Identifying the Trustworthiness of COVID-19 Information Sources

About the authors

Professor Ben Seyd is a Senior Lecturer in Politics at the University of Kent. Dr Joseph A Hamm is an Associate Professor at Michigan State University. Professor Will Jennings is a Professor of Political Science and Public Policy at the University of Southampton.

About the COVID-19 Recovery: Building Future Pandemic Preparedness and Understanding Citizen Engagement in the USA and UK

The programme is the result of a partnership between the British Academy, the Social Sciences Research Council (SSRC) and the Science & Innovation Network in the USA (SIN USA) to award funding to 10 transatlantic studies focusing on UK-US COVID-19 vaccine engagement. It follows a pilot study exploring levels of vaccine engagement in four locations across the US and UK, while the larger programme was expanded to study multiple locations. The programme was funded by the UK’s Department for Business, Energy and Industrial Strategy.

This report arises from a project conducted by researchers at the universities of Kent and Southampton (in the UK) and Michigan State University (in the US). The main authors of this report are Ben Seyd, Will Jennings and Joseph A Hamm. The report draws on additional research conducted by assistants at the universities of Kent and Southampton (through the TrustGov project) and at Michigan State University. The project was funded by a collaboration between the UK British Academy and the US Social Science Research Council and Science and Innovation Network. All the information provided in this report is the work of the project investigators and does not necessarily represent the views of the funders.

Readers interested in the project and its research are invited to visit the project website, where the full project report and details of further publications – including academic journal articles and commentaries – will be posted. The project website can be accessed at: https://research.kent.ac.uk/information-trust/

The project was designed to identify how far people trust sources of information on the coronavirus, and which factors shape people’s trust judgements. Trust represents an important element in whether people accept information about personal health and risk during crises like viral pandemics.(1) A noteworthy feature of the coronavirus pandemic over the last two years has been the range of actors and bodies providing the public with information; not just government ministers, but also different media outlets, scientists, doctors and a range of local and community agencies. If trust is an important element in whether people listen to this information, and comply with any guidance offered, we need to understand more clearly how far people trust these different information sources, and for what reasons. Only by understanding the reasons for trust will we be in a position to design effective communication strategies that maximise people’s receptiveness to important information and guidance during a health emergency.

In addition to exploring how far people trust different sources of information about the coronavirus, and for what reasons, the project also explored how far levels of trust in different information sources vary between groups within the population. Particular attention was paid to variations in trust between individuals located in different socio-economic and ethnic groups.

This report sets out the project’s key findings on four key issues:

1. What are the principal sources used by people to gain information about the coronavirus?
2. How far do people trust these different information sources?
3. Which factors shape individuals’ trust in different information sources?
4. What are the effects of trust in different information sources on people’s attitudes and behaviour?

The project conducted research in two countries: the United Kingdom (UK) and the United States (US). There were two main components to the research in each country: a set of focus groups among small samples of citizens, and surveys among larger, and nationally representative, samples of the population.

The project commissioned Ipsos-Mori to conduct six focus groups in the UK in December 2021, and (as part of a separate research project) eight focus groups in the US in February and May 2021. The groups in each country were conducted online among samples drawn from a variety of geographical locations. Each group comprised between five and seven participants. The UK groups contained participants from different socio-economic and ethnic groups. In both countries, the groups were stratified by age, and each group contained a broadly equal gender mix. Participants in the groups were asked to identify which features of politicians they associated with trustworthy and untrustworthy behaviour. Participants in the UK groups were asked to identify the same features of scientists.(2)

The project also commissioned Ipsos-Mori to conduct surveys in the UK and the US, involving nationally representative samples of adults aged 18+ who were interviewed online. Quotas were set in both countries on age, gender, region and working status. Results were also weighted to known population proportions on various demographic and socio-economic factors. The total number of respondents achieved was 1,501 in the UK and 1,499 in the US. In addition to a wide range of questions, the surveys also included a choice (or conjoint) experiment, in which survey participants were presented with a pair of either politicians or scientists, and asked to identify which of each pair of information source they trusted more to provide reliable information. Each actor within the pair was identified as possessing a different set of characteristics. By measuring respondents’ trust choices across the sample, the project was able to identify which characteristics of the actor are associated with higher or lower levels of trust. This exercise forms the basis for the results reported in Figure 4.

