Voters weighing economic costs of deportations
65% of voters say the administration is focusing too much on deportations and not enough on the cost of living, and only 21% want the government to deport every undocumented immigrant regardless of whether they have committed a crime.
A new national survey of 1,591 U.S. registered voters finds broad rejection of how immigration enforcement is being carried out. Sixty-five percent agree the administration is “focusing too much on deportations and not focusing enough on fixing the economy, inflation, and the rising cost of living”; 27% disagree. Voters draw the line inside enforcement itself: 87% call deporting violent criminals acceptable, and majorities of 57–74% say every other tactic the survey tested goes too far. The rejection runs deepest among Black, Latino, and lower-income voters. Key findings:
1.There is no majority for mass deportation: only 21% of voters want it.
If mass deportation had a popular mandate, it would show up here. It does not: asked which approach to immigration enforcement comes closest to their view, only 21% of voters choose deporting all undocumented immigrants regardless of whether they have committed a crime. Nearly three-quarters choose something narrower — 37% would deport serious criminals while offering a path to legal status to those who work and pay taxes, the most popular of the four options; 27% would focus on the economy and the cost of living instead of mass deportations; and 10% would scale enforcement back across the board. The position voters actually endorse is explicit: asked to agree or disagree with a single statement, 73% say both — deport undocumented immigrants who commit a crime, and give hard-working undocumented workers a path to legal status. The same battery shows the mirror image: a hardline statement — that it is “time to deport immigrants who illegally crossed the border regardless of how long they’ve been here” — is underwater, 41% agree to 50% disagree. Side by side, the hardline frame nets −9; the enforce-and-legalize frame nets +49.
2.Voters separate border security from deportations.
Fifty-six percent of voters disapprove of the administration’s handling of deportations and immigration enforcement and 41% approve, a 14-point deficit; 47% disapprove strongly, against 30% who approve strongly. Border security is the administration’s best issue of the six the survey tested, and it still runs level: 47% approve, 49% disapprove. Approval of border security sits six points ahead of approval of deportations, and its net rating 13 points ahead — the same voters, rating two parts of the same agenda differently.
The August reading is not a blip. Wave has fielded the same handling battery in each of its three national surveys this year, and the gap holds in every one: net approval of Trump’s handling of border security ran +5, even, and −1 across February, May, and August, while deportations sat at −17, −14, and −14 — border security ahead of deportations every time, and deportations deeply underwater in all three.
3.Immigration enforcement focus and tactics are broadly unpopular.
Eighty-seven percent of voters call deporting undocumented immigrants who have committed violent crimes an acceptable part of enforcement. Nothing else comes close: between 57% and 74% say each of the other ten actions goes too far, from deporting long-time residents with no criminal record (57%) to using deadly force against someone who is not armed (74%). The shootings are widely known — 53% of voters had heard a lot about federal immigration agents shooting and killing people during enforcement operations, and 37% a little — and voters do not want the agencies involved investigating themselves: 26% would leave those investigations to the Department of Homeland Security, against 65% who would hand them to an independent, state, or local investigator. Detention runs the same way: 69% say immigrants who pose no flight risk and attend their hearings should not be held while their cases proceed, and 63% call building more ICE detention centers a waste of taxpayer dollars.
4.The deportation focus is weakest with Black, Latino, and lower-income voters.
The groups rejecting the deportation focus most emphatically are the ones Democrats rely on. Eighty-six percent of Black voters and 72% of Latino voters agree the administration is focusing too much on deportations and not enough on the cost of living, against 65% of all registered voters; among voters in households earning under $50,000 it is 71%. Disapproval of the administration’s handling of deportations reaches 77% among Black voters and 64% among Latino voters. But these are not open-borders blocs: only 9% of Black voters and 13% of Latino voters choose deporting everyone, and their preferred alternatives differ — the leading answer among Black voters is focusing on the economy instead of mass deportations, at 44%, while Latino voters are the strongest constituency for pairing deportation of serious criminals with a path to legal status, at 41%.
5.Voters find economic messaging on immigration convincing.
