3.1. Credibility of the WHO

According to the literature on attitude change or persuasion in social psychology, information source, a speaker delivering a message, is considered one of the key components of persuasive communication (Pornpitakpan, Reference Pornpitakpan2004; Brinol and Petty, Reference Brinol and Petty2009). There are many studies that show information sources can influence how people process information and thus how likely they are to change their beliefs or attitudes. It serves as a cue (Briñol et al., Reference Briñol, Petty and Tormala2004), influences the confidence of receivers (Petty et al., Reference Petty, Brinol and Tormala2002), affects the amount of thinking (DeBono and Harnish, Reference DeBono and Harnish1988), or biases the direction of reasoning (Chaiken and Maheswaran, Reference Chaiken and Maheswaran1994).

Scholars particularly emphasize the concept of source credibility. Early social psychology studies by Hovland and Weiss (Reference Hovland and Weiss1951) and Hovland et al. (Reference Hovland, Janis and Kelley1953) demonstrate that more credible sources produced greater attitude change than less credible sources. In general, the effect of source credibility on persuasiveness is widely considered as a well-established finding (Petty et al., Reference Petty, Brinol and Tormala2002). Source credibility is a multi-dimensional concept commonly believed to consist of expertise and trustworthiness. Expertise refers to how much a source is seen as capable of making accurate and valid statements, while trustworthiness refers to the extent to which an audience believes the source’s willingness to deliver accurate and valid information (Hovland et al., Reference Hovland, Janis and Kelley1953; Pornpitakpan, Reference Pornpitakpan2004). Perceived expertise and trustworthiness contribute to source credibility. Existing studies show that regardless of whether the emphasis is on expertise or trustworthiness, high-credibility sources are more persuasive than low-credibility sources.

This implies that IOs can shape citizens’ attitudes towards certain issues when they are perceived as more credible, and thus the credibility of IOs can influence their capacity to craft and promulgate new ideas and norms. It is particularly important for IOs because, unlike states, they typically lack coercive power (Franck, Reference Franck1990; Hurd, Reference Hurd1999). The importance of credibility to shape individuals’ attitudes applies to the WHO. The WHO, as the foremost IO for public health, is a specialized agency of the United Nations that acts as ‘the directing and coordinating authority on international health work’ (World Health Organization, 1946). Like many other IOs, it is mainly charged with solving coordination problems by providing knowledge and information, in this case, on global health issues (Snidal, Reference Snidal1985). Globalization has increased states’ vulnerability to global health threats. Faced with common challenges regarding global health concerns, states delegate authority to the WHO to coordinate international responses that promote health and prevent the spread of diseases. Although the WHO has no legal authority to enforce its policies, it does have broad legal authority to serve as a forum for international conventions and agreements that address global public health issues (Meier et al., Reference Meier, Taylor, Eccleston-Turner, Habibi, Sekalala and Gostin2020). It is also able to develop and codify non-binding recommendations and regulations. The organization seeks to identify and disseminate information about the most effective ways to address global health concerns. It has also been at the forefront of global health emergencies, including global pandemics such as SARS and COVID-19, natural disasters, and humanitarian crises.

Importantly, the WHO maintains a reputation for scientific and technical expertise in health issues. It aims to provide guidelines and recommendations based on the best available scientific evidence, and these help states coordinate to resolve global health problems (Benvenisti, Reference Benvenisti2020). The organization is composed of more than 8,000 professionals, including doctors, epidemiologists, and scientists (World Health Organization, 2025). It also operates a team of experts in such areas as economics, statistics, and emergency relief. Thanks to expertise owned by the WHO, we expect that the WHO as an information source commands credibility, which will boost people’s willingness to trust the guidelines and recommendations it issues. The WHO might be especially effective under uncertain information environment during a health crisis, like the one during the COVID-19 pandemic.

