Misperceptions about Refugee Policy

This letter explores the prevalence of misperceptions about refugee policy and tests whether correcting these misperceptions changes attitudes toward refugees. Large numbers of people hold misperceptions about both the nature and effects of refugee policy. An experiment directly compares the effects of correcting misperceptions about existing refugee policy (e.g., the refugee admission process) with correcting misperceptions about the outcomes of refugee policy (e.g., the proportion of refugees in the United States and the percentage who receive welfare benefits). Corrective information about existing policy substantially increases support for refugees, but corrective information about policy outcomes has no effect on attitudes. The results suggest that including descriptive information about existing U.S. policy in media coverage of refugees could both correct misperceptions and change attitudes.


INTRODUCTION
R efugee policy in the United States is a matter of intense debate, especially in recent years. Former President Donald Trump's January 2017 executive order suspending the entry of Syrians indefinitely and of all other refugees for 120 days was subject to multiple legal challenges. For his part, President Biden pledged to raise the annual refugee admissions ceiling. Public opinion plays an important part in these policy debates (Levy, Wright, and Citrin 2016), and in the wake of Trump's decisions, the public's distrust of refugees has received heavy media coverage. In 2016, the Guardian noted that 80% of Trump supporters saw refugees as a threat (Smith 2016). Americans were divided over admitting refugees, reported PBS Newshour in 2017 (Kellman and Swanson 2017), and the USA Today noted that conservative Republicans' attitudes toward refugees had become more negative (Gomez 2018).
Although most surveys assessing public opinion toward refugees implicitly assume that people have a basic understanding of existing refugee policy, some evidence suggests that public knowledge of both what a refugee is and how they enter the country may be limited (see, e.g., Ipsos 2016). This letter seeks to (1) measure the nature and prevalence of misperceptions about refugees in the US and (2) test whether correcting those misperceptions can affect attitudes toward refugee policy. We find that substantial numbers of people hold misperceptions about both existing refugee policy and the outcomes of those policies.
Correcting misperceptions about existing policy increases support for refugees. However, correcting misperceptions about policy outcomes (including the proportion of refugees, their crime rate, and their welfare dependency) has no effect on attitudes.

MOTIVATION
This section briefly outlines some of the demographic and contextual variables associated with attitudes toward refugees and then discusses an additional potential contributing factor: factual beliefs.

Attitudes towards Refugees
Across 22 European countries, lower education, lower income, older age, and membership in a religious denomination are associated with higher resistance to refugees (Coenders, Gijsberts, and Scheepers 2004). Contextual factors also play a role: for example, people in European countries with higher levels of ethnic diversity are less supportive of refugees compared with those in less ethnically diverse countries (though this finding depends on the measure of diversity used) (Steele and Abdelaaty 2019). Characteristics of the refugees themselves also matter. Using a conjoint experiment, Adida, Lo, and Platas (2019) find that Americans preferred Syrian refugees who are female, highly skilled, English-speaking, and Christian.
Personal interactions with, and exposure to, refugees can also affect opinions. For example, Ghosn, Braithwaite, and Chu (2019) found that interactions with refugees increased Lebanese respondents' support for hosting, hiring, and allowing their children to marry refugees (although see Hangartner et al. 2019). One potential mechanism for this effect is that these interactions increase empathy and perspective taking: Adida, Lo, and Platas (2018) found that respondents who engaged in a perspective-taking exercise ("Imagine that you are a refugee fleeing persecution in a wartorn country") were more likely to write a letter in support of Syrian refugees. However, even though exposure can increase tolerance, some people resist it: U.S. residents are less supportive of refugee resettlement within their own communities compared with elsewhere in the country (Ferwerda, Flynn, and Horiuchi 2017).

