It’s common to see surveys ask Americans what they think about immigrants, but how much can the way those surveys' questions are phrased affect their answers? Do Americans have different attitudes toward immigrants as a whole than they do toward immigrants living in the country illegally? And do Americans express different opinions depending on what words pollsters use when asking about this group of immigrants? There is a fierce debate over terminology, but does it affect results if surveys ask about “undocumented immigrants” or “illegal immigrants”?
At YouGov, we aim to ask questions that prompt our panelists to express their views accurately. Part of striving to achieve that goal is testing a variety of approaches to asking about major issues such as immigration. A year ago we conducted an experiment randomly assigning panelists to answer questions about their opinions towards “immigrants,” “illegal immigrants,” or “undocumented immigrants.” Our results show that small changes in how immigrants are asked about can lead to big changes in how Americans answer.
While the term "immigrants" encompasses a broad range of people in the U.S., "illegal immigrants" and "undocumented immigrants" generally are used in the news media and other contexts to describe similar groups of people — and they evoke similar levels of negative sentiment. But positive sentiment differs in a surprising way — even if the two terms' commonly understood meanings are similar.
Americans are more likely to express unfavorable opinions about “illegal immigrants” and about “undocumented immigrants” than about “immigrants” as a whole. While 22% of Americans say they have an unfavorable opinion of “immigrants,” nearly twice that share express unfavorable views of “illegal immigrants” (38%) and “undocumented immigrants” (39%).
This pattern holds among Democrats, Independents, and Republicans, although the size of the gap varies. There is only a modest difference in the shares of Democrats who have unfavorable attitudes towards “immigrants” (12%), “illegal immigrants” (15%), and “undocumented immigrants” (17%). The differences among Independents and Republicans are considerably larger. While only 18% of Independents have an unfavorable view of “immigrants,” more than twice as many hold unfavorable views of “illegal immigrants” (41%) and “undocumented immigrants” (45%). Republicans are most likely to express an unfavorable opinion of “immigrants” (42%), but unfavorable attitudes towards “undocumented immigrants” (68%) and “illegal immigrants” (69%) are even more widespread.
On the other hand, the shares of Americans with favorable attitudes towards these immigrant groups show a surprising difference. The shares of Americans expressing favorable opinions towards “immigrants” (20%) and “illegal immigrants” (20%) are the same, while the share expressing a favorable opinion of “undocumented immigrants” (13%) is smaller.
This pattern holds within political parties. Democrats are most likely to express positive opinions of immigrant groups regardless of wording, but favorable attitudes towards “immigrants” (31%) and “illegal immigrants” (31%) are more widespread than those towards “undocumented immigrants” (24%). Republicans show the same trend, although attitudes are less positive overall. While 8% of Republicans express a favorable opinion of “immigrants” and 6% do so for “illegal immigrants,” only 2% are favorable toward “undocumented immigrants.” Independents show the largest gap of all: while 22% have a positive opinion of “immigrants” and 21% say the same about “illegal immigrants,” just 6% express a favorable view of “undocumented immigrants.”
This raises a question: Why do fewer Americans express positive attitudes about “undocumented immigrants” than about “illegal immigrants”? This seemingly runs counter to the argument from some advocates that the term “illegal immigrant” incites anti-immigrant prejudice. If “illegal immigrant” is an inherently negative term, one would expect that Americans would be especially unlikely to express favorable attitudes about the group when this wording is used.
One possible explanation: Americans may be less likely to express a favorable attitude when asked about “undocumented immigrants” because they are less familiar with the term. 48% of Americans say their opinion of “undocumented immigrants” is neither favorable nor unfavorable, compared to 42% who say that about “illegal immigrants.” This may help to explain why Independents, who may be least exposed to partisan rhetoric, are least likely to express a positive attitude about undocumented immigrants. 49% of Independents are neither favorable nor unfavorable toward “undocumented immigrants,” compared to only 37% with neither a favorable nor unfavorable opinion of “illegal immigrants.”
Another possible explanation is that using the term “illegal immigrants” sparks a backlash among Americans who are sympathetic towards members of this group. Americans who have been persuaded that the term “illegal immigrants” should no longer be used may have a strong negative response to the term. Seeing what they perceive as politically loaded wording may lead these respondents to give a more positive response toward people being described by the term. Americans with pro-immigrant attitudes may not exhibit this same kind of backlash effect when asked about “undocumented immigrants” because this term does not carry the same charged connotations.