Further details of the methods used in the project can be found in its Final Report, which is available at: https://research.kent.ac.uk/information-trust/

People have used a wide variety of sources to gain information about the coronavirus.

Some of this information may have been accessed directly from a source such as a government spokesperson or a scientist or medic, for instance via a televised government press conference. People may also have accessed information from a secondary source, such as a media outlet or a personal contact. We asked our survey respondents which sources of information they had found most useful in understanding the coronavirus.(3)

The results are presented in Figure 1, and show the importance of scientific and medical experts, with information from these actors seen as being most useful by two-thirds (67%) of respondents in the UK and just over half (54%) of respondents in the US. Local healthcare workers and doctors are also seen as useful sources of information. While the media is seen as useful by more than one-half of respondents in the UK, the figure among respondents in the US is rather lower. By contrast, state governments are seen as useful sources of information among US respondents more than local councils are in the UK. In both countries, government (government ministers in the UK; the federal government in the US) is designated as a useful source of information by only around three in 10 respondents, while social media platforms are deemed useful by only around two in 10 respondents.

We find some variation in this picture when we explore the figures among different socio-economic and ethnic groups within the population. Individuals holding higher educational qualifications (notably a university-level degree or above) are more likely than individuals with lower educational qualifications (below a university-level degree) to see scientific and medical experts as useful sources of information. In the UK, there is a small gap by ethnicity, with members of ethnic minority groups slightly less likely than White individuals to see information provided by scientists as useful. Yet in the US, there is no such gap between different ethnic groups. Where there is a gap among US citizens is in attitudes towards information provided by local doctors, where Black individuals are, along with Asian individuals, less likely than their White counterparts to see information from this source as useful.

Not only are scientists and medical experts seen as useful sources of information, they are also widely trusted by citizens. We asked the survey respondents how much they trusted a range of sources when it came to providing information about COVID-19, and asked them to rate their trust on a 0-10 scale (where 0 = do not trust at all and 10 = have full trust in them). We show the results in Figure 2. These point to high levels of trust among the UK and US populations in both local doctors and scientists advising the government. Local authorities – local councils in the UK and state governments in the US – also gained a trust rating around the scale midpoint.

In the UK, television news is trusted a little more than information from newspapers. The media may not be highly trusted by people, but is seen as a useful source of information on the coronavirus, as we saw in Figure 1. In both countries, trust in the national government falls below the scale midpoint.

There are variations in levels of trust in information sources between people in different social groups. We show these figures in Table 1. In both the UK and US, trust tends to be higher among more educated and affluent people than among less educated and poorer people. When it comes to ethnicity, we do not find variations in trust in scientists between White individuals and members of ethnic minority groups.(4) Yet in both countries, members of ethnic minority groups (in the US, specifically Black individuals) are less trusting in local doctors than are White individuals. And although trust in ethnic group leaders and faith leaders is fairly low across the population, it tends to be higher among ethnic minority group members than among White individuals.

Identifying why people trust different information sources comprised the main element of the project, and so forms the largest part of this report. The project focused in particular on why people trust government officials and scientists. As we have already seen, people trust scientists much more than they do government ministers in the UK, or the federal government in the US. The project examined why this might be the case; in particular, are people’s trust judgements in scientists based on similar or different considerations to their trust judgements in politicians?

Notes: Education: low = below university degree; high = university degree or above. Income (annual household income): low = bottom income band; high = top income band.

To explore this, we firstly examined what reasons our survey respondents gave for trusting scientists. We asked them a question about the importance of various considerations in deciding whether to trust scientists on information about COVID-19. The considerations were: how scientists behave in their everyday life, their scientific background, the accuracy of the information they provide, the extent to which they care about people and whether they are independent of the government.

