Displacement is a pressing geopolitical matter. During the COVID-19 pandemic, some countries enacted dramatic measures to reduce migration by closing their international borders; nevertheless, 11.2 million people were forced to flee their homes in 2020. Today, 1% of the global population is forcibly displaced, continuing a decade-long trend. Between 2010 and 2020, the number of forcibly displaced individuals more than doubled, from 40 million to 80 million people (United Nations High Commissioner for Refugees 2022).
Concurrently, the United States often has applied refugee policies unevenly, haphazardly, and unjustly for political expediency (Bhardwaj Reference Bhardwaj2023). For example, in March 2022, the Biden administration announced a program that would allow 100,000 Ukrainians to stay and work in the United States for two years. Conversely, the 85,000 Afghan refugees who arrived in the United States after the withdrawal from Afghanistan in August 2022 could receive legal work permits for only 18 months.
Furthermore, US policies do not currently accommodate populations that likely will constitute a growing portion of future displaced migrants: those who are displaced by climate change and natural disasters. Climate-induced migrants, or those “forced to leave their traditional habitat, temporarily or permanently, because of marked environmental disruption” (El-Hinnawi Reference El-Hinnawi1985), are frequently absent from the global debate on those seeking refuge. This is especially worrisome given that the Institute for Economics & Peace (2020) estimates that more than 1 billion people will be displaced due to climate change by 2050.
US policies do not currently accommodate populations that likely will constitute a growing portion of future displaced migrants: those displaced by climate change and natural disasters.
The US government has done little to formulate policies to protect climate migrants (Mahmud Reference Mahmud2022), and only a few studies have explored the US public’s response to climate migrants. This dearth of federal policies coupled with the inevitable and substantial increase in climate-induced migrants across the globe and also in the United States leaves open a pressing question: Are Americans willing to increase the number of climate migrants in response to this humanitarian challenge? Or are they more willing to accept those seeking refuge from conflict in their country? This article presents the results of a survey experiment on a national sample of Americans fielded in April 2022 in the throes of the Ukraine crisis. In particular, we examined whether climate migrants face similar levels of acceptance compared to others who require humanitarian intervention, such as those who are fleeing violence in Ukraine and Afghanistan. Understanding how climate-induced migrants are perceived relative to other refugees provides important information for elected officials and policy advocates to anticipate public responses to increasing influxes of climate-induced migrants.
We found that migrants displaced by climate change are less likely to receive support compared to other forced migrants fleeing conflicts, particularly those fleeing Ukraine. Subgroup analyses reveal that these results persist most consistently among Republican, white, and Christian Americans. These results are important because climate displacement will continue to require refuge, and assessing the public’s reaction to these refugees has implications for their incorporation in receiving countries. Because the effects of climate change disproportionately affect marginalized populations, an informed public dialogue that embraces the personhood and place of refugees is a growing international concern (Williams Reference Williams2008). However, our results suggest that there are limits to the American public’s humanitarianism.
We found that migrants displaced by climate change are less likely to receive support compared to other forced migrants who are fleeing conflicts, particularly those who are fleeing Ukraine. Subgroup analyses reveal that these results persist most consistently among Republican, white, and Christian Americans.
LITERATURE REVIEW
As the number of people displaced by climate events increases (Portner et al. Reference Portner, Roberts, Trisos and Simpson2022), those who are seeking refuge from conflict often find themselves displaced to climate “hotspots,” which are uniquely susceptible to environmental hazards and even further displacement (United Nations High Commissioner for Refugees 2022). Vulnerable people will be forced from their homes, and other groups and countries will need to admit and incorporate refugees into their communities. Our study considers this population and compares it to others motivated by different push factors, such as conflict.
Climate migrants can be displaced by many different types of natural disasters that require immediate assistance, or they can be uprooted due to more long-term patterns inflicted by climate change: droughts, wildfires, floods, and hurricanes (Mahmud Reference Mahmud2022). Currently, developing nations are disproportionately impacted by climate events (Bathiany et al. Reference Bathiany, Dakos, Scheffer and Lenton2018; Blair, Grossman, and Weinstein Reference Blair, Grossman and Weinstein2022). However, the downstream effects of climate change mean that nations such as the United States will face these challenges to a greater degree in the future (Marotzke, Semmann, and Milinski Reference Marotzke, Semmann and Milinski2020). Although there are few legal protections for these migrants (McAdam Reference McAdam2012), the anticipated growth of this population and the lack of existing public policy means that public opinion may play an important role in the coming years.