In addition to its expertise in global health, there are other aspects of the WHO that make it a trustworthy information source. First, the WHO’s authority to coordinate and lead international responses to global health concerns in the collective interest confers its trustworthiness. The WHO has played important roles in coordinating ground-level emergency operations, often working with non-government organizations, at the forefront of global health emergencies, and this role as a first responder to health emergencies contributes to its trustworthiness. Second, its scientific knowledge can contribute to its trustworthiness. The WHO, staffed by these scientific and bureaucratic experts, emphasizes its technical and scientific character as a non-politicized organization that can establish global standards and provide reliable advice to countries (Gruszczynski and Melillo, Reference Gruszczynski and Melillo2022). For example, the director of the WHO’s Health Emergencies programme said in an interview that ‘[t]he power that we have is the power to persuade through science, persuade through evidence, persuade by demonstrating what other countries are doing and showcasing good examples of good practice (Huang, Reference Huang2020)’.

The WHO’s credibility, grounded on its expertise and trustworthiness, may be especially important in conveying the COVID-19-related information, since there has been considerable uncertainty among the public over the causes and consequences of COVID-19, and domestic discussions of COVID-19 have been highly politicized in the U.S. and elsewhere. Compared to domestic public health authorities that could be heavily influenced by domestic politics, the information provided by the WHO could be perceived as more credible. Based on the WHO’s formal mandate and substantial technical resources, we establish a baseline expectation for source credibility effects. The WHO’s institutional design assumes that technical expertise and international authority should enhance the persuasiveness of health information relative to domestic sources.Footnote ² This reflects the classic source credibility model from social psychology, where expertise and trustworthiness are expected to increase persuasion (Hovland et al., Reference Hovland, Janis and Kelley1953).

However, this baseline expectation may not hold universally, and even when it does, the magnitude of source effects may vary significantly based on how the WHO is presented. The effectiveness of expert sources depends on how they are perceived and framed rather than their objective characteristics alone (Bayram and Shields, Reference Bayram and Shields2021). We theorize that public acceptance of WHO information will vary based on how the organization is presented to audiences. Specifically, we test whether emphasizing different aspects of the WHO’s institutional profile affects its credibility relative to domestic sources.

First, if the WHO is perceived to have high expertise and sufficient competency in addressing the COVID crisis, the WHO’s guidelines and recommendations are likely to be credible, and the WHO is able to persuade the public. The emphasis that the WHO is an organization whose authority is based on technical expertise and scientific knowledge will increase the public’s perception of its expertise and knowledge and strengthen the credibility of the WHO as an information provider by activating positive competence assessments. We thus expect people to be more likely to accept the WHO’s guidelines and recommendations when they are presented as having scientific and professional expertise.

Meanwhile, when the WHO is perceived to represent the collective will and shared interests of its many member states, it will be perceived as more trustworthy (Lupia and McCubbins, Reference Lupia and McCubbins1998; Greenhill, Reference Greenhill2020). The flip side of this is that when the public perceives it as representing a particular member’s preferences or as being influenced by a single member, it will likely be perceived as less trustworthy. Indeed, in handling the COVID-19 pandemic, the WHO has weathered many criticisms, one of which is being too lenient towards China: the WHO did not publicly condemn China’s belatedness in sharing information about the COVID-19 outbreak and praised China for its transparency and cooperation. This criticism could pose a significant threat to the WHO’s trustworthiness and thus harm its credibility (Taylor, Reference Taylor2021). Thus, when people believe that the WHO maintains close political connections with certain countries for ideological, geopolitical, or economic reasons, they will be more likely to question its credibility and less likely to accept information – recommendations and guidelines – the WHO supplies. We propose that people are less likely to accept the WHO’s guidelines and information when it is presented as partial to China because such presentation undermines perceptions of impartiality, a key component of source trustworthiness.

3.1. Modifying effects of pre-existing attitudes

We theorize that the effectiveness of WHO source credibility depends critically on recipients’ pre-existing attitudes and the thoughts they generate in response to health messages. According to the self-validation approach in social psychology (Brinol and Petty, Reference Brinol and Petty2009), source credibility can either enhance or undermine persuasion depending on whether recipients’ initial thoughts about a message are favourable or unfavourable. This suggests that the effects of WHO credibility should vary systematically across individuals with different pre-treatment attitudes towards international health authorities and COVID-19.

