Democratic backsliding damages favorable US image among the global public

Abstract

US democracy appears to have weakened during the 21st century. Scholars have raised concerns that this democratic backsliding will reduce favorable views of the United States among foreign citizens in other democracies. In turn, observers predict that the eroded global image of the United States will undercut its ability to win foreign policy cooperation from international partners. We assess these views using three multinational survey experiments fielded in 12 countries with 11,810 respondents. The results show that information about US democratic backsliding indeed decreases respondents’ favorability toward the United States. However, in our exploratory analysis, we find little evidence that it decreases support for cooperating with the United States. While America’s global image may suffer from international reporting focused on the degradation of its longstanding democratic system, its ability to garner support for critical policies seems resilient in some important partner countries.

Significance Statement.

This article examines whether signs of democratic backsliding affect the US global image. Scholars and policymakers raise concerns that democratic backsliding causes foreign citizens to view the United States less favorably. Additionally, backsliding may undercut United States ability to win foreign policy cooperation. Our multinational survey experiments show that information about US democratic backsliding decreases favorability toward the United States. However, we find little evidence that this information decreases support for cooperation with the United States. It appears that America’s ability to garner support for critical policies is resilient to perceptions of diminished democracy in some countries.

Introduction

Does US democratic backsliding hurt its international image? Cultivating a positive international image is important to US foreign policy strategy. “Soft power” promotes cooperation via attraction—favorable attitudes toward the United States among ordinary citizens abroad (1, 2). The United States’ longstanding commitment to democracy constitutes one crucial element of this multifaceted power to attract (2).

That asset is now at risk: Democratic backsliding may diminish America’s soft power resources. Scholars report that recent years have revealed cracks in the foundations of US democracy (e.g. (3)) and investigate the causes and consequences of American democratic backsliding from theoretical and empirical perspectives (e.g. (4–7)). Policies and events that undercut American democracy have garnered significant attention not only within the United States but also globally, drawing both censure and concern from foreign leaders and media (8). Such backlash might signal that foreign citizens are pulling away from the United States as its claim to global democratic leadership fades. Consequently, soft power weakness may carry significant implications for US security interests. A diminished US image may hinder its capacity to respond to contemporary challenges that rely on cooperation from allies and partners—from global pandemics to climate change, the war in Ukraine, and beyond.

We examine the empirical support for these concerns using three preregistered multinational experiments in 12 US economic and security partners in regions that are the focus of international tensions—Europe and Asia, and older and newer democracies that can be expected to share the democratic values that the United States has traditionally espoused. Specifically, with 11,810 total respondents, we randomly assigned information about American democratic backsliding to some respondents before measuring views about the United States. The first study collected responses from nine European and/or Anglosphere democracies: Australia, Canada, France, Germany, the Netherlands, New Zealand, Norway, Sweden, and the United Kingdom. We fielded the second study in New Zealand and the third in India, Japan, and South Korea. These additional studies replicate key findings from the first and further examine alternative explanations. Importantly, they also include exploratory analyses that probe US backsliding’s effects on public support for specific cooperative policies that the United States is pursuing in each country.

The results confirm what many suspect (9): Democratic erosion undermines the US image around the globe. However, in the exploratory analysis in studies 2 and 3, we find little evidence that the diminished American image corresponds with decreased public support for cooperation with the United States. This latter finding holds across four countries and policies that range from prioritizing economic relations with the United States over China, strengthening the Quadrilateral Security Dialogue and curtailing arms sales to Russia. This strongly suggests that American democratic backsliding does not shift public support for cooperating with United States interests within these democratic partners.

The project makes three contributions to research about the link between democratic backsliding and foreign public opinion. First, it provides clear, causally identified evidence that information about democratic backsliding damages the US image among democratic partners. Prior work on soft power implies or assumes such a causal relationship (10, 11).^a Our work provides a robust empirical foundation for these claims.

Second, our findings illustrate an additional dimension of democracy’s role in the global power system. Existing studies show that democratic institutions strengthen a state’s “hard power” through several avenues, including stronger civil-military relations, increased human capital, and greater motivation during military conflict (12–15). Democratic institutions also strengthen a state’s bargaining power, ameliorate information problems in negotiations, and incentivize states to enter conflicts that they are likely to win (16–18). Our project considers how US domestic institutions shape foreign public opinion, a complementary route to generating and maintaining the US’s global influence.

