What Is the Flag We Rally Around? Trust in Information Sources at the Outset of the COVID-19 Pandemic in Latvia

Abstract

Trust in information sources about COVID-19 may influence the public attitude toward the disease and the imposed restrictions, thus determining the course of the COVID-19 pandemic in a given country. Acknowledging an increase in trust in the government or the so-called rally ‘round the flag’ effect around the world at the beginning of the COVID-19 pandemic, this study explores possible determinants of this effect in Latvia, looking at such variables as the perceived disease risk, gender, age, education, income, and language spoken in the family. Presuming that risk perception may be amplified by trust in various information sources, we investigate a spill-over of the rally ‘round the flag’ effect on healthcare professionals, media, and interpersonal networks. Studying data from a nationally representative sociological survey conducted in September 2020, we confirm a positive relationship between trust in all information sources, except friends, relatives, and colleagues, and perceived disease risk. Correlations are also strong regarding trust in almost all information sources and the measured socio-demographic variables, except gender. Interpersonal trust seems to be relatively stable, and in most cases the correlations are statistically insignificant. With this study we suggest that increase in trust in government institutions as well as other information sources, even in crisis situations, does not depend on any single element, but instead presents a more complex phenomenon.

1. Introduction

At a moment of crisis, government communication is critical for mobilization of crisis-mitigation action. The media are important intermediaries in the process, transmitting information about the expected behaviour and restrictions to the population. However, when the crisis continues to spread, communication may cease to be effective in the long term if trust in information sources, either government or media, or both, is low. Observing such a tendency in Latvia, we offer one way of correlating trust in information sources and perceived disease risk through the lens of the rally ‘round the flag’ effect (Hetherington and Nelson 2003; Mueller 1973). This effect refers to cases when, during a crisis, governments tend to receive higher public trust than in routine situations. The rally ‘round the flag’ effect has been observed in many countries at the outset of the COVID-19 pandemic (The Economist 2020; Bol et al. 2021; Bækgaard et al. 2020; Kritzinger et al. 2021; Cunningham 2020).

Latvia offers an outstanding example of this effect during the first wave of the pandemic (from March to September 2020): 51% of the population trusted the government as an information source in September 2020. In a country where only 31% of the population reports confidence in national government (OECD 2021), the government institutions have received a considerable credit of trust as information sources during the pandemic. Latvia has one of the lowest levels of trust in government among the OECD countries and Latvian trust is essentially lower than trust in the EU-27 on average in government (23%), public administration (27%), parliament (21%), political parties (7%), and healthcare system (66%) (EC 2020/2021), who were among the most active providers of information during the first wave of the pandemic examined in this article.

At the beginning of the pandemic Latvia had one of the lowest rates of COVID-19 infections in the world: between 3 and 151 cases per week, the highest peak being 5 April 2020, when, in terms of cumulative infection rates, Latvia had 21 infection cases per 100,000 inhabitants (SPKC 2021). However, since September 2020 the number of infection cases started to grow rapidly, reaching the peak on 10 January 2021, when Latvia had 694 cases per 100,000 inhabitants (SPKC 2021). Despite the imposed restrictions that severely limited personal freedoms, including a weekend-holiday curfew, the infection rates remained high until July 2021, when the first positive impact of vaccination began to appear. This dynamic suggests that low trust in democratic institutions and the media might have contributed to the low impact of restrictions in the second wave of the pandemic in autumn 2020, while a short-lived rally ‘round the flag’ effect might have influenced the relatively high efficiency of restrictions in the first wave of the pandemic (from March to September 2020) which is the focus of this study.