It is possible that many people had already developed firm attitudes about the WHO and/or the COVID-19 pandemic by the time we conducted our survey experiment in April 2022. Particularly, many studies demonstrate that partisanship plays a critical role in shaping pre-existing attitudes on highly politicized issues like the COVID-19.Footnote ³ To understand how partisanship shapes these attitudes, we draw on social identity theories of behaviour. Social identity theories of behaviour argue that individuals want to increase their self-esteem by differentiating their in-group from out-groups and favouring the in-group (Tajfel, Reference Tajfel1979). Social identity, particularly partisanship, is very influential in shaping people’ actions and beliefs.

Existing scholarship has shown that people tend to adopt the positions of co-partisan elites, particularly in highly polarized situations (Zaller, Reference Zaller1992; Druckman et al., Reference Druckman, Peterson and Slothuus2013). This is true even in non-political issues, including attitudes towards science (Blank and Shaw, Reference Blank and Shaw2015) and IO legitimacy (Dellmuth and Tallberg, Reference Dellmuth and Tallberg2020). Elite positions serve as heuristics, helping regular people determine their attitudes (Zaller, Reference Zaller1992). Beyond elite cues, another important mechanism is social norm and partisan conformity. Partisan identities as social identities come with social norms that affect political behaviour and attitudes (Pickup et al., Reference Pickup, Kimbrough and de Rooij2022). Several studies show that people care about their peers in social groups and their attitudes and behaviours are shaped by social pressures (Klar, Reference Klar2014; Suhay, Reference Suhay2015; Toff and Suhay, Reference Toff and Suhay2018). Particularly, Toff and Suhay (Reference Toff and Suhay2018) demonstrate that information about co-partisan peers’ policy preferences is just as influential as elite cues in shaping citizens’ own positions, with this peer influence being particularly strong among those with high partisan social identity. Within partisan networks, conformity pressures operate through social feedback mechanisms where individuals adjust their views to align with perceived group consensus. These dynamics can be reinforced by identity-protective cognition, where individuals unconsciously filter and interpret information in ways that protect their valued group relationships and maintain their standing within partisan communities (Kahan, Reference Kahan2013). This suggests that in the context of COVID-19 and WHO attitudes, partisans may have conformed not only to elite messaging but also to perceived norms within their partisan peer groups, potentially reinforcing resistance to WHO information among those whose co-partisans expressed scepticism. These partisan dynamics are particularly relevant for attitudes towards the WHO, as pre-existing partisan differences in support for IOs and science were already well-established. Republicans have historically been less supportive of IOs and multilateralism (Milner and Tingley, Reference Milner and Tingley2015; Smeltz et al., Reference Smeltz, Daalder, Friedhoff, Kafura and Sullivan2022) and less likely to trust science and scientists (Hamilton et al., Reference Hamilton, Hartter and Saito2015), compared to Democrats. Furthermore, since the outbreak of COVID-19 entered political conversations in the U.S. in January 2020, the pandemic has arguably been one of the most salient issues in public discourse, generating competing claims about preventive guidelines, false or misleading information, or even conspiracies about COVID-19 response efforts. The U.S. President Donald Trump continuously downplayed the threat of the pandemic and attacked the health policy community’s expertise at addressing the pandemic (Rutledge, Reference Rutledge2020). He frequently condemned the WHO’s handling of the coronavirus pandemic and accused the organization of being too deferential to China. Other Republican politicians and conservative media commentators articulated similar positions on the COVID-19 and the WHO, while Democratic politicians and liberal media commentators demanded stricter COVID-19 prevention measures and expressed support for the WHO (Allcott et al., Reference Allcott, Boxell, Conway, Gentzkow, Thaler and Yang2020; Grossman et al., Reference Grossman, Kim, Rexer and Thirumurthy2020; Clinton et al., Reference Clinton, Cohen, Lapinski and Trussler2021).

