for policies in the respective groups' interests post-treatment. In other words, despite self-identifying with Christianity or the working-class, the respondents' opinions on, for example, same-sex marriage or minimum wage, remain unchanged. In sum, the study shows that identities are flexible in response to realistic elite cues and that immigration attitudes not only replace traditional social cleavages, but can change their meaning.