Scholarship is beginning to explore public responses to these migrant populations. In these studies, scholars find similar but slightly lower support for international migrants who are fleeing the consequences of climate change than those who are fleeing persecution (Arias and Blair Reference Arias and Blair2022). This research also finds that people in receiving countries are more willing to settle climate migrants compared to economic migrants. Our research contributes to this literature by investigating whether people support increasing the number of refugees admitted to the country in response to climate change rather than by studying decisions made about individual refugees.
In the United States, research has centered on support for those displaced by conflict and economics. Adida, Lo, and Platas (Reference Adida, Lo and Platas2019) found that Americans perceive high-skilled, English-speaking, Christian women as the most desirable Syrian refugees. Other studies found that country of origin, jobs, gender, and religion all shape favorability toward individual refugees and/or migrants (Steele, Abdelaaty, and Than Reference Steele, Abdelaaty and Than2023), mirroring the concerns of elected officials (Shaffer et al. Reference Shaffer, Pinson, Chu and Simmons2020).
The following section compares support for those who are forced to leave their homes by climate to those who are forced to leave by conflict. We also inquired about those who are fleeing Ukraine and Afghanistan—countries with different religious compositions and migrants who experience the threat of violent persecution.
EXPECTATIONS
Recent research finds that climate migrants occupy a middle category in the public view between economic and persecuted migrants (Arias and Blair Reference Arias and Blair2022). Because their migration is due to climatic factors beyond their control, they are perceived as involuntary migrants, which spurs greater empathy and support from citizens of receiving nations than voluntary labor and economic migrants (Arias and Blair Reference Arias and Blair2022). In other words, these migrants are simultaneously perceived as fleeing conditions beyond their control and as economic threats requiring less support than other refugees (Arias and Blair Reference Arias and Blair2022). Additional research also suggests that changing these attitudes can be challenging. Even climate disasters increase support for climate migrants for only a short duration (Arias and Blair Reference Arias and Blair2024). We anticipated a similar distinction, in which Americans will be most sympathetic toward conflict-fleeing refugees and less sympathetic toward climate migrants.
H1a: Those reading about climate migrants will be less likely to support them relative to those reading about migrants who are fleeing conflict.
H1b: Those reading about climate migrants will be less likely to support them relative to those reading about refugees in general.
Similarly, we anticipated that individuals will respond differently to those fleeing conflict depending on their country context and religion—two factors that are conflated in real-world questions about which refugees to admit. We expected that refugees who are European, Christian, and mostly white to elicit more support relative to other forced migrants. We expected this because extant research found that support for refugee policy tends to increase when those refugees are white (Rosenberg Reference Rosenberg2022), and Christian refugees historically have provoked greater support in both the US and the European contexts (Adida, Lo, and Platas Reference Adida, Lo and Platas2019). It is important to note that these two factors are intimately linked in real-world discussions of refugee policy. For example, Grace and Heins (Reference Grace and Heins2021, 556) found that the discourse around South Carolina’s refugee policy combines race, religion, and meaning, constructing refugees as “Brown, Muslim, Terrorist, Third World.” Given that most climate migrants come from the Global South and thus are unlikely to share the same race, national origin, and religion as most refugees from Ukraine, we expected greater support for Ukrainian refugees relative to others, including those seeking refuge for climate concerns.
H2: Those reading about Christian refugees from Ukraine will be more likely to support accepting them relative to those reading about other forced migrants, including climate migrants.