Third, we find limitations in the conditions under which information that harms the US image also weakens soft power, the ability to influence foreign policy outcomes. In exploratory analysis, we examine foreign public views about policy debates that implicate both US interests and their partner countries’ interests. We find that although backsliding dampens the American image, the same information about backsliding does not decrease support for cooperation with the United States. This result suggests that some US foreign policy priorities can withstand backlash against domestic democratic backsliding. It also raises questions about the extent to which soft power provides essential leverage for promoting US interests (1, 19).

Democratic backsliding and foreign public opinion

Nye (20, x) defines soft power as “the ability to get what you want through attraction rather than coercion or payments.” States expend considerable effort to enhance this ability by generating soft-power resources. They advertise their country’s positive attributes through many activities, such as cultural exchange programs and high-level visits. By investing in foreign public goodwill and favorability, states can shift international outcomes in desirable directions (1).

Among various “currencies” of US soft power, Nye argues that democracy is among the “deeply seductive” values that “can be powerful sources of attraction” (19, p. x, 15). In other words, the US reputation as a strong democracy contributes to its positive image. Therefore, we expect that democratic decline damages favorable views toward the United States. We consider three reasons for this effect.

First, the US centers democracy as a key “value” in the identity it projects to the world. Democracy is a defining feature of its domestic institutions and a priority in foreign policy. The majority of citizens in other democratic countries admire the American idea of democracy, separate from beliefs about American values in general (20, 56). Global citizens favor states that adhere to liberal values, even when those states have nondemocratic regimes (10). Accordingly, a logically consistent expectation is that if democratic decline weakens the American claim to core liberal values, the United States will lose some of its international allure.

Second, historically resilient and effective government institutions strengthen the US image. Stable institutions can serve as a soft-power resource by projecting an image of competence, success, and strength. Controversial laws that could fundamentally undermine free and fair elections, growing distrust in the legitimacy of those elections, and violent opposition to the outcome of elections could damage global public confidence in these historically reliable political institutions.

Third, promoting democracy worldwide has been central to American foreign policy and public diplomacy. However, this significant soft power asset can weaken if the foreign public views the US’s democracy promotion as insincere or hypocritical when US democracy itself declines (11, 20).

In sum, liberal and democratic values, stable institutions, and sincere democracy promotion increase US attractiveness to global audiences. Policies and events that undermine any of these elements should, in turn, degrade the US image abroad. Therefore, we expect the following:

• Hypothesis:  Exposure to information about American democratic backsliding deteriorates the positive image of the United States among citizens in foreign states.

This hypothesis does not imply that the global public’s evaluations of the United States are damaged in all respects. It is critical to distinguish between the US image and its translation into soft power through foreign public support for international outcomes which the US pursues. The theory of soft power suggests that efforts to improve the state’s image abroad will also shift public opinion toward more robust support (or lower opposition) for cooperating with an attractive foreign power (1, 20). However, the scope and validity of this theory is a matter of debate, including criticism that the theory fails to consider important conditioning factors such as the connections to hard power or the role of competing narratives (21, 22). Events that damage favorability, such as democratic decline, may or may not affect foreign public support for cooperating with the United States on contemporary foreign policy issues. This empirical question has not been systematically examined in the existing literature. We thus explore a general conjecture that backsliding will reduce support for cooperation.

• Conjecture:  Exposure to information about American democratic backsliding decreases support for policies aimed to increase cooperation with the United States

This conjecture specifies that we will examine whether American democratic backsliding has a direct effect on change in support for cooperative foreign policies. We do not examine whether any observed change in favorability mediates this relationship, a question beyond the scope of our experimental design. Further, this conjecture is exploratory and is not a preregistered hypothesis.

Research design

Our empirical inquiry includes three studies.^b In Section A, we first describe the design of study 1. Then, in Section B, we explain the rationale and additional features of studies 2 and 3.

Study 1

We fielded study 1 on 7,295 total respondents in nine countries during the spring of 2023. Our sample includes adult residents in Australia, Canada, France, Germany, the Netherlands, New Zealand, Norway, Sweden, and the United Kingdom. All countries are liberal democracies, making them essential cases for testing our argument: If US democratic decline has any implications for foreign public opinion, we should observe them most strongly among the citizens in advanced Western democracies sharing governance norms with the United States (10).^c

We randomly treated half the participants with an article about US democratic backsliding. All participants then answered a question about their favorability toward the United States and completed other posttreatment questions.