To discover factors that might have contributed to the rally ‘round the flag’ effect, we examine a correlation between trust in government institutions and the perceived threat of COVID-19 during the first wave of the pandemic in Latvia. As various sources of information modify people’s risk perceptions, we explore if the ‘rally ‘round the flag’ effect has spilled over to other sources of information, such as healthcare professionals, media, and interpersonal networks. We add to our analysis correlations between trust and socio-demographic variables, such as gender, age, education, income, and the language spoken in the family, to demonstrate the variety of influences on individual decisions to trust sources of information. With this, we suggest that increase in trust in government institutions as well as other information sources, even in crisis situations, does not depend on any single element, but instead presents a more complex phenomenon, consisting of various, potentially interacting variables.

Our study contributes to understanding of significant differences in the rally ‘round the flag’ effect between countries even in similar epidemiological circumstances. Previous research demonstrates that various factors, i.e., perceived severity of the COVID-19 pandemic (Cunningham 2020), perceived threat to health, or partisanship alignment (Kritzinger et al. 2021) have contributed to political support in Western and Nordic democracies. However, less is known about the rally ‘round the flag’ effect in Eastern Europe, especially countries with initially low levels of trust to government. The objective of this study is to reduce this gap, discovering factors that might have contributed to the rally ‘round the flag’ effect in Latvia. Our specific objective is to test whether, in crisis situations, trust in government as an information source spills over to other information sources. Correlations between the perceived risk of disease and various socio-demographic factors and people’s trust in democratic institutions, including the media, provide useful insights into complexities governments face in curbing the health crisis. These correlations have practical implications for government communication planners and media professionals.

2. Materials and Methods

2.1. Literature Review

Rally ‘round the flag’ effect points to an increasingly supportive public opinion toward politicians and governments at the outset of crisis. Early research on the impact of COVID-19 lockdowns on political support in 15 western European democracies shows that the support for the status quo decision makers, institutions, and regimes has grown, affirming citizens’ approval of strict confinement measures (Bol et al. 2021). Bækgaard et al. (2020) have observed that the rally ‘round the flag’ effect in Denmark has spilled over to other democratic institutions, particularly, the judicial system and the public sector, and possibly to the parliament and the media. Strömbäck (2021) concluded that trust in media during crisis either fluctuated or increased. Esaiasson et al. (2020) show that in Sweden not only institutional but also interpersonal trust has increased despite the ongoing public contradictions on crisis management.

During the COVID-19 pandemic, differences in the rally ‘round the flag’ effect across countries have been attributed to a range of factors. Studying Austria and France, Kritzinger et al. (2021) demonstrate that the perceived threat to health has been the most important variable for augmented government support in Austria, whereas the perceived threat to health has not been essential in France—instead, the partisan alignment with the position or opposition parties determined the level of trust (Kritzinger et al. 2021). In addition, in both countries, the perception on the appropriateness of the government measures has correlated positively with trust (Kritzinger et al. 2021). Meanwhile, Johansson et al. (2021) illuminate that in Sweden the political ideology, alongside the perceived effects of the pandemic on the country, have become significantly more important at later phases of the pandemic, and it explains the declining support to the government in Sweden after the initially elevated levels of approval. The rally ‘round the flag’ effect has been associated with the levels of severity of the pandemic, claiming that in the countries hit harder by the pandemic, for example, the United States, Italy, and the United Kingdom, the rally effect has been weaker (Cunningham 2020). On the contrary, exploring the case of the Netherlands, Schraff (2020) argues that it is emotionally determined collective anxiety, emerging because of the severity of the pandemic, i.e., the increase in the COVID-19 infections, which fosters the citizens’ support for political institutions. The study affirms that with the increasing numbers of COVID-19, the impact of social trust and economic evaluations on political trust decreases (Schraff 2020).

In sum, studies show that such factors as perceived severity of the pandemic, perceived threat to health, perceived effects of the pandemic on the country, perceived appropriateness of the government measures, as well as collective anxiety and partisanship alignment have been important in driving political support in Western and Nordic democracies. However, less is known about the rally ‘round the flag’ effect in Eastern Europe, especially ethnically heterogenous, lower income societies with less established partisanship traditions such as Latvia.