The results (presented in Figure 3) show that, in both the UK and US, the most important reported considerations relate to scientists’ expertise and competence. The most important consideration identified by individuals for their trust in scientists is the accuracy of the information provided, along with the scientific background of the actors. Scientists’ independence from government is also important. Less relevant to people’s assessment of their trust are scientists’ concern for other people and how they behave in their everyday lives.

The project also explored the reasons people hold for trusting different sources of information through collective discussions in the focus groups. A summary of the results of these discussions are provided in Table 2. This shows how people’s trust in different sources of information quite often rests on different considerations. People report trusting or distrusting politicians for a number of reasons. These cover how much care and concern politicians have for other people (features that we label ‘benevolence’), how honest, faithful to their promises, authentic in upholding a clear set of personal values and objective in the information they provide politicians are (which we label ‘integrity’) and how far politicians are open when it comes to information (which we label ‘transparency’). In the UK, the focus group discussions rarely identified competence as a key consideration in the trustworthiness of politicians, although this did arise in the US focus groups in the form of ‘getting things done’.

Compared to this wide base of considerations that people identify as relevant when assessing the trustworthiness of politicians, the considerations identified as relevant for other actors is rather narrower. In the case of scientists, the focus groups in the UK (the US focus groups did not explore people’s trust in scientists; see footnote 2) identified technical expertise and experience (which we label ‘competence’) as a key consideration. Alongside this, trust in scientists was seen to reflect the objectivity of the information scientists provide, which in turn also reflects their independence from government and any political pressures this might entail. If scientists are open and transparent in the information they provide, this was associated with trustworthiness; yet concerns over any suppression of information was associated with distrust. In contrast to politicians, however, benevolence – care and concern for others – was not identified by UK focus group participants as a key consideration in scientists’ trustworthiness.

As with scientists, the trustworthiness of local doctors was seen to reflect technical expertise. Doctors were also seen to be trustworthy because of their care ethic for their patients and for people in the local community. The UK focus group discussions on other community actors such as local councils, the local media and community leaders did not identify many considerations which made these sources of information trustworthy. While some focus group participants saw community leaders, for example, as concerned with local people’s needs, others saw them as pursuing particular agendas rather than serving the whole community. The national media – in particular newspapers – was widely derided in the UK focus groups, on the grounds of pursuing its own interests and of lacking objectivity.

These results suggest that many people form trust judgements about actors such as government officials and scientists on different grounds. Trust in scientists seems to reflect appraisals largely of competence and expertise, while trust in politicians seems to reflect appraisals largely of benevolence and integrity. Trust is a complex judgement to form, and is rarely reducible to one or two factors. Yet our results highlight a different balance between these factors when it comes to people’s trust judgements about scientists and government officials.

We find a broadly similar finding in the results of a conjoint experiment we ran as part of the surveys of the UK and US populations. Conjoint experiments are a way of identifying the effects of different characteristics of a product or individual on people’s judgements or choices. In this case, we presented our survey respondents with a choice between pairs of scientists or pairs of politicians, and asked them each time to select which one within the pair they would trust the most. Each of the scientists and politicians presented to the survey respondents was listed as possessing a particular set of characteristics or features. We can use the results from this analysis to identify the effects of each feature on survey respondents’ trust choices.

These results are presented in Figure 4. The dark blue dots represent the effects of each feature on people’s trust in scientists. The light blue dots represent the effects of each feature on people’s trust in politicians (government ministers in the UK and state governors in the US). Dots appearing to the right of the dotted line (which is fixed at the mean value of 0.5) indicate a feature associated with higher levels of trust; dots appearing to the left of the dotted line indicated a feature associated with lower levels of trust.

The results for scientists (dark blue dots) show the substantial effect that considerations of competence have on people’s trust. When survey respondents were faced with examples of scientists whose work was described as being low quality, their levels of trust fell considerably. The effect of competence on people’s judgements of scientists is greater than the effects of features like benevolence and representativeness (in that the dark blue dots for low competence sit further to the left, below the mean, than the dark blue dots associated with low benevolence and representativeness). Yet for politicians (light blue dots), the effect of competence on people’s trust judgements is no greater than the effects of benevolence (measured by the degree of concern with ordinary people’s lives), representativeness (measured by how in touch with ordinary people the actor is described as being) and honesty (measured by the extent to which the actor’s claims are supported by evidence).