Finally, we expected support for climate migrants to be driven in part by the respondents’ personal characteristics. Namely, Americans’ partisanship, race, and religion will shape their willingness to support admitting additional refugees. Republicans are less likely to support refugees in general (Newman Reference Newman2018) and are more resistant to frames that encourage action on climate change (Zhou Reference Zhou2016). Similarly, white Americans are unsupportive of permissive immigration policies (Abrajano and Hajnal Reference Abrajano and Hajnal2015), and they prefer Christian over Muslim refugees (Nassar Reference Nassar2020). They also are more likely than their non-white counterparts to have more politically polarized climate-change attitudes (Schuldt and Pearson Reference Schuldt and Pearson2016). Furthermore, scholars have found that Christians—namely, Evangelical Protestants—hold relatively higher levels of animus toward immigrants (McDaniel, Nooruddin, and Shortle Reference McDaniel, Nooruddin and Shortle2011) and refugees (Whitehead and Perry Reference Whitehead and Perry2020); are less alarmed about climate change; and take fewer climate-change–related actions (Morrison, Duncan, and Parton Reference Morrison, Duncan and Parton2015). Although support for these individual hypotheses is found in the previously discussed studies, it is worth noting that these are not completely discrete. As social and political identities (e.g., Republican, Christian, and white Americans) become more aligned, this amplifies political polarization and results in bias, prejudice, and anger having a more prominent role in political decision making (Mason Reference Mason2018). Based on this research, we derived the following hypotheses:
H3a: Self-identified Republicans will be more likely than their partisan counterparts to support admitting Christian forced migrants from Ukraine.
H3b: White Americans will be more likely than non-white Americans to support admitting Christian forced migrants from Ukraine.
H3c: Christians will be more likely than their non-Christian counterparts to support admitting Christian forced migrants from Ukraine.
DATA AND METHODS
In April 2022, we fielded a survey experiment on a national sample of US adults collected by IPSOS Knowledge Panel service in collaboration with the Public Religion Research Institute (PRRI). IPSOS uses address-based sampling and online recruitment to create nationally representative omnibus survey samples. Our original sample size was 1,027 respondents, with a total analytic sample of 1,008 after removing 19 cases with missing data on the dependent variable (Stewart et al. Reference Stewart, Lajevardi, Williams and Whitaker2024).Footnote 1
To assess whether Americans distinguish between those migrating to the United States due to different contexts or push factors, respondents were randomized into one of four treatment conditions asking whether additional refugees should be admitted to the United States. They included (1) a control condition, (2) a Muslim Afghan refugee condition, (3) a Christian Ukrainian refugee condition, and (4) a climate refugee condition.Footnote 2 Table 1 lists the full wording of the questions.
Table 1 Experimental Conditions and Design
Note: Emphasis added.
We chose these treatment conditions because each group helped us to understand refugee questions currently facing decision makers in American politics. Our control condition, albeit not a pure control, was intended to help us understand respondents’ general reaction to refugees as a whole without priming any specific group. With a growing population of climate refugees and ongoing crises facing those forced out of Ukraine and Afghanistan, these questions reflect the real-world scenarios facing leaders and voters. Although these prompts do not allow us to disentangle nationality, religion, and race, and although there is not perfect information equivalence (Dafoe, Zhang, and Caughey Reference Dafoe, Zhang and Caughey2018), they mirror the factors that individuals weigh about refugee admissions, thereby adding greater verisimilitude. Respondents were instructed to read the policy statement into which they were randomized and then to rate their support for the policy. Each policy statement began with the same prompt but the end of the statement was randomized (see table 1).Footnote 3
After reading the policy statement, respondents indicated how much they agreed with the policy on a 1–4 Likert scale ranging from “strongly disagree” (1) to “strongly agree” (4). The forced migrant condition, which provided no information about the type of refugees admitted, served as our control group. We compared average support for refugee policy across conditions.
RESULTS
This section describes our aggregate results, addressing H1a, H1b, and H2. Table 2 displays the differences in means between the different conditions for all respondents, as well as among other respondent subgroups. Our outcome measure was an ordinal scale, which could not be treated necessarily as a continuous outcome. Therefore, each row in table 2 includes two test statistics: the results of ANOVA F-tests (for differences in average scores across conditions) and Chi-Square tests for differences in the distribution across conditions and response categories.
Table 2 Main Hypotheses Tests Across Respondent Subsamples
Notes: The table includes comparisons across conditions and within different subsamples of our respondents, using both ANOVA (F statistics) and Chi-Square analysis.
Several points are noteworthy. First, when we examined the entire respondent pool, significant differences (p<0.001) emerged between those willing to admit those refugees in the climate condition (𝜇=2.32) and those in the other three conditions (i.e., Uk𝜇=2.79, Af=2.50, and Gen𝜇=2.48), which confirms H1a and H1b. This pattern persists for the three other subgroups that were examined—Republicans, whites, and Christians—whereas significant differences did not emerge for Democrats, non-whites, and non-Christians. Moreover, among each subgroup examined, Democrats were most supportive of climate migrants and the three other treatments that focused on different groups of migrants (i.e., Cl𝜇=2.88, Uk=3.03, Af=3.05, and Gen=3.01), whereas Republicans were the least supportive across all types of forced migrants (i.e., Cl𝜇=1.73, Uk𝜇=2.61, Af𝜇=1.89, and Gen𝜇=1.85).