Treated respondents received a summary of a hypothetical report titled “New Report: Democracy is Declining in the United States.” Figure 1 displays the complete treatment text. The treatment describes US events and policies linked to democratic decline in bullet point statements. It uses language and information from actual news items and published reports, spliced together and edited to meet the study’s goals.^d

Fig. 1.

“Backsliding” treatment.

The treatment text includes several features designed to maximize participant comprehension of the core message that US democracy is declining. First, the bold headline conveys the central theme—akin to summaries that participants might peruse browsing the internet or skimming a newspaper. Second, a summary statement below the headline repeats that theme and introduces the term “backsliding” to describe the state of US democracy. Third, the text presents detailed information about critical events in a list format. We expect this presentation style to increase respondents’ ability to comprehend key details in the treatment text. This form of information stimulus also has ecological validity: reports published by think tanks, governments, and other organizations often feature an “executive summary” similar to our treatment.

After reading the text, treatment group participants completed a question asking them to select all items that the report considers “a threat to democracy in the United States.” The question has six answer options: Five statements repeat a theme or fact from the report and constitute correct responses; they can also select “none of the above.” This “treatment booster” question aims to reinforce and strengthen the information treatment. Therefore, the treatment effect estimates in our analyses reflect the combined effect of asking respondents to read the hypothetical report summary and encouraging them to carefully consider the report’s contents via the supplemental question.

The “compound” nature of our treatment is intentional. We wanted to provide holistic information about the multidimensional US democratic decline and raise the salience of US backsliding. This approach captures the critical concept that we seek to manipulate: Democratic backsliding occurs via multiple, simultaneous channels. It entails widespread, incremental changes that reduce electoral competition and participation (7, p. 95).

We measure favorability, our primary dependent variable, using a well-validated, direct question common in soft-power research (23). Participants reported whether they view the United States favorably or unfavorably on a 4-point scale ranging from “very unfavorable” to “very favorable.”

A posttreatment questionnaire examines the mechanisms underlying our hypothesis. First, we asked respondents to rate whether the United States is a full democracy using a 100-point scale (24). American democratic backsliding may also diminish global trust in democracy itself, with implications for the positive image of the United States. Therefore, we also asked whether participants agreed or disagreed that “Democracy is preferable to any other kind of government.”

Studies 2 and 3

We fielded studies 2 and 3 to 1,501 respondents in July 2023 and 3,014 in February 2024 to answer two additional questions. First, is the backsliding treatment effect merely based on negative information about the United States? We assess this possibility with another treatment containing damaging information about the US economy. Second, does democratic backsliding place US foreign policy priorities at risk? To examine this, we added questions asking whether American backsliding reduces support for cooperating with the United States on salient foreign policy issues.

We chose New Zealand for study 2 for two reasons. First, we observed the most prominent effect from study 1 in this country. Study 2 aims to probe this causal effect further, thus requiring a sample where we anticipate confirming Hypothesis 1. Second, New Zealand’s geographic location and economic and security situations make the population a vital target for US soft power. Indeed, China is by far its largest trading partner.^e US policymakers hope to persuade New Zealanders that joining AUKUS—Australia, the United Kingdom, and the United States’s Trilateral Security Partnership—is a crucial opportunity to address growing and common security concerns stemming from the rise of China as a military superpower in the Indo-Pacific region. However, political leaders in New Zealand have resisted over concerns about needlessly provoking China.

Studies 1 and 2 focused on geographically and/or culturally “Western” states, limiting the inferences we can draw about how democratic backsliding affects US influence in other key global regions. Fully understanding how American democracy is perceived among citizens of US partners requires samples from different regions that are important to US interests.

For study 3, therefore, we selected three countries from the Indo-Pacific region—Japan, India, and South Korea. They are democracies, critical US partners or allies, and vital for US security interests. The United States has bilateral defense treaties with Japan and South Korea, and both are stable democracies. India is the world’s largest democracy and part of the Quadrilateral Security Dialogue alongside the United States, Japan, and Australia. Each country is important for US capacity to meet future security challenges. These samples significantly expand the scope of our inferences about how democratic backsliding affects US influence outside European and/or Anglosphere countries.