Being portrayed or perceived as lacking competence and expertise therefore appears to be particularly damaging for people’s trust in scientists. Trust in politicians appears to be more sensitive to a broader set of considerations: not only how competent they are, but also how in touch with other people they are and how concerned they are with ordinary people and their lives.

The results of the conjoint experiment also point to the importance for scientists of being seen as independent from government. Scientists who are described as considering the scientific evidence but adjusting their decisions to reflect what politicians believe are marked down on trust, while scientists who do not adjust their decisions in this way are marked up. We note that a similar – if rather weaker – effect is seen for politicians themselves. Where government ministers (in the UK) or state governors (in the US) are presented as limiting their decisions to the scientific evidence alone, people’s trust in them rises. Where those actors are presented as adjusting their decisions to reflect what other politicians believe, people’s trust in them falls.

This suggests that government officials can gain the public’s trust by sticking close to ‘the science’. Yet our conjoint experiment also tested whether trust in scientists and politicians is affected by whether these sources focus only on the scientific evidence, or balance that evidence by also taking into account other considerations like the needs of business. We can see in Figure 4 that the dark blue dots for the scientists sit on, or very close, to the line running up from the mean level of 0.5. This means that whether scientists do, or do not, take into account the needs of business has no significant effect on whether people judge them as trustworthy. Yet politicians who focus only on the scientific evidence and ignore the needs of business are marked down on trust, while politicians who incorporate the needs of business into their decisions are marked up on trust.

It appears, then, that when scientists stick to the scientific evidence, their stock of trust increases. Yet for politicians, there is a trickier balancing act. On the one hand, they need to follow the scientific evidence and not be swayed by political pressures. Yet on the other hand, they must adjust the scientific evidence to include the needs of business. Here, at least, people’s trust in politicians seems to reflect a combination of factors that is not easily managed.

We highlighted in Section 5 how levels of trust in key information sources often vary quite significantly between different groups within the population. To pursue this issue, we examined how far the effects on trust of the factors and considerations just explored varied between different groups within the population. Do certain groups – more economically deprived individuals and members of ethnic minorities, say – trust different actors for different reasons? Our research findings suggest not. We lack the space here to explain these findings in detail (they can be found in the project’s Final Report, available at: https://research.kent.ac.uk/information-trust). In summary, though, the results of the focus group discussions and conjoint experiment both suggest that individuals in different social groups tend to trust government officials and scientists for broadly similar reasons, not for distinctive reasons.

We know from previous studies that levels of vaccination among the population are often highly sensitive to people’s feelings of trust: vaccination levels tend to be lower among people who don’t trust government than among people who do trust government. (5) We explored the role of trust on vaccine behaviour by measuring trust both in government and in scientists, and also by measuring people’s likely future vaccination behaviour. To capture this, we asked people whether they would be likely to get any future COVID-19 vaccination immediately or whether they would not get any future vaccination. Among the UK population, 62% indicated they would get vaccinated immediately, while 10% indicated they would be likely to refuse a vaccination (the remaining proportions are accounted for by other response options to our survey question). Among the US population, 52% indicated they would get vaccinated immediately, while 17% indicated they would be likely to refuse a vaccination.

People’s future vaccination choices are shaped by their feelings of trust. As shown in Figure 5, people in both the UK and US who express high trust are more likely to indicate that they will get vaccinated immediately in future than people who express low trust. Yet the association with future vaccination behaviour is stronger for people’s trust in scientists than for their trust in government ministers. This is particularly the case in the UK. Here, only one in five people (21%) expressing low trust in scientists report being willing to get vaccinated immediately in future, while among people expressing high trust in scientists, the figure is 73%. By contrast, more than half (54%) of those expressing low trust in government ministers indicate that they will get vaccinated in future, not that far below the level among people expressing high levels of trust (73%).