2.2. Latvia as a Case Study

Latvia is an ethnically mixed country: 62% of the population identify as Latvian, 25% as Russian, and 13% as other (Central Statistical Bureau of Latvia 2021). Such heterogeneity is largely a result of industrialisation and the immigration policy of the Soviet Union, of which Latvia was a part between 1944 and 1991. Additionally, media consumption in Latvia is largely determined by one’s ethnic group; therefore, different ethnic groups in Latvia may acquire their daily information from various sources. For example, research shows that a part of Latvia’s Russian-speaking population prefer TV channels controlled by the Russian government (NEPLP 2020; Valtenbergs et al. 2018).

Moreover, the political party system in Latvia also represents a division along ethno-linguistic lines instead of a more traditional political divide between the right and left. The Latvian political party system forms a two-dimensional ideological space, representing the identity and the socioeconomic dimensions (Ikstens 2018). Though the centre-right and right parties differ from the centre-left parties with more liberal or socially conservative policies, they work for strengthening the Latvian identity, whereas the centre-left and left parties fight for political and cultural rights of the Russian-speaking population (Ikstens 2018). Consequently, the ethnic structure of party support in Latvia is clearly polarised, with the Latvian-speaking population voting for right-wing parties, while the Russian-speaking population choosing left-wing parties (Eihmanis 2019). The latter are relatively well represented in the parliament but have never formed government.

Finally, although economically Latvia’s living standards continue to converge with the EU-average (Latvia’s gross domestic product per capita is 72% of the EU average (Eurostat 2021a)), unemployment, high inequality, and low income accompany the growth. For example, Latvia has the fourth lowest minimum wage in EU (Eurostat 2021b) and one fourth of the Latvian population lives at a risk of poverty (Eurostat 2021c). Massive emigration of economically active part of the population has also contributed to the country’s economic hardship. Since 2000, when Latvia started the EU accession negotiations, approximately 260 thousand people, or 11% of the population, have emigrated from Latvia (Lulle 2021).

2.3. Conceptual Framework and Research Questions

Adding these three factors—ethnic diversity, absence of established partisanship, and low income—to permanently low level of trust in democratic institutions in Latvia, our examination of the rally ‘round the flag’ effect adds new knowledge to the broader, global debate about the role of this effect during the COVID-19 pandemic. In the rally ‘round the flag’ effect studies, the focal point of analysis is trust, which is a key factor to mobilize the public support toward crisis mitigation action, proposed by the government. Studies confirm that a lockdown, imposed after the outbreak of the COVID-19 pandemic, increased the support to democratic institutions (Bol et al. 2021; Bækgaard et al. 2020). Besides curbing the pandemic, the rally effect provides benefits for politicians, for instance, enhancing their chances in elections (Yam et al. 2020). Nevertheless, the rally effect lasts for a relatively short term (Johansson et al. 2021), fostering scholarly interest in the causes of its rise and decline. We observe a similar effect also in Latvia during the first wave of COVID-19 pandemic from March to September 2020. The rally effect, we argue, partly explains the relatively lower infection rates at the beginning of the pandemic in Latvia. Although the term ‘trust’ has multiple meanings (Luhmann 1979) and there is no single agreed definition (Enli and Rosenberg 2018), in the context of government institutions, it means that people rely on the institution to do its duty, serve in the interest of the society, and not misuse its powers (Pjesivac 2017). Trust is especially important in crisis situations such as the first wave of the COVID-19 pandemic, when the initial information is scarce, yet the situation demands an immediate response from the government and population. Regarding the rally ‘round the flag’ effect, Bækgaard et al. (2020) observe that heightened trust in government after the COVID-19 lockdown has been spilling over to other democratic institutions. Studies demonstrate that the pandemic’s course at the early stage was determined by the authorities’ ability to deliver trustworthy information (Zarocostas 2020) to various audiences and by the society’s attitude toward this information (Nielsen 2020). In agreement with these observations, for our examination of the rally ‘round the flag’ effect we focus on two major groups of democratic institutions as information sources: government and the media (including public and private broadcasters, printed press, and online news portals (ONPs)). In addition, we evaluate trust in international organisations such as World Health Organization (WHO), which was the main institution the governments followed in their communication strategies.