Factual Beliefs about Immigrants and Refugees
In addition to demographic and contextual factors, factual beliefs about who refugees are and how they enter the country may also play a role in shaping attitudes. People are more supportive of refugees when they perceive them as involuntary (Verkuyten, Mepham, and Kros 2018), and respondents often hold more positive attitudes toward individuals facing persecution compared with those seeking economic opportunities (Abdelaaty and Steele 2022;Bansak, Hainmueller, and Hangartner 2016;Czymara and Schmidt-Catran 2017).
However, these factual beliefs are not always accurate. The public's understanding of the term "refugee" can change when media coverage casts refugees as voluntary economic migrants in disguise (Findor et al. 2021). The notion of "bogus refugees" has proven to be widespread, with a 22-country study showing that 51% of those surveyed on average (49% in the US) agreed very much or somewhat that refugees were economic migrants rather than people escaping persecution (Ipsos 2016). Similarly, in a study by McKay, Thomas, and Kneebone (2012), 56.8% of respondents believed that asylum seekers came to Australia "for a better life," whereas only 24.4% believed it was "to flee persecution" (see also Bjånesøy 2019).
Several other studies also suggest that misperceptions about refugees abound, and they can shape attitudes. Two-thirds of participants in Australia believed that "most asylum-seekers are queue jumpers," and such false beliefs were highly correlated with negative attitudes (Pedersen, Attwell, and Heveli 2005). In contrast, when Australians saw asylum seekers' claims for refugee status as legitimate, they supported more lenient policies regarding mandatory detention (Hartley and Pedersen 2007). In short, public attitudes may be shaped partly by the public's factual beliefs (and misperceptions) about refugee policies, including the qualifications for refugee status and the admission process.

Attitudinal Effects of Corrective Information
Can correcting misperceptions about refugees have downstream effects on attitudes? Although little existing research has examined this question in the context of refugees, a number of studies have examined how corrective information affects support for immigrants. This section briefly summarizes this literature and then discusses how the findings might shed light on how different types of factual beliefs shape attitudes toward refugees.
Hopkins, Sides, and Citrin (2019) conduct seven separate survey experiments in which they correct individuals' factual misperceptions about the percentage of the U.S. population that is foreign born. Although the interventions successfully increased accuracy, they had no effect on attitudes. Similarly, randomly assigning respondents to receive information about the proportion of immigrants and their incarceration and unemployment rate was effective at correcting misperceptions, but it did not change policy preferences (Grigorieff, Roth, and Ubfal 2020). Corrective information about immigrants' welfare dependency, crime rate, and proportion of the population led participants to update their factual beliefs but not their policy preferences (Jørgensen and Osmundsen 2019). Adida, Lo, and Platas (2018) find that providing information about the number of Syrian refugees admitted by the US compared with other democracies also fails to affect attitudes.
The experiments outlined above provided respondents with information about the effects (either direct or indirect) of refugee policy. These numbers are conceptually similar to so-called "performance" indicators like the unemployment or inflation rate in that they are measures of the outcomes of policies. Although performance measures are critical for retrospective voting (Healy and Malhotra 2013), they are also highly susceptible to motivated reasoning, which can affect not just peoples' factual beliefs but also how they interpret these beliefs (Bisgaard 2019). For example, even when partisans can agree on economic conditions, they interpret these indicators in different ways, blaming the outparty and crediting the in-party (Bisgaard 2015). This process of "partisan rationalization" means that even when partisans share the same set of factual beliefs about policy outcomes, they may use these beliefs to reinforce rather than change their preexisting attitudes (Gaines et al. 2007).
In the context of refugees, these outcomes include refugees' dependence on entitlement programs and their rate of criminal offenses. These quantities are policy outcomes in that they are directly shaped by U.S. refugee policy (e.g., the stringency of background checks). But in addition to holding false beliefs about policy outcomes, people may also be misinformed about basic facts about existing policy, including how people are classified as refugees and the admission process. Information about existing policy may be less subject to partisan-driven motivated reasoning than is information about policy outcomes, for two major reasons. First, because people intuitively seek to attribute responsibility for policy outcomes (Bisgaard 2019), they are more likely to blame (or credit) partisan actors, which in turn colors their interpretations of outcome information (Gaines et al. 2007). In addition, because existing policies (especially ones enacted prior to the current era of political polarization) are usually a product of at least some bipartisan compromise, they may be received more positively across party lines.
The following section outlines a study designed to (1) examine the prevalence of different types of misperceptions about refugees and (2) compare the downstream attitudinal effects of providing two different types of information: information about existing refugee policy (e.g., the legal definition of refugee and the refugee admission process) and information about the outcomes of refugee policy (e.g., the proportion of refugees in the US and their rates of welfare dependency and crime). This design allows for a direct comparison of how each type of information affects attitudes.