The survey content and structure are the same in studies 2 and 3 except for country-specific policy questions. Respondents were randomly divided into two treatment groups and one control group. Respondents in the treatment groups received either the backsliding treatment, identical to that used in Study 1, or the economy treatment, titled “New Report: The Economy is Declining in the United States” (Fig. 2). The economy treatment aims to test whether general negative information reduces US favorability. It closely matches the structure, length, and tone of the backsliding report. Both treatments contain only factual information, and both highlight a dire circumstance characterized by decline and with no indication that things will improve. If the economy treatment does not affect foreign public opinion about the United States, we can be more confident that any effect of the backsliding treatment pertains to democracy rather than negativity. Control group respondents proceeded directly to the posttreatment questionnaire.

Fig. 2.

“Economy” treatment.

After the treatment, all participants received the US favorability question as in study 1. They then answered three foreign policy questions. Each question taps into practical foreign policy consequences that might follow declining US soft power resources. The first question asked whether the respondent’s country should prioritize economic relations with the US or China. We tailored two other questions to each partner country’s specific US policy priorities. In New Zealand, respondents indicated (i) support for increasing military aid to Ukraine and (ii) joining AUKUS. Respondents in India reported (i) their views about reducing arms imports from Russia and (ii) strengthening the state’s commitment to the Quadrilateral Security Dialogue (the Quad), including the United States, Australia, India, and Japan. Japanese respondents also received questions about (i) strengthening Japan’s commitment to the Quad and (ii) their support for Japan to make a public statement committing to defend Taiwan from any Chinese aggression. In Korea, we asked about (i) support for joining the Quad and (ii) support for the Chip 4 Alliance, which aims to reduce China’s position in the global semiconductor supply chain.

Finally, three posttreatment questions asked about the stability of the US government, whether the United States is getting stronger, and agreement that “other countries should be like the United States” regarding values. These questions help us explore the mechanisms that drive the backsliding treatment effect that we expect. Furthermore, we included two “placebo” questions measuring respondents’ beliefs about the quality of the US military and US universities, which are not directly associated with the strength of American democracy. Therefore, we do not expect the backsliding treatment to degrade evaluations of these institutions.

Results

Backsliding reduces favorability toward the United States

We evaluate the hypothesis by comparing US favorability ratings between the treatment and control groups. The results using the pooled sample from all three studies are presented in Fig. 3, which provides clear evidence that democratic backsliding hurts foreign public opinion about the United States. The left panel of Fig. 3 shows that the distribution of responses shifts “left” (less favorable) among respondents in the treatment group compared to those in the control group. The right panel presents each group’s average favorability rating and its 95% CI. It shows that the treatment causes a statistically significant decline in American favorability, decreasing the rating by 0.095 units on the 4-point scale ($P<0.01$). This effect size is about 41.9% of the SD of aggregated favorability scores across the countries in our sample, representing a moderately sized effect.

Fig. 3.

Effect of American democratic backsliding treatment on favorability toward the United States.

The three studies use different countries and survey contexts. Yet, we find strikingly similar results. We observe a statistically significant effect of $-0.095$ units in the sample of nine European and Anglosphere countries from study 1. The findings from studies 2 and 3 are similar—a significant effect of $-0.096$ units. These consistent results increase our confidence that democratic backsliding has a robust causal effect on US favorability.

The results of other tests further support our conclusion.^f First, we find that the backsliding treatment does not significantly shift respondents’ answers to those placebo questions about US institutions that are not directly related to the quality of American democracy (Fig. S4 in Section D of the Supporting Information). These null effects mitigate concerns that the backsliding treatment simply creates a broad negative halo around the United States.