One’s peers, community opinion leaders, or other respected individuals tend to influence the attitudes toward government and media. Esaiasson et al. (2020) show that interpersonal trust has increased during the pandemic, thus helping to face the crisis. Thus, in addition to government and the media, we also included interpersonal communication because the informal networks of friends, relatives, and colleagues traditionally are notable sources of information in Latvia (Kalniņa 2018; Vasiļjeva 2016). We included also social networking sites (SNS) in our research as these sites have become a notable source of information for some individuals. The COVID-19 pandemic motivated people to increase their search for health information on the Internet, which, however, oftentimes caused more confusion and anxiety instead of helping (Mohammadi et al. 2020; Shetata 2020).

Finally, considering that the COVID-19 pandemic is a medical and public health emergency and healthcare professionals have had a visible place in public debates since the beginning of the pandemic, we also included this group in our research, although the medical field usually is not analysed alongside the democratic institutions.

Overall, we aimed to find out if trust or distrust in these four types of sources influence the perceived disease risk and contribute to the rally ‘round the flag’ effect.

In our research, trustworthy sources are those that the respondents believe to be deserving of their trust and confidence (Ardèvol-Abreu et al. 2018; Takahashi and Tandoc 2016; Tsfati and Ariely 2013). The media are essential intermediaries in the process of agenda-setting by selecting the issues, which present to the public (McCombs 2005; McCombs and Shaw 1972), and transmitting information, still defined as ‘social glue’ (Enli and Rosenberg 2018), yet the effect of the transmitted information will depend on the perception of the media trustworthiness. Personal factors determine an individual’s trust in the media (Yamamoto et al. 2016). Individuals who generally do not trust other people and politicians are more likely to distrust the media (Cappella 2002; Cappella and Jamieson 1997). If the audience does not trust the media, then it is more likely to also distrust the information the media conveys (Tsfati and Ariely 2013). Distrust in qualitative media can negatively affect the public’s opportunities to obtain diverse information (Kohring and Matthes 2007) and lead to a situation where the audience may not receive vital information and instead relies on alternative sources such as SNS or friends (van Dijck and Alinejad 2020), which, in a longer-term, may modify individuals’ perception of reality, as the cultivation theory suggests (Gerbner et al. 2002), or provide alternative means to cope with the cognitive dissonance the pandemic may cause (Festinger 1957). Notably, not only trust in government, but also trust in radio and television rose sharply in many countries during the COVID-19 pandemic (Newman 2020). Furthermore, recent research has shown that higher attention to news media can be related to a higher degree of adherence to the preventive behaviour guidelines (Faasse and Newby 2020; Jiang et al. 2021).

Political and social trust have been reported as significant factors in COVID-19-related policymaking and behaviour (Devine et al. 2021; Fridman et al. 2020; Guillon and Kergall 2020; Heydari et al. 2021; Karić and Međedović 2021; Mohammadi et al. 2020; Vardavas et al. 2021; Ye and Lyu 2020). For example, people who trust the government will be more willing to follow the rules even if they are not too favourable to them (Rudolph and Evans 2005). Distrust, meanwhile, results in reluctance to observe the rules (Marien 2011). Higher trust tends to provide a wider scope of policy choices for the authorities, whereas lower trust implies harsher measures and enforcement (Harring et al. 2021). Distrust in the public sector and media paves the way for consumption of alternative sources, including those that spread disinformation and false news (Karić and Međedović 2021), which makes the fight with the pandemic even more complicated.

