Do Immigrants’ Partisan Preferences Influence Americans’ Support for Immigration?
Do Immigrants’ Partisan Preferences Influence Americans’ Support for Immigration? (DOI: https://doi.org/10.1017/XPS.2025.10013) was written by Daniel McDowell and David A. Steinberg in 2025. It was published in the Journal of Experimental Political Science.
Web survey of the U.S. general population, with random treatment by information about an immigrant group's partisan leaning.
- 3000 respondents from Dynata's panel
- balanced through quotas on gender, age, and education
- fielded Oct 24-Nov 1, 2024
- random treatment is built directly into question for key metric
- 11-point scale for agreement with permitting higher levels of immigration
"In the last ten years, a significant number of immigrants into the United States have come from [COUNTRY]. A majority of immigrants from [Vietnam/Venezuela] support Donald Trump and the Republican Party. Do you agree or disagree that the United States should permit higher levels of immigration from [COUNTRY]?"
expectation is that no reverse treatment is necessary because the general U.S. population is primed to believe immigrants support Harris
Assignment of Vietnam vs Venezuela is also random, forming a 2x2 experimental design. The two countries are selected because there is reasonable evidence about turnout for Trump in these groups.
Used a manipulation check to evaluate the treatment. About 20% of the untreated group indicated that the immigrant group would vote for Harris. Compare to about 60% of the treated group.
- validates the authors' expectations
Estimates from the control group demonstrate that "Democrats are most supportive of more immigration (5.3) while Independents (4.2) and Republicans (3.0) are decreasingly supportive. Differences between the groups are statistically significant". Compare to "the treatment condition, the gap in immigration attitudes between Democrats and Republicans completely disappears."
- partisanship is self-identification
Authors also estimate the treatment effect using OLS. Among Democrats it is -0.9; among Republicans it is -1.3.
- Control for income, education, gender, age, and race.
- Estimated by interacting partisanship with treatment indicator. Independents are the left-out level.
- Not statistically significant among independents.
- Further stratifying partisanship by strong and weak Democrats/Republicans reveals that the magnitude of the treatment effect is correlated with strength of partisanship. Again, interacted partisanship with treatment indicator, but now Strong Democrats are the left-out level.
- Also tested using vote intention as opposed to partisanship. Fit data worse.
The authors add that "all three partisan groups prefer Vietnamese over Venezuelan immigrants".
The authors found that their estimates were robust to adding post-stratification weights on gender, age, and level of education.