DESIGN
In June 2019, a pretest was conducted to investigate potential misperceptions about refugees. In total, 200 respondents, recruited via Amazon's Mechanical Turk, answered open-ended questions asking them to give their own definition of what a refugee is and how they differ from an immigrant. These responses provided insights into common misperceptions about refugees. Many respondents described refugees as seeking a better life, escaping poor living conditions, or fleeing events that would not ordinarily qualify someone for refugee status (such as natural disasters or crime). The responses suggested that misperceptions about how people are classified as refugees and how refugees are admitted to the United States are common. These findings were used to inform the experiment described below (see Appendix for more details on how the pretest shaped the experimental design).
The survey experiment was fielded by Lucid in December 2020. Lucid matches U.S. Census demographics by using quota sampling, and participants recruited by Lucid behave similarly to respondents in representative samples on several experimental benchmark surveys (Coppock and McClellan 2019). In total, 2,565 people completed the survey, and all were included in the analyses. First, all respondents answered several demographic questions including education, party identification, and whether one or both parents were born outside the US. All of these were asked prior to the treatment so that they could be used as covariates in the experimental analyses.
Participants were then assigned to one of five conditions in a 2 (policy outcomes vs. existing policy) Â 2 (corrective information vs. no corrective information) plus pure control experiment. The full experimental design is illustrated in Figure 1, and the questionnaire is available in the Appendix.
Respondents in the existing refugee policy condition were asked four questions. The first question asked them to indicate what reasons qualify someone for refugee status (e.g., persecution for religious beliefs or coming from a country with high levels of corruption). The next three asked about refugee admission policy: whether refugees were required to undergo background checks, whether they received resettlement assistance, and whether they applied directly to the US or to the United Nations. Then, half were randomly assigned to see the correct answers to those policy questions.
Respondents in the refugee policy outcomes condition were asked to estimate the size of the refugee population living in the United States, the percentage convicted for terrorism-related offenses, and the percentage dependent on welfare. These questions were based on similar measures used in past corrective interventions directed at correcting misperceptions about immigrants (Hopkins, Sides, and Citrin 2019; Jørgensen and Osmundsen 2019). Half then saw the correct answers to these factual questions.
Respondents in a pure control condition saw neither the set of questions nor the answers. The inclusion of a pure control condition is necessary for establishing a baseline because the factual questions themselves may shape attitudes by making particular considerations (for example, background checks or crime rates) more salient.

Refugee policy outcomes Existing refugee policy
Questions only x What proportion of U.S. residents are refugees x What proportion of refugees receive welfare benefits x What proportion of refugees have been convicted of terrorism-related offenses x Reasons that qualify someone for refugee status (check all that apply) x Whether background checks are required x Whether refugees receive assistance x Whether refugees apply directly to US.

Questions followed by corrective information
x Proportion of U.S. residents who are refugees (.06%) x Proportion of refugees who receive welfare benefits (6%) x Proportion of refugees who have been convicted of terrorismrelated offenses (.00074%) x Legal definition of refugees x Description of refugee application process x Description of refugee resettlement process

No questions or corrective information
Next, all respondents answered a number of questions assessing their attitudes toward refugees along two different dimensions: policy support and support for admitting specific refugees. We chose to explore these two outcomes because they are common approaches for measuring attitudes toward refugees and immigrants (see, e.g., Alrababa'h et al. 2021). To indicate policy support, respondents indicated whether they supported or opposed five policies (1-5 scale, M = 2.8, α = 0.74): giving loans to refugees to finance their travel to the US, allowing refugees to receive food stamps, allowing them to bring their immediate family members to the US, implementing stricter background checks, and temporarily pausing all refugee admissions during the COVID-19 pandemic. To assess support for refugee admission, they were shown images and brief biographic information about three refugees and asked whether they would support admitting them to the United States (1-5 scale, M = 3.8, α = 0.74). Finally, respondents answered the open-ended question "When you think of refugees who seek to come to the United States, what thoughts come to mind?"

RESULTS
At the end of the survey, respondents were asked slightly different versions of the same factual questions that they were asked earlier. These questions served as a manipulation check. In the existing refugee policy condition, respondents who received the corrective information answered 4.8 of the 7 questions correctly, compared with 3.7 among those who did not, t(1013) = 4.6, p<0.001. In the refugee policy outcomes condition, respondents who received the corrective information answered 1.2 out of the 3 questions correctly, compared with 0.7 for those who did not, t(1035) = 8.9, p<0.001. The results suggest that both treatments were quite successful at changing factual beliefs.