During the first wave of the pandemic, healthcare professionals became highly visible information sources. Trust in sources about health issues determines the extent to which people will perceive the provided information as meaningful, credible, and relevant (Rimer et al. 1999; Alexander et al. 2011). Demographic, socio-economic, and lifestyle factors also influence the perception of sources on medicine and health in general (Hesse et al. 2005) and COVID-19 in particular (Jahangiry et al. 2020). Plohl and Musil (2020) suggest that trust in scientists in general is a pre-condition to curbing the spread of disinformation and conspiracy theories. However, Eichengreen et al. (2021) warn that the pandemic might eventually diminish trust in scientists and experts.

Perceived risk of disease is one of essential factors contributing to the public trust in information sources. Attempts to combine trust in information sources with the perceived risk of disease have become increasingly popular in researching the COVID-19 pandemic (e.g., Niu et al. 2021; Mohammadi et al. 2020; Newton 2020). A person’s participation in the preservation of their physical, mental, and emotional health is related to their awareness of the risk. Recognition of the risks is necessary to change a person’s health behaviour or maintain changes in it (Siaki and Loescher 2011). Perceived disease risk is an integral part of health communication models (Becker et al. 1978; Petty and Cacioppo 1986) and health behaviour change theories (Siaki and Loescher 2011; Brewer et al. 2007). Extensive research has shown that a higher level of perceived risk leads to a higher degree of adherence to restrictions and engagement in preventive behaviours (Bruine de Bruin and Bennett 2020; Dryhurst et al. 2020; Faasse and Newby 2020; Harper et al. 2021; Siegrist et al. 2021; Wise et al. 2020).

To measure the perceived disease risk, we focus on subjective self-assessment of a person’s vulnerability to health threats (Brewer et al. 2004). Relying on the framework of health psychology, Health Belief Model (Rosenstock 1974; Becker et al. 1978) can be helpful for understanding the main determining factors of health behaviour during the COVID-19 pandemic. According to the model, any stimuli related to health create an assessment of perceived risk and the possible overcoming of this risk (Sarwar et al. 2020). This model has been successfully employed in COVID-19 research (Guillon and Kergall 2020; Heydari et al. 2021; Jiang et al. 2021; Sperling 2020). Publicly available information helps individuals to understand their disease risk (Lee et al. 2010), because perceived risk is not always related to actual risk (van der Weijden et al. 2007). A failed communication can cause uncertainty about a person’s ability to evaluate their disease risk (Orom et al. 2020). Government institutions, the media, and other people’s opinions are among those external factors that shape the risks alongside individual factors (genetics, age, experience, etc.) and culture (Millstein and Halpern-Felsher 2002; Siaki and Loescher 2011). Media literacy can also contribute to perceived risk of disease, as individuals’ ability to recognize disinformation may modify their perception of risk (Rožukalne et al. 2021).

Relying on these conceptual approaches, we formulated the following research questions:

RQ1: What was the correlation between trust in information sources and such variables as perceived disease risk, gender, age, education, income, and language spoken in the family during the first wave of the COVID-19 pandemic in Latvia from March to September 2020?

RQ2: Does trust in government institutions spill over to other sources of information, most notably the media, the healthcare professionals, and interpersonal networks?

We expected to find a positive correlation between the perceived risk of disease and trust in information sources at the beginning of the pandemic, thus contributing to the rally ‘round the flag’ effect. Hypothetically, the rally effect should be similar across different sociodemographic and ethnolinguistic groups, given that the pandemic is universal in its character, and the imposed restrictions to curb the spread of the virus are the same for everyone. Yet we expected to see variations in the strength of the correlation across the measured variables.

2.4. Method and Data Collection

We collected the data for our study in a nationally representative sociological survey using an Internet panel (CAWI method) that took place in September 2020. The sample of research consisted of one group of respondents in September (n = 1005). Adult residents of Latvia aged 18 to 75 participated in the survey. The sample of quotas was used to form the sample of the survey; the data were weighted according to the data of the Population Register of the Office of Citizenship and Migration Affairs of the Ministry of the Interior of the Republic of Latvia on 27 January 2020. A random sampling method provided a sample of adequate representations of all socio-demographic and ethno-linguistic groups of Latvia. Females represent 53% of the sample, whereas males represent 47%. The most widely represented age groups in survey are 35–44 years and 45–55 years, 21% and 20% of respondents, respectively. Furthermore, 62% of the respondents mainly use Latvian as a language of communication in the family, 36% communicate in Russian, and 2% in another language.