In the US, trust in scientists is similarly closely associated with people’s reported likelihood of getting an immediate vaccination in the future. Yet trust in the federal government is also quite strongly associated with future vaccination intention. Among US citizens who express high trust in the federal government, 70% indicate they will get a COVID-19 vaccination as soon as it becomes available. This figure drops to just 30% among people with low trust in the federal government. We can speculate that this is because, in the US, attitudes to national government and feelings of trust are shaped by people’s party identity, and partisanship is strongly associated with attitudes to vaccination.

We also explored how far people’s attitudes to coronavirus lockdown measures are associated with their trust in different information sources. The lockdown measures we examined were mandatory mask-wearing in public places, requiring people to work from home, the closure of schools and restricting admittance to entertainment and leisure venues to individuals who have been vaccinated (the so-called ‘vaccine passport’).

Much as with individuals’ intentions towards vaccinations, we find that trust has a positive association with people’s support for different lockdown measures. The results are presented in Figure 6. In most cases, support for the measures is greater among people expressing high levels of trust than among people expressing low levels of trust. Yet in the UK, the associations are greater in the case of people’s trust in scientists than in the case of their trust in government. When it comes to measures like compulsory mask-wearing, working from home and COVID-19 vaccine passports, support is substantially greater among people expressing high trust in scientists than among people expressing low trust. The exception is restrictions involving the closure of schools, which is less closely related to trust in scientists in the US, and not at all related to trust in scientists in the UK.

In the UK, trust in the government only appears to shape support for restrictions in the case of ‘vaccine passports’; otherwise, support for social restrictions is not substantially associated with people’s trust in their political rulers. By contrast, in the US, trust in the federal government is more closely related to people’s support for lockdown measures. Again, this points to the politicisation in the US of people’s feelings of trust in the government and of their support for lockdown measures.

People’s attitudes to social restrictions thus appear to be bound up with their evaluations of national government in the US in a way that does not appear to apply in the UK.

In this section, we identify the main implications of our findings, focusing on who are the key trusted sources of information on the coronavirus, and which factors shape people’s trust in government officials and scientific experts.

Who does the public trust?

Our findings point to a wide variation in people’s trust in different sources when it comes to providing information about the coronavirus. We have found high levels of trust across the UK and US populations in scientific and medical experts and local doctors, somewhat lower levels of trust in the media and government actors (in spite of the success of the mass coronavirus vaccination programme), and very low levels of trust in social media platforms. We have also found differences within the population in levels of trust. Levels of trust tend to be higher among more affluent and educated individuals than among their poorer and less educated counterparts. This suggests that particular effort needs to be devoted to conveying information to individuals in more marginalised social groups who, given their lower levels of trust, are likely to be less prone to access and act on potentially important health-related information.

Among members of ethnic minority groups, we did not find significantly lower levels of trust in scientists than among White individuals, although among Black individuals in both the UK and US, we did find evidence of lower trust in doctors.6 Although members of ethnic minority groups are not particularly prone to believe the media provides useful information about the coronavirus, they do appear to trust media outlets somewhat more than do White individuals. In addition, ethnic minority group members also trust local ethnic group leaders rather more than do White individuals, while also seeing friends and family and social media platforms as more useful sources of information about the coronavirus than do White individuals. These results point to the importance of disseminating health-related information via multiple sources in order to attract attention and acceptance among ethnic minority groups.

Why do people trust different sources?

This report has focused on trust in government officials and scientific experts. It suggests that people’s trust judgements in these two information sources rest on somewhat different considerations. However, it does not show that individuals located in different social groups within the population draw on different considerations in assessing the trustworthiness of these sources. There is thus some variation in bases of trust when it comes to different actors, but little variation in the bases of trust between different individuals.

People’s trust in scientists is strongly shaped by considerations of competence and expertise. Integrity and transparency are also important considerations in people’s trust in scientists. Scientists tend to be trusted more in the public eye if they are seen as objective and independent of political pressures. There is clearly an important balancing act here for scientific and medical advisers; to engage with, and to inform, the decision-making process (as many scientists have done during the coronavirus pandemic), while also maintaining an independence from political influence.