Investigating further with a list experiment
To better understand Americans’ responses to “illegal immigrants” and “undocumented immigrants,” we conducted a test known as a list experiment. Half of this survey’s respondents were randomly selected to participate in the list experiment. These participants did not directly answer questions about their attitudes towards immigrants. Instead, we showed them several groups at the same time and asked them to give a single number describing how many of those groups they viewed favorably. List experiments are intended to mitigate reasons that survey respondents may exaggerate or downplay their opinions. Even though all YouGov survey responses are anonymous, respondents still may feel pressure to respond in a way that aligns with other members of a group they belong to or that advances their political beliefs. In a list experiment, respondents can see that it is impossible to determine how they view any specific group. Instead, we can only observe the total number of groups each participant views favorably. Research has shown that when asked to answer questions in this more indirect manner, survey participants are less affected by pressures to answer in a particular way.
Of course, this creates an obvious problem: If we cannot observe respondents’ answers directly, how can we learn what they think from a list experiment? The key is to randomly vary only one item in the list. In this experiment, participants were asked, “How many of the following 6 groups do you have a favorable opinion of?” and were told, “You don't have to indicate which groups you have a favorable opinion of, just how many.” The following five groups were shown to all respondents: liberals, moderates, conservatives, atheists, and Evangelical Christians. The sixth group was randomly assigned to be either “illegal immigrants,” “undocumented immigrants,” or “immigrants.” Because only the last group differed among respondents, we can compare the numbers of groups viewed favorably by respondents who were shown each of the different groups, and know that any differences can only be explained by different attitudes towards that last group. That doesn't tell us about individual respondents’ attitudes toward immigrants, but in the aggregate it shows the average effect of language on expressed opinion.
Americans who are asked about their opinions of liberals, moderates, conservatives, atheists, Evangelical Christians, and “immigrants” hold favorable opinions of an average of 2.53 groups. The average drops to 2.27 when “immigrants” is replaced with “undocumented immigrants,” and to 2.25 when it is replaced with “illegal immigrants.” Because the average response with “immigrants” on the list (2.53) is 0.26 higher than the average for lists with “undocumented immigrants” (2.27), we can deduce that 26 percentage points more Americans hold a favorable opinion of immigrants than hold a favorable opinion of undocumented immigrants. In contrast, because the average responses to the list with “undocumented immigrants” and the list with “illegal immigrants” differ by only 0.02, there appears to be only a 2 percentage point difference in the share of Americans who hold a favorable view of undocumented immigrants and the share who hold a favorable view of illegal immigrants. This difference is so small that we cannot conclude that it is statistically significant. This means we do not have evidence that there is any difference in attitudes towards “illegal immigrants” and “undocumented immigrants.”
Americans of all parties give higher average responses about “immigrants” than about “illegal immigrants” or “undocumented immigrants,” according to the list experiment. Democrats show the largest difference. Their average response is 2.96 when their list includes “immigrants,” but only 2.44 with “undocumented immigrants” and 2.41 with “illegal immigrants.” This difference of a little over 0.5 suggests that about 50 percentage points more Democrats hold a favorable view of “immigrants” than of “undocumented immigrants,” and a similar gap holds with “illegal immigrants.” For Independents, this difference is 21 percentage points, and for Republicans it is 17 percentage points.
In contrast, the difference in the averages given by Americans seeing lists with “illegal immigrants” or “undocumented immigrants” is smaller than this survey’s margin of error, overall and among Democrats, Independents, and Republicans. This suggests that Americans’ opinions of unauthorized immigrants do not truly depend on which label is used. This contradicts our findings when Americans were asked directly about their opinions of “illegal” and “undocumented” immigrants. The differences that we observe when Americans are asked about their opinions directly may therefore reflect pressure that respondents feel to answer in a certain way depending on whether a more positive label (“undocumented”) or a more negative label (“illegal”) is used. When that pressure is relieved using a method such as a list experiment, there is no difference.
Given these findings, it is important to take care when interpreting Americans’ attitudes about immigrants. Questions that use unfamiliar terms risk downplaying Americans’ attitudes, while questions that use terms that some consider biased may prompt more extreme responses.
— Taylor Orth and Carl Bialik contributed to this article
See the results for this YouGov survey
Methodology: These questions appeared across three YouGov polls conducted online from December 20, 2023 - January 2, 2024; December 21 - 30, 2023; and January 12 - 17, 2024. 3,000 respondents were randomly assigned to answer questions about opinions towards “immigrants,” “illegal immigrants,” or “undocumented immigrants,” and were randomly assigned to answer these questions either directly or in a list experiment. Respondents were selected from YouGov’s opt-in panel to be representative of adult U.S. citizens. The sample was weighted according to gender, age, race, education, 2020 election turnout and presidential vote, baseline party identification, and current voter registration status. Demographic weighting targets come from the 2019 American Community Survey. Baseline party identification is the respondent’s most recent answer given prior to November 1, 2022, and is weighted to the estimated distribution at that time (33% Democratic, 31% Republican). The margin of error for the overall sample selected to answer a question is approximately 6.5 percentage points.
Image: Getty (Chip Somodevilla / Staff)