The survey consisted of question blocks with closed questions. The respondents were asked to evaluate specific sources of information (friends and relatives, colleagues, healthcare professionals, Latvian government institutions, TV, radio, newspapers, ONPs, SNS, international organisations such as the World Health Organization (WHO)) on a four-point Likert scale of very trustworthy to very untrustworthy. To consider trust in sources in correlation with the perceived disease risk, we asked the respondents to evaluate their perceived risk on a five-point scale from Low and almost unreal to Very high and very real. Base for average value: respondents who gave a specific answer, i.e., the frequency of mentioning the answer I do not want to answer/hard to say was not taken into the account.

The SPSS programme was used for the initial processing of the data, their comparison, and grouping, using the functions of descriptive statistics and correlation. The survey data were weighted, and a correlation analysis was performed using Spearman’s rank correlation coefficient to determine statistically significant correlations between respondents’ risk perceptions and trust in various sources of information. The Kruskal-Wallis H test (also called the ’one-way ANOVA on ranks’) was used in the data analysis, focusing on data that could characterize differences between parts of population (criterions employed—gender, age, education, income, and language used in the family) and their trust in information sources. Comparing the results verified with mentioned tests, Kruskal-Wallis H test data were chosen for the further analysis.

3. Results and Discussion

3.1. Trust in Information Sources

Data reveal that people show the highest trust toward healthcare professionals (85%), followed by international organisations such as WHO (69%), television, radio, and government institutions, whereas the least trusted source is SNS (

Table 1. Trust in information sources during the first wave of the COVID-19 pandemic in Latvia (in %).

	Trustworthy	Untrustworthy
Healthcare professionals	85	11
International organisations (e.g., WHO)	69	27
Latvian government institutions	51	45
TV	49	46
Radio	48	46
Friends and relatives	47	44
Newspapers	44	48
Colleagues	42	48
ONPs	38	57
SNS	26	68

WHO = World Health Organization; ONP = online news portals; SNS = social networking sites.

). It is noteworthy that the government institutions received relatively high support (51%). Such level of trust is higher than the general level of trust of Latvia’s population in the government. It suggests that during a crisis, government may receive a higher than usually credit of trust, if it demonstrates access to reliable information, mobilizes means to mitigate the crisis, and exercises its power. Our observation is in alignment with studies conducted in other countries, confirming that the trustworthiness of government institutions has increased during the first wave of the COVID-19 pandemic (e.g., Bol et al. 2021; Esaiasson et al. 2020; Bækgaard et al. 2020), thus confirming the rally ‘round the flag’ effect in Latvia.

Notwithstanding the rally ‘round the flag’ effect, data suggest that most of these information sources evoke a prominent level of controversy regarding their trustworthiness, except healthcare professionals and international organisations such as WHO. Most sources balance between trust and distrust. Distrust was high also in traditional media sources: less than a half (48%) of respondents trusted television and radio, while nearly as many respondents (46%) did not trust them. Our findings about traditional media sources such as television and radio are particularly puzzling, as the level of trust in them is lower than before the pandemic. These findings contradict the conclusions of Bækgaard et al. (2020) that the rally ‘round the flag’ effect potentially spills over to other democratic institutions and go in line with Strömbäck (2021) that the trust in media is fluctuating.

3.2. Trust in Information Sources and the Perceived Disease Risk

To estimate the correlation of trust in sources with the perceived disease risk, we divided the respondents into five groups according to their level of concern about the COVID-19: the unconcerned, little concerned, moderately concerned, rather concerned, and very concerned (

Table 2. Perceived disease risk during the first wave of the COVID-19 pandemic in Latvia (in %).