Next, we visually represented how climate migrants fare relative to each of the other three disaggregated treatment conditions. Figure 1 plots the means for each experimental condition. On a scale from 1 to 4, respondents averaged 2.48 in response to the admission question after randomization into treatment across all conditions. Categorized by treatment, the means support for admitting those in the general (control), Ukraine, Afghanistan, and climate conditions were 2.48, 2.79, 2.50, and 2.32, respectively. As illustrated in figure 1 and as pairwise T-tests confirmed, significant differences emerged between the climate condition and the Ukraine condition (p<0.0001), which confirms H2.
Figure 1 Average Support for Refugee Admission, Full Sample
Plot includes group means for each experimental condition and 95% Confidence Intervals. The dotted horizontal line represents the grand mean in average support for admission (2.48 on the four-point scale).
Moreover, substantive differences persist between the climate condition and the general treatment (control) (p=0.10) in line with H1b, as well as between the climate condition and the Afghanistan condition (p<0.10). In other words, respondents were less likely to support migrant increases when they were climate induced compared to each of the other treatment conditions and were most likely to support migrants when they were Christians from Ukraine. There is a clear advantage for those who are seeking refuge from the conflict in Ukraine and comparative reticence to admit climate-induced migrants, in line with the Arias and Blair (Reference Arias and Blair2022) study. Furthermore, support for the Afghanistan condition is indistinguishable from the control condition, although it fell significantly below the Ukraine condition (p<0.001). This means that our sample reacted similarly whether respondents thought of refugees in general or specifically Muslim refugees from Afghanistan.
Finally, we evaluated whether respondents’ personal characteristics moderated their preferences for the four treatments. Results of these tests for heterogeneous effects are reported in figure 2. Overall, among the subgroups examined, Republicans (𝜇=2.61 versus 𝜇=1.72, p<0.0001), white Americans (𝜇=2.82 versus 𝜇=2.23, p<0.0001), and Christians (𝜇=2.74 versus 𝜇=2.17, p<0.0001) were significantly more likely to prefer increasing admissions in the Ukraine condition than in the climate condition, which confirms our three subgroup hypotheses H3a, H3b, and H3c.Footnote 4
Figure 2 Average Support for Refugee Admission, Key Sample Subgroups
Plot includes group means for each experimental condition and 95% Confidence Intervals. The dotted horizontal line represents the grand mean in average support for admission (2.48 on the four-point scale).
Delving deeper into the partisan findings, the top two panels in figure 2 depict group means for each experimental condition with 95% confidence intervals among Democrats (left) and Republicans (right). First, across the board, Democrats randomized into all four treatments were significantly more likely to support refugee admissions, regardless of the experimental condition. Second, Republicans generally opposed refugee admissions; however, when we reviewed each condition, we observed distinctions by refugee type. Namely, Republicans were significantly more likely to support admitting Ukrainian refugees relative to the other conditions. In fact, only in this condition were Republicans as likely as not to support increasing refugees.
The second row in figure 2 examines differences by racial group. Due to relatively small numbers of non-white people in our sample, we examined whites and non-whites separately rather than in a more granular cross-racial comparison. In particular, we examined whether whites were more likely to make distinctions for white refugees compared to other groups. In fact, we did find this pattern. Among non-whites, there were no significant differences in refugee support across experimental conditions. However, among white Americans, reading about white Christian refugees made them more likely to support expansive refugee policy. Although they were not statistically different from the control condition, white Americans also were less supportive of climate-induced migrants.
The third row in figure 2 explores differences in refugee support among Christians and non-Christians. Although our treatments also explicitly invoked Muslims, the sample size prevented us from examining this group separately. Among the Christians in our sample, there was a clear difference in support for white Christian refugees from Ukraine compared to all other conditions. By contrast, non-Christian respondents appeared to be more supportive of admitting additional refugees in general. Moreover, there were no statistically significant differences in support across experimental conditions.