In a randomized experiment, each respondent rated one of two versions of the same message about deportations raising grocery and health care costs. The version ending “…while billionaires get another tax break” rated higher than the economic-cost version alone: 55% called it convincing, 7 to 10 on a 0-to-10 scale, against 45%, and its mean rating was 6.1 against 5.3. The arms were not perfectly matched to begin with — before either message, 59.5% of the populist arm and 53.4% of the economic-cost arm already said mass deportation is hurting the economy and raising costs — so part of the ratings gap reflects who landed in each arm. And neither version moved that underlying question: re-asked after the message, the two shares stood at 60.1% and 54.0%, six-tenths of a point higher in each arm. This survey shows the two messages rating differently; it does not show either one changing what voters think about deportation.
Methodology
This national survey of 1,591 U.S. registered voters was conducted by Wave Polling and Research. Fieldwork was carried out online August 7–11, 2026 by Verasight. All respondents were affirmatively matched to the L2 voter file database as registered voters.
Weighting. The data are weighted by Wave Polling and Research using iterative proportional fitting (raking) to registered-voter targets on eleven dimensions: age, race/ethnicity, sex, metropolitan status, household income, education, Census region, and white/non-white race by college or non-college educational attainment, benchmarked to the Current Population Survey; leaned three-way party identification (Democrats including independents who lean Democratic, pure independents, and Republicans including independents who lean Republican), benchmarked to the Pew Research Center’s 2026 National Public Opinion Reference Survey (NPORS); 2024 presidential vote, benchmarked to the certified national results scaled to the registered-voter frame’s turnout, with the non-voter share held at its benchmark value; and local quintile of Trump vote share, in which each respondent’s zip code is matched to precinct-aggregated presidential results calibrated to certified county, congressional-district, and state totals, and binned by the surrounding area’s Trump share of the two-party vote averaged over 2020 and 2024, with each quintile holding one-fifth of the national two-party vote. Weights were raked from uniform starting values and iteratively trimmed to the range [0.2, 5] with margins re-imposed after each trim.
Margin of error. The margin of sampling error accounts for the survey’s design effect, a measure of the loss of statistical power due to survey design and weighting. The Kish design effect of the weights is 1.45, for an effective sample size of 1,097 and a margin of sampling error of plus or minus 2.96 percentage points on a 50% proportion. Real-world error includes sources of error beyond sampling error — such as nonresponse bias, coverage error, and measurement error in self-reports — and margins of sampling error understate total survey error. One such source is error in the weighting benchmarks themselves: re-weighting across the NPORS partisanship benchmark’s approximate 95% sampling band increases potential survey error by about 1.01 percentage points for questions highly correlated with vote choice — which, combined in quadrature with the margin of sampling error, yields an adjusted margin of error of plus or minus 3.14 percentage points. A full sensitivity analysis appears in the topline appendix.
Fieldwork provider methodology. The following describes Verasight’s sampling and data-quality procedures for this project. The sampling criteria were: adult (age 18+), U.S. resident, and affirmatively matched to the L2 voter file database as a registered voter; the final sample includes only respondents who passed all data quality assurance checks, outlined below. The margin of sampling error calculated by Verasight using the classical random sampling formula, before accounting for the design effect of weighting, is plus or minus 2.5%.
All respondents were recruited via email from the Verasight Community, which is composed of individuals recruited via random address-based sampling, random person-to-person text messaging, and dynamic online targeting. All Verasight community members are verified via multi-step authentication, including providing an SMS response from a mobile phone registered with a major U.S. carrier as well as within-survey technology verifying the absence of non-human responses, such as Google reCAPTCHA v3.
To further ensure data quality, the Verasight data team implements a number of post-data-collection quality assurance procedures, including confirming that all responses correspond with U.S. IP addresses, confirming no duplicate respondents, verifying the absence of non-human responses, and removing any respondents who failed in-survey attention and/or straight-lining checks. Respondents who completed the survey in less than 30% of the median completion time were removed. Unmeasured error in this or any other survey may exist. Verasight is a member of the American Association for Public Opinion Research Transparency Initiative.