We next examine the effect of the economy treatment in studies 2 and 3 (Figs. S5–S7 in Section D of the Supporting Information). We find that information about the decline in the American economy does not reduce favorability toward the United States. The effect is close to zero and not statistically significant ($b=-0.03$, $P>0.10$).^g This finding supports our expectation that the substance of the backsliding treatment causes reduced favorability toward the United States.^h

Finally, Figs. 4 and 5 display the effects of each treatment on a panel of auxiliary posttreatment questions. We rescaled each outcome to range from 0 to 1 when calculating the average treatment effects. The backsliding treatment decreases participants’ rating of US democracy, their estimation that the US government is stable, and their estimation that the United States is getting stronger. These results are consistent with our expectations and suggest that our treatment affects US favorability at least partly via views about US democracy, its stability, and its strength.ⁱ The negative effect on democracy ratings is 42.0% of the SD. We observe smaller effects on perceptions of stability (18.9%) and US strength (19.7%). We also note that, as expected, the economy treatment does not affect participants’ rating of US democracy nor their estimation that the US government is stable. However, it does have a negative effect on perceptions that the United States is getting stronger. This effect is of similar size to the effect of the backsliding treatment, 20.0% of 1 SD of the aggregated means on this variable. This indicates that the economy treatment does move opinion about the United States and can cause the same movement as the backsliding treatment.

Fig. 4.

Effect of American democratic backsliding treatment on additional outcomes.

Fig. 5.

Effect of American economic decline treatment on additional outcomes.

We provide results from a posttreatment factual manipulation check in the Materials and methods section below, and additional analyses in Sections C and D of the Supporting Information.

Backsliding does not shift (most) opinion on policy

Although favorable US image is damaged, we find no evidence that the backsliding treatment reduces public support for policies that entail cooperation with the United States. Figure 6 shows the estimated average treatment effects for the policy-related questions. Across this range of policies and geographic locations, in most cases, American backsliding does not change public support for US-connected foreign policies.^j Specifically, the backsliding treatment does not shape support for foreign economic policies in these four countries in the Indo-Pacific region. There is no shift in respondents’ preference for prioritizing economic relations with China over those with the United States. Similarly, there is no decline in Koreans’ support for the Chip 4 alliance. Neither is there a shift in support for almost all policies addressing security issues. American democratic backsliding does not decrease Indian, Japanese, or Korean respondents’ support for commitment to (or joining) the Quadrilateral Security Dialogue. It does not decrease responsiveness to American calls for aid to Ukraine in India, Japan, or New Zealand, nor make Japanese respondents less supportive of defending Taiwan. These findings soften concerns that American democratic backsliding weakens existing US security partnerships.

Fig. 6.

Effect of American democratic backsliding treatment on support for policy.

Characteristics of our survey design and large sample increase our confidence that these null results (although exploratory) reflect the absence of effects in the population of interest. Our survey omits respondents who did not pass an attention check, increasing our confidence that our respondents are reasonably attentive. The significant effect of the backsliding treatment on ratings of US democracy indicates that our information treatment does, indeed, move the respondents’ opinion in a way that we expect theoretically. Therefore, we can conclude that the respondents were not pretreated in a way that would prevent the treatment from moving any opinions of their general attitude (favorability) toward the US change.

In spite of this, their specific policy attitudes do not move. Furthermore, the wide range of policy questions we included in our studies indicates that an unusual issue area does not drive this null. We similarly observe a null effect when pooling all policy-relevant questions. We estimate a mixed-effect model with the backsliding treatment, country-fixed effects, and question-random effects. The coefficient on the backsliding treatment is $-$0.008, which is almost precisely zero and not statistically significant. Baseline support for these policies is high, ranging from 0.481 to 0.843, making floor effects unlikely. Still, our exploratory analyses include a small number of policy questions in four countries. It is possible that there are substantive effects not captured by our studies, which calls for additional research.

Conclusion

There is a growing concern that the US faces increased challenges to free, fair elections, and the peaceful transfer of power. Through survey experiments in 12 democracies, we provide causal evidence that information about the US failing to maintain the integrity of its internal democratic system has global consequences for the country. Namely, international news reports about democratic backsliding weaken US attractiveness among citizens of economic and security partners.

However, evidence from India, Japan, New Zealand, and South Korea suggests that this same information does not reduce policy support in crucial partner states. The null results for policy attitudes raise essential questions about the connection between the US global image and foreign public support for cooperation with the United States. While a fundamental expectation from the theory of soft power is that the same factors that deteriorate the US image will decrease support for cooperation with the United States, US influence seems sufficiently robust to withstand some of the alleged blowback from its democratic backsliding. Some evidence shows that international opinion matters for US foreign policy goals (1, 25, 26), and research should examine whether the null results we report hold among the Western democracies that the United States has long relied on for security cooperation. Our findings nevertheless raise questions about the conditions under which opinion changes indicate less soft power in terms of influence over outcomes.