Having been exposed to a flood of information and misinformation on COVID-19 from various sources, citizens may already have formed strong pre-treatment attitudes towards public health agencies and the WHO. Indeed, existing studies have demonstrated substantial partisan differences in how people have evaluated COVID-19 policies and the public officials responsible for them (Gadarian, Reference Gadarian, Goodman and Pepinsky2022; Goldstein and Wiedemann, Reference Goldstein and Wiedemann2022; Golos et al., Reference Golos, Hopkins, Bhanot and Buttenheim2022). Bayram and Shields (Reference Bayram and Shields2021) also show that Democrats tend to trust the WHO more than Republicans.

Taken together, this discussion suggests that people are likely to have strong pre-existing attitudes towards the WHO and COVID-19-related issues and that partisanship is a main driver of these attitudes. Compared to Republicans, Democrats are more likely to support and accept the WHO’s guidelines and recommendations. Independents will likely be in the middle. Accordingly, we expect the WHO’s guidance to be less effective at influencing the attitudes of strongly partisan respondents, who already hold strong attitudes about COVID-19 and related policies and recommendations.

This raises the question of whether those without strong partisan attachments might be more receptive to WHO guidance. Independents are often viewed as moderate, politically unsophisticated and uninformed, lacking policy views and political commitments (Kinder and Kalmoe, Reference Kinder and Kalmoe2017; Freeder et al., Reference Freeder, Lenz and Turney2019; Fowler et al., Reference Fowler, Hill, Lewis, Tausanovitch, Vavreck and Warshaw2023). They tend to be less politically attentive, interested, and involved than other partisans. Thus, they are less likely to hold strong pre-existing attitudes towards COVID-related policies and the WHO’s guidance compared to Republicans and Democrats. Of course, we recognize that independents are a heterogeneous group since they include independent leaners who behave like strong partisans (Keith et al., Reference Keith, Magleby, Nelson, Orr and Westlye1992) and voters holding extreme policy views poorly described by a single ideological dimension (Broockman, Reference Broockman2016). Nevertheless, the proportion of moderate voters is indeed greater among self-identified independents than among other partisan groups (Fowler et al., Reference Fowler, Hill, Lewis, Tausanovitch, Vavreck and Warshaw2023). We expect that political independents are more likely to be receptive of the WHO’s information since they are less politically attentive and informed and are more likely to lack pre-existing policy views and commitment.

6.1. Partisan differences in pre-treatment attitudes towards public health officials and the WHO

Before testing H4 in a systematic manner, we investigated whether partisans had different baseline attitudes towards public health officials, the WHO, and COVID-19. These attitudes were measured before participants were exposed to the experimental conditions.Footnote ¹⁰ First, we examine respondents’ pre-existing beliefs regarding the handling of the COVID-19 pandemic by domestic public health officials and the WHO. Figure 2(a) shows that trust in public health officials is far higher among Democrats than among Independents or Republicans. When we asked how much respondents trusted the public health agencies on a 4-point scale of 1 (not at all) to 4 (very much), the mean value of Democrats’ responses was 2.54 (around the mid-point between a little and moderately), while the mean responses among Independents and Republicans were only 1.77 and 1.67, respectively. The results are largely consistent with what is reported in (Bayram and Shields, Reference Bayram and Shields2021). Partisan differences were also observed when assessing pre-existing beliefs regarding the effectiveness of the WHO. Figure 2(b) illustrates how effective respondents think the WHO has been at handling the COVID-19 pandemic on a 5-point scale. Democrats tend to judge the effectiveness of the WHO more positively than other partisans. The average values of Independent or Republican respondents’ evaluations of the effectiveness of the WHO are much lower, although Independents’ opinions are slightly more favourable than those of Republicans. We find a similar pattern regarding trust in scientific experts in Figure 2(c).

Figure 2.

Partisan responses to pre-treatment questions about COVID-19 and information sources.

