A pathway to citizenship is a priority for voters, and “amnesty” attacks change few minds.
79% of registered voters support a path to citizenship for immigrants who have lived here ten years, pay taxes and pass a background check. An “amnesty” attack and response cuts net support by just 11 points and leaves a 64% majority.
Three October surveys of registered voters, one national and one each in Florida and Pennsylvania, asked respondents to rank 14 immigration policies for a new Congress, tested support for three versions of a path to citizenship, and ran a randomized experiment exposing that path to an “amnesty” attack and a rebuttal. The memo carries the findings. The toplines carry every question, its wording, and its weighted results; the crosstabs break each question out by subgroup. A separate file reports the policy-priority (MaxDiff) rankings for Hispanic voters and voters ages 18–29 in the national survey.
Methodology
Each survey was conducted by Wave Polling and Research, with fieldwork carried out online by Verasight. All respondents were matched to the voter file as registered voters; the Florida and Pennsylvania respondents were also matched to the voter file as registered in that state and confirmed their state of residence in the survey.
| Survey | Field dates | Sample | Design effect | Margin of error |
|---|---|---|---|---|
| National | Sept. 30–Oct. 5, 2026 | 1,000 | 1.23 | ±3.4 pts |
| Florida | Oct. 1–5, 2026 | 1,030 | 2.08 | ±4.4 pts |
| Pennsylvania | Oct. 1–5, 2026 | 502 | 1.65 | ±5.6 pts |
Weighting. Data are weighted by raking to registered-voter targets for age, race/ethnicity, sex, household income, education, and race by college attainment, benchmarked to the Census Bureau’s November 2024 Current Population Survey Voting and Registration Supplement for the relevant geography; to the 2024 presidential vote, benchmarked to certified results; and to the local quintile of Trump vote share, in which each respondent’s zip code is matched to precinct-level presidential results. The national survey is also weighted to Census region and to leaned three-way party identification, benchmarked to the Pew Research Center’s 2026 National Public Opinion Reference Survey. Weights were trimmed to the range [0.2, 5].
Margin of error. Each margin of sampling error above accounts for the survey’s design effect and applies to a 50% proportion on the full sample. Margins are larger for subgroups and for comparisons between experimental conditions. Margins of sampling error understate total survey error, which also includes nonresponse, coverage, and measurement error. The full methodology statement, including Verasight’s sampling and data-quality procedures, is on the first page of each topline.