Perceive the Disease Risk as	Group	September 2020
Low and unreal	Unconcerned	10
Low but real	Little concerned	42
Medium high and real	Moderately concerned	33
High and real	Rather concerned	9
Very high and real	Very concerned	2
N/A	-	5

N/A = not applicable.

). We created the indicator by asking people how high the risk of COVID-19 is for (a) themselves, (b) family, relatives, and (c) Latvia’s population as a whole. They could choose among these risk evaluations: low and almost unrealistic, low but realistic, medium, and realistic, high and realistic, or very high and very realistic. Then we divided the results into quintiles.

Our data indicate that the perceived disease risk in Latvian society is relatively low: only 9% of respondents were rather concerned, while just 2% were very concerned about the pandemic (Table 2). Most respondents remained unconcerned, seeing the risk of disease as low and unreal (10%) or little concerned, perceiving the risk of disease as low but real (42%). A moderately concerned part of the population, which considered the risk of disease as medium high and real, comprised nearly one third (33%). The sparse numbers of the rather or very concerned respondents point to the optimism bias, a tendency to ‘underestimate the likelihood of negative events’ which is observed in majority of the human population (Sharot 2011). The effect of this bias is not specific to Latvia as it has been also discovered elsewhere in studies of the perception of the COVID-19 threat (Seehuus et al. 2021; Siegrist et al. 2021; Wise et al. 2020).

We assessed if the level of concern is correlated with trust in information sources. To explore the correlation, we compared the overall indicators of trust in all sources and the level of trust by groups of different perception of disease risk (

Table 3. Trust in information sources by the perceived disease risk (in %).

% Who Trust	All Respondents	Uncon-Cerned	Little Concerned	Moderately Concerned	Rather Concerned	The Very Concerned
Healthcare professionals	85	64	89	90	91	100
International organisations (e.g., WHO)	69	31	72	78	76	94
Latvian government institutions	51	23	53	59	58	81
Television	49	18	52	57	52	75
Radio	48	18	52	56	52	75
Relatives, friends	47	53	47	47	46	63
Newspapers	44	15	46	53	52	63
Colleagues	42	48	43	42	46	52
ONPs	38	13	39	43	46	63
SNS	26	16	25	29	32	45

).

We discovered that trust in all information sources, except relatives, friends, and colleagues increased with the rise in the disease risk perception, i.e., fear about the disease augments the trust, which follows the pattern, represented by the case study on Austria (Kritzinger et al. 2021). Correlation analysis confirms this finding, revealing statistically significant differences between the respondents’ risk perception and trust in distinct groups of information sources (

Table 4. Correlation analysis of data on trust in information sources and the perceived disease risk.

Spearman’ s Rho	Correlation Coefficient	Sig. (2-Tailed)	n
Friends and relatives	0.039	0.25	884
Colleagues	0.007	0.837	874
Healthcare professionals	−0.234 **	0	927
Latvian government institutions	−0.206 **	0	917
Television	−0.204 **	0	912
Radio	−0.197 **	0	909
Newspapers	−0.206 **	0	879
ONPs	−0.187 **	0	907
SNS	−0.128 **	0	892
International organisations (e.g., WHO)	−0.252 **	0	913

** Correlation is significant at the 0.01 level (2-tailed).

). First, we found no statistically significant correlation between perceptions of disease risk with trust in relatives, friends, or colleagues as sources of information. Second, we found a statistically significant correlation between risk perception and trust in government and the media. Third, the correlation analysis shows that respondents with a higher level of perceived disease risk are more likely to trust a variety of sources.

However, our data do not prove causality; thus, it is not known whether the perceived disease risk increases a person’s trust in information sources or whether higher trust in sources increases their level of perceived disease risk. Nevertheless, the correlation demonstrates that the perceived risk of disease has an impact on the rally ‘round the flag’ effect. Moreover, we observe the effect both regarding government institutions and the media, thus suggesting that in cases where people perceive the disease risk as high, the rally ‘round the flag’ effect is spilling over to the media, as Bækgaard et al. (2020) propose. Interpersonal trust follows another dynamic—trust in friends, relatives, and colleagues is not influenced by perceptions of disease risk, thus showing that deeply embedded social trust in Latvia is rather resilient against sudden shocks.