Second, we asked a series of questions about COVID-19. To capture respondents’ pre-existing openness to the COVID-19 guidelines, we asked respondents how worried they were about the impact of the coronavirus on themselves or their families, and how safe they thought the COVID-19 vaccine was on a 4-point scale that ranged from not at all (1) to very much (4). Figure 2(d) and (e) clearly illustrate partisan differences in respondents’ attitudes toward COVID-19. Democrats were more concerned about COVID-19 and more trusting of the vaccine’s safety compared to other groups. Republicans showed less worry about COVID-19 and more distrust of the vaccine. Independents’ views fell between these two, showing moderate levels of concern and vaccine trust. Last, we examined respondents’ knowledge of COVID-19 with a true-false quiz. Specifically, we presented 10 true-false questions from the myth-busting statements on the WHO’s website.Footnote ¹¹ Figure 2(f) shows the mean number (out of 10) of respondents’ correct answers by party. We find that knowledge levels did not differ meaningfully across party affiliations. Nevertheless, Independents tended to be the least knowledgeable, while the difference from other groups’ means was quite small (around 0.3). To summarize, a descriptive exploration of respondents’ pre-existing beliefs regarding domestic public health agencies, the WHO, and COVID-19 clearly illustrates partisan differences. Both Democrats and Republicans have strong pre-existing attitudes toward COVID-19-related issues and the WHO, although their attitudes are totally different from each other. This suggests that both partisan groups can be less likely to be influenced by the WHO treatment, which explains the null findings in the main analysis. Contrarily, Independents have weaker pre-existing attitudes, being more prone to WHO-related treatment effects. Intriguingly, however, the partisan differences in COVID-19 knowledge were quite small.

6.2. Heterogeneous Treatment Effects Across Party Identification

We present the marginal effects on the average level of trust in the guidelines in Figure 3.Footnote ¹² Figure 3(a) shows that the treatment effect of general WHO condition is larger among Independents, compared to Republicans and Democrats. However, it is not statistically significant at the 5% significance level. Contrarily, Figure 3(b) shows the positive and statistically significant effects of the ‘Professional WHO’ treatment at the 5% level in the subgroup of Independents. When Independents are provided with information that the WHO is a credible source of information due to its professionalism, their guidelines become more effective. As Figure 3(c) illustrates, however, the positive effects of the WHO as the information source disappear when the respondents are informed that the WHO is politically biased: the estimated effects of the politicized WHO condition are close to zero and statistically insignificant.Footnote ¹³ The findings related to Democrats are similar to those related to Independents, although the treatment effects for them are not statistically significant in all cases and are smaller in magnitude than those for Independents. We find that the results remain similar when different DVs are used, such as the average trust in each guideline or the average behavioural intention to follow each guideline (Appendix Figures A1 and A2).Footnote ¹⁴

Figure 3.

Treatment effects on trust in the COVID-19 guidelines by party affiliation. The 90% (thicker lines) and 95% (thinner lines) confidence intervals are presented.

The results about Republicans suggest an adverse effect of the WHO as the information source. Republican respondents exposed to the general WHO condition were less likely to trust the guidelines than those in the public health officials condition. This may reflect that Republicans are more likely to question the credibility of an IO (Johnson and Rickard, Reference Johnson and Rickard2017; Funk and Tyson, Reference Funk and Tyson2022). Even the professionalism cue fails to mitigate this negative effect (from −0.23 to −0.21), implying that positive information about the WHO’s credibility may not change Republicans’ trust in the organization’s guidance. When it comes to the politicized WHO condition, we find null findings even for Republicans. Republicans may already have strong pre-existing beliefs that the WHO is politicized, which is consistent with the view that Republican media outlets have frequently expressed (Stecula and Pickup, Reference Stecula and Pickup2021).Footnote ¹⁵

We conducted additional analysis using the conditional average treatment effect approach proposed by Imai and Li (Reference Imai and Li2022), with various individual-level predispositions towards public health officials, the WHO, and COVID-19 serving as moderators. The results, presented in Appendix Section A5, demonstrate that multiple dimensions of trust – in domestic authorities, IOs, and scientific expertise – jointly determine receptivity to IO information. This suggests that IO effectiveness in public engagement depends on complex interactions between organizational credibility signals and individual predispositions across multiple domains. This reinforces our main conclusion that IO direct engagement is most effective among individuals with moderate prior beliefs and higher baseline trust in relevant authorities, but its effectiveness is limited among those with strong opposing predispositions.