People’s trust in politicians rests on a slightly different, and somewhat broader, set of considerations. It is not enough for them to be seen as competent and objective alone. They must also be seen to understand, and to care about, people’s needs, and to be seen as honest (by admitting their mistakes, for example). Politicians who are seen as uncaring and out of touch are unlikely to be effective guides of the public during a health crisis. Moreover, unlike in the case of scientists, people’s trust in politicians appears to require that they listen to the scientific evidence while also meeting the wider needs of the economy. In all this, people’s trust in government officials rests on a broad, and potentially difficult to balance, set of considerations.

On the basis of our research, we make the following recommendations around public communications in a health emergency such as a viral pandemic:

General

1. The role of scientific and medical experts in fronting the provision of public information about the coronavirus – and on future health risks – is effective and should be continued.
2. A variety of other sources – such as ethnic group and faith leaders, and different media outlets – should also be used to provide health-related information in a way that appeals to a range of individuals, particularly those among different ethnic minority groups.
3. Further research is needed to identify sources of information, and types of messages, that may appeal in particular to individuals in more deprived socio-economic groups within the population.

Scientific sources of information

4. Scientific and medical experts providing the public with information need to emphasise, and reinforce, their technical knowledge and expertise, as this is the primary feature on which people assess their trustworthiness.

5. Scientific and medical experts also need to pay close attention to the integrity of their work, as people’s trust in scientists is sensitive to perceptions of objectivity and transparency.

6. While scientific and medical experts need to be aligned with governments in their role in informing and guiding decision-makers, they also need to make clear their independent status. Any public perceptions that scientists are not independent of government are likely to lower people’s assessments of their trustworthiness.

Political sources of information

7. Part of the long-term planning for a health emergency needs to include politicians – in particular senior government figures – acting in ways likely to increase public perceptions of trustworthiness. People evaluate the trustworthiness of politicians against a range of factors, which complicates the task of boosting trust. Yet trust is an important resource in a health emergency, and politicians should give thought in advance to ways in which they can act to increase public perceptions of their trustworthiness.

Brodeur, Abel, Idaliya Grigoryeva and Lamis Kattan (2021) ‘Stay-at-home orders, social distancing, and trust’, Journal of Population Economics, 34: 1321–1354

Jennings, Will et al (2021) ‘Lack of trust, conspiracy beliefs, and social media use predict COVID-19 vaccine hesitancy’, Vaccines, 9:6, 593

Larson, Heidi J et al (2018) ‘Measuring trust in vaccination: A systematic review’, Human Vaccines and Immunotherapeutics, 14:7, 1599-1609

1. Abel Brodeur, Idaliya Grigoryeva and Lamis Kattan (2021) ‘Stay-at-home orders, social distancing, and trust’, Journal of Population Economics, 34: 1321–1354; Heidi J Larson et al (2018) ‘Measuring trust in vaccination: A systematic review’, Human Vaccines and Immunotherapeutics, 14:7, 1599-1609; Will Jennings et al (2021) ‘Lack of trust, conspiracy beliefs, and social media use predict COVID-19 vaccine hesitancy’, Vaccines, 9:6, 593.
2. The US focus groups were conducted under a different research programme, which primarily focused on people’s trust in politicians not in scientists.
3. Respondents were allowed to select up to three sources. The data in Figure 1 thus represent the number of times a source was identified as first, second or third most useful, expressed as a proportion of the total.
4. In the UK, limits on sample sizes mean we only report results for members of an ethnic minority group. In the US, we distinguish between members of Black, Hispanic and Asian ethnic groups.
5. Larson et al (2018) ‘Measuring trust in vaccination’; Jennings et al (2021) ‘Lack of trust, conspiracy beliefs’.
6. We should note that our surveys, while comprising samples of 1,500 people, did not contain sufficient numbers of people from ethnic minority groups to provide detailed breakdowns within this category. We would commend further survey research, using bespoke samples or booster samples, to further explore the findings reported here.