Abstract
The literature on Latino politics has emphasized the importance of immigration as a galvanizing issue for the community. Numerous studies have found that, in the aggregate, Latinos are attentive to immigration policy and oppose candidates who take restrictionist positions. However, it is unclear how Latinos make decisions when deeply-held issue positions apart from immigration conflict with group-relevant policies. In this paper, I design a dynamic tailored conjoint experiment that leverages large language models (LLMs) to assess whether immigration or core issues elicited in an open-ended question are stronger determinants of candidate choice. I consistently find that the effects of core issue positions on candidate choice are larger than the effects of immigration stances across three online samples of Latinos conducted on CloudResearch and YouGov (n=2,421). Factors such as group identity and proximity to the immigration experience narrow -- but rarely close -- the gap between the two issues.