DISCUSSION AND CONCLUSION
Because climate change forces people from their homes and home countries, it creates tremendous challenges. For refugees, acceptance in a new country can lessen the trauma of displacement. However, when Americans think about refugees, climate migrants do not receive similar levels of support. Results from our survey experiment indicate that not all migrants experience the same type of welcome. Average support was marginally lower for those displaced by climate change. In contrast, respondents expressed significantly higher support, on average, for Christian Ukrainian refugees. Our findings underscore that both the group and the context shape public-policy attitudes. Despite the growing challenge of vulnerable populations displaced by climate change, it appears that the public is less supportive of their refuge than of those individuals who are fleeing violence. Additionally, the American public is more likely to support refugee admissions when those migrants are Christian and Ukrainian, compared to Muslim refugees from Afghanistan and climate migrants. This is particularly true for respondents who are Republican, white, and Christian. This finding reinforces the idea that the public can be galvanized in support of some refugees but that this support is contingent on context.
…the American public is more likely to support refugee admissions when those migrants are Christian and Ukrainian, compared to Muslim refugees from Afghanistan and climate migrants. This is particularly true for respondents who are Republican, white, and Christian.
Our study is not without limitations. First, given our small sample size, our subgroup findings are suggestive and should be evaluated in other studies. Future work should evaluate how specific racial and ethnic groups—particularly those with large subsets of immigrants—support refugee admissions due to various push factors. Second, when we conducted our study, the crisis in Ukraine ushered in significant urgency in the national media and discourse, whereas the US military exit from Afghanistan had largely faded from public view. Thus, timing may have shaped our results. However, it is worth noting that interest in Afghanistan was simply lower than Ukraine, as illustrated by the Google Trend results in figure 3. Certainly, American interest in Afghanistan was high in Summer 2021 after the US exit, but it lagged somewhat in comparison to its interest in Ukraine at the beginning of the Russian invasion in February 2022.
Figure 3 Trends in Google Search Interest for Refugee Countries of Origin, 2021–2022
Reports of Google Trends normed search interest for “Afghanistan” and “Ukraine” for one year between July 2021 and July 2022.
Third, we cannot address specifically why these different populations provoke different reactions, and we encourage future research to more explicitly investigate notions of refugee deservingness. A weakness of our experimental design is that we bundled treatments in two of our conditions with race, religion, and push factors. This did not allow us to assess the independent influence of these factors. Therefore, although this did not provide a “clean” way to assess the independent influence of race, religion, and push factors on support for refugees, it did provide verisimilitude. Even simply mentioning a country associated with one of these conflicts would provoke connections among race, religion, nation, and conflict. Although precisely isolating the effects of each part of the treatment is an important research goal, it is not the only one.
Fourth, it may be that because our climate condition provided no information on the religion or nationality of the migrants, there may be less equivalence across treatment conditions. Because the effects of climate change will not be felt evenly (Bathiany et al. Reference Bathiany, Dakos, Scheffer and Lenton2018), future work would be well served to examine how identity, nationalism, and religiosity moderate these relationships.
In recent years, the American public has been asked to address humanitarian challenges by supporting the relocation of those displaced from conflict. In coming years, the effects of climate change on vulnerable populations will require similar humanitarianism. However, our results demonstrate that, at present, when the American public is faced with these two humanitarian challenges, it is ready to rise to the challenge when Christian refugees are fleeing conflict in Ukraine but are less willing to admit climate-induced migrants. Given the anticipated growth of this population, our results suggest that building public support to incorporate these refugees is a pressing need.
Supplementary material
To view supplementary material for this article, please visit http://doi.org/10.1017/S1049096524000398.
ACKNOWLEDGMENTS
The research in this article was funded by PRRI, a nonpartisan, non-profit research institute and was developed and conducted when the coauthors served as PRRI Public Fellows in 2021–2022. The authors thank PRRI for funding this research and facilitating this collaboration. The authors also are grateful to five anonymous reviewers and the editors for their helpful suggestions about this article.
DATA AVAILABILITY STATEMENT
Research documentation and data that support the findings of this study are openly available at the PS: Political Science & Politics Harvard Dataverse at https://doi.org/10.7910/DVN/USRTNJ.
CONFLICTS OF INTEREST
The authors declare that there are no conflicts of interest or ethical issues with this research.
Open access