Prevalence of Misperceptions
In the policy outcomes condition, the median number of factual questions correct was two out of three, and in the existing policy condition, the median number was four out of six. 1 Thus, the modal respondent in both groups answered at least 50% of the questions correctly, suggesting that both types of misperceptions are prevalent but not omnipresent. Table 1 shows the five the most commonly held misperceptions about existing refugee policy. Almost two-thirds of respondents believed that refugees apply directly to the U.S. government rather than to the UN and that coming from a country with high levels of crime, poverty, or corruption qualifies a person for refugee status. More than one in four respondents believed that no background checks are required for refugees.
Misperceptions about policy outcomes were also widespread. Similarly to previous surveys measuring misperceptions about immigrants, respondents in the outcomes condition substantially overestimated the prevalence of refugees (median answer = 19%, correct answer = 0.06%), their dependence on welfare (median answer = 19%, correct answer = 6%), and their rate of terrorism-related convictions (median answer = 5%, correct answer = 0.00074%). Figure 2 shows the effect of the two treatments on support for refugee policy and support for admitting individual refugees as compared with the pure control condition. Party identification, education, gender, parental origin, race, and political interest are included as covariates (analyses without covariates are included in the Appendix).

Effects of Treatments
Neither being asked questions about policy outcomes nor being additionally provided with the answers to those questions affects attitudes toward refugees as compared with the pure control. In contrast, those who were asked questions about refugee admission without being given the correct answer were slightly more supportive of refugee-friendly policies than were those in the control condition. In other words, simply making aspects of existing refugee policy salient increased policy support (although this was not the case for admission support). In addition, receiving the correct answer to those questions substantially increased support for refugee-friendly policies and admission, moving respondents by about 0.3 on a five-point scale on both indices. To put this effect size in context, it is of a magnitude similar to the difference between Independents and Democrats. The effect of receiving information about existing policy is not conditional on party: effects on both policy attitudes and support for admission are similar for Democrats, Republicans, and Independents (analyses in Appendix).
We also conducted an exploratory analysis of the open-ended responses to the question "When you think of refugees who seek to come to the United States, what thoughts come to mind?" Compared with the control group, respondents who received information about existing policy were significantly more likely to reference the dangers and persecution that refugees face and less likely to mention terrorism and negative effects on the US (see Appendix). In contrast, the responses in the policy outcomes condition were indistinguishable from those in the control group.

CONCLUSION
This study examines the prevalence of false beliefs about existing refugee policy as well as the outcomes of those policies. In line with previous studies, we find that people overestimate the number of refugees admitted into the US, their dependence on welfare, and their involvement in terrorist activity. However, correcting these misperceptions does not shift attitudes toward refugees. This null result parallels the findings in studies that randomly assign people to receive similar information about immigrants (Grigorieff, Roth, and Ubfal 2020;Hopkins, Sides, and Citrin 2019;Jørgensen and Osmundsen 2019), suggesting that facts about policy outcomes may be particularly subject to partisan interpretation (Bisgaard 2015;Gaines et al. 2007). However, we find that people also hold a range of inaccurate factual beliefs about existing refugee policy, including the legal definition of refugees and how they are admitted to the United States, and providing corrective information about these policies substantially increases support for refugees.
Our findings suggest that researchers should be wary of assuming that respondents understand the term "refugee" and what it entails. Indeed, the enduring finding in the literature that education is positively correlated with pro-refugee attitudes may be related to knowledge about refugee status and associated policies. Journalists might consider including basic background information about refugee policy in their coverage of the issue (although the effects of this information might vary depending on levels of media trust). In addition, these results suggest that for policy makers or advocacy organizations who wish to garner public support for refugees, simply describing existing policy might (perhaps counterintuitively) be a more effective strategy for increasing support across party lines than offering statistics about the outcomes of those policies.
Of course, misperceptions about refugees are likely to vary by national and temporal context, and additional studies are needed to describe patterns of misperceptions about refugees both across countries and over time. It will also be valuable for future research to examine whether correcting policy information rather

Policy Support Admission Support
Existing Policy Policy Outcomes than performance measures alters attitudes in other issue areas as well as to investigate other potential mechanisms for this effect. For example, information about existing policy (as opposed to policy outcomes) may be cognitively easier to integrate into existing attitudes. Alternatively, it may change which exemplars come to mind (e.g., women versus men) or prime a different facet of the refugee experience (e.g., the admission process versus assimilation), which may in turn shape attitudes.
With the dramatic increase in the global refugee population, it is all but certain that refugee policy will continue to be a matter of public debate in the United States and elsewhere. Research on public opinion toward refugees should pay attention to how factual beliefs and misperceptions about refugee status and the admissions process shape attitudes.

SUPPLEMENTARY MATERIALS
To view supplementary material for this article, please visit http://doi.org/10.1017/S0003055422000910.

DATA AVAILABILITY STATEMENT
Research documentation and data that support the findings of this study are openly available at the American Political Science Review Dataverse: https://doi. org/10.7910/DVN/MB8FGY.