Materials and methods

This study received approval from the Institutional Review Board (IRB) at Dartmouth College, CPHS# STUDY00032562. All participants provided informed consent.

We fielded study 1 via Qualtrics panels. We fielded studies 2 and 3 via PureSpectrum. This change in the survey platform was driven primarily by the relative cost of administering surveys. We implemented three tools to maintain and evaluate the high quality of samples in each study. First, we implemented demographic quotas matched to key characteristics from each country to collect a diverse range of participants. In study 1, we set national quotas for respondents’ age, sex, and region within each country. In study 2, we targeted 50% men and 50% women. We did not include other quotas due to the difficulty of recruiting large survey samples in New Zealand. For study 3, we included quotas for age and sex in all three countries. For age, we used three categories: 18–34, 35–54, and 55 or older. Study 3 also included education quotas in India and Japan (but not South Korea), targeting people with or without a university degree in proportion to each country’s national education attainment.

Second, we used several pretreatment filters. Specifically, we terminated the survey and excluded responses if they met any of the following conditions: (i) A GeoIP filter identified a different location from the target country; (ii) Any Qualtrics fraud detection tools indicated that a response was likely a bot or duplicate respondent; (iii) A respondent incorrectly answered 1 of 2 simple attention checks (study 1) or a single attention check (studies 2 and 3); (iv) A respondent selected a language that is not widely used in the target country; and (v) In study 1 only, the respondent was identified as a speeder. This condition eliminated respondents who completed the entire survey—which included other, unrelated studies—within 10 min. Studies 2 and 3 did not exclude respondents based on their completion time. Criteria a–d are preregistered, whereas we deviated from our plan to eliminate speeders in study 1 based on quality assurance consultations with the panel provider.

Third, we included a closed-ended factual manipulation check at the end of the survey. In study 1, we asked “During the survey, you may have been given a summary of a report about the United States. Which of the following was the headline of the article you read?” Answer options included: “New Report: Democracy is Declining in the United States”/“New Report: Human Trafficking Persists in the Global Community”/“New Report: Americans of Color have been hit hardest by the Covid-19 Pandemic”/“New Report: The US/is more Economically Divided than other Wealthy Democracies”/None of these. I did not read an article./I don’t remember.

In study 1, we find high rates of correct responses to the manipulation check in the treatment (59.0%) and control (65.8%) groups. In study 2, we find similar rates of correct responses in the backsliding group (49.9%), economy (53.7%), and control (54.0%) groups. In study 3, the proportion of correct responses was relatively lower in the backsliding (47.4%), economy (45.3%), and control (37.2%) groups.

These results suggest that study participants were reasonably attentive, though less so in study 3. The presence of inattentive respondents adds noise to our outcome questions, thereby attenuating any observed treatment effect. Our estimates are conservative: Without the measurement error arising from this noise, the effects could be larger.

Fourth, we include a subjective manipulation check to evaluate whether the treatment affected participant perceptions of the quality of US democracy. The question asked: “On a scale from 0 to 100, where 0 is least democratic and 100 is most democratic, how would you rate the political system of the United States as of today?” Figure 4 shows that participants rate the United States significantly less democratic ($-0.022$, $P<0.001$) when they receive the backsliding treatment compared to the control. Figure 5 shows that the economy treatment does not affect views about democracy ($-0.011$, $P>0.1$). These results increase our confidence that participants received the correct treatment information and that the treatment was sufficiently strong to change perceptions about the targeted concept.

See the Supplementary Information for a detailed description of all materials and methods used in this study. They include our preregistered analysis plan, survey design, and additional robustness analyses.

Supplementary Material

Supplementary material is available at PNAS Nexus online.

Funding

We gratefully acknowledge funding from Dartmouth College and the Dickey Center for International Understanding.

Author Contributions

B.E.G., Y.H., K.M., and K.E.P. designed research and wrote the article; Y.H. analyzed and visualized data.

Data Availability

Anonymized data, complete replication code, and documentation necessary to produce the analysis and figures presented in this article are available in the Harvard Dataverse: https://doi.org/10.7910/DVN/M88SXU.