3.3. Trust in Information Sources and Socio-Demographic Variables

The study analysed the level of trust in various sources of information depending on the socio-demographic characteristics of the respondent (gender, age, level of education, income, spoken language in the family). Concerning gender, the differences are statistically significant only regarding the SNS (

Table 5. Correlation analysis of data on trust in information sources and gender.

Kruskal Wallis Test	Chi-Square	df	Asymp. Sig.
Friends and relatives	3.332	1	0.068
Colleagues	3.081	1	0.079
Healthcare professionals	1.518	1	0.218
Latvian government institutions	1.593	1	0.207
Television	2.624	1	0.105
Radio	1.932	1	0.165
Newspapers	3.126	1	0.077
ONPs	0.875	1	0.35
SNS	9.083	1	0.003
International organisations (e.g., WHO)	13.498	1	0

), i.e., women tend to trust social media more than men. It suggests that gender has no notable impact on the rally ‘round the flag’ effect.

Our findings show that age is a statistically significant variable regarding government institutions and SNS (

Table 6. Correlation analysis of data on trust and age.

Spearman’s Rho	Correlation Coefficient	Sig. (2-Tailed)	n
Friends and relatives	−0.03	0.359	913
Colleagues	−0.028	0.402	905
Healthcare professionals	0.047	0.148	966
Latvian government institutions	0.102 **	0.001	958
Television	−0.001	0.977	954
Radio	−0.003	0.922	951
Newspapers	−0.012	0.716	921
ONPs	−0.012	0.702	950
SNS	−0.117 **	0	935
International organisations (e.g., WHO)	0.161 **	0	956

** Correlation is significant at the 0.01 level (2-tailed).

). It implies that younger people tend to trust government institutions more than older people, whereas the trend is the opposite for SNS: older people trust them more than younger people. This finding affirms that age is an important variable in the rally ‘round the flag’ effect, but in this case, it does not spill over to the traditional and digital media but pertains only to the SNS.

Regarding education, the results show that differences are statistically significant for all information sources, except colleagues and SNS (

Table 7. Correlation data analysis of data on trust and education levels.

Spearman’s Rho	Correlation Coefficient	Sig. (2-Tailed)	n
Friends and relatives	0.102 **	0.002	913
Colleagues	0.026	0.426	905
Healthcare professionals	−0.078 *	0.016	966
Latvian government institutions	−0.171 **	0	958
Television	−0.107 **	0.001	954
Radio	−0.108 **	0.001	951
Newspapers	−0.096 **	0.004	921
ONPs	−0.083 *	0.01	950
SNS	−0.027	0.416	935
International organisations (e.g., WHO)	−0.084 **	0.009	956

* Correlation is significant at the 0.05 level (2-tailed). ** Correlation is significant at the 0.01 level (2-tailed).

). Lower levels of education correlate with lower trust in all information sources, except friends, relatives, and colleagues. It demonstrates that the level of education contributes to rally ‘round the flag’ effect. We observe such a trend not only regarding the government institutions but also the traditional and digital media.

Regarding income, i.e. the average monthly income per person in household over the last six months (after tax) considering all income—salaries, stipends, allowances, pensions, etc., the correlations are statistically significant in all cases, except SNS and colleagues (

Table 8. Correlation analysis of data on trust in information sources and income.

Spearman’s Rho	Correlation Coefficient	Sig. (2-Tailed)	n
Friends and relatives	0.101 *	0.01	648
Colleagues	0.027	0.496	645
Healthcare professionals	−0.110 **	0.004	685
Latvian government institutions	−0.204 **	0	677
Television	−0.124 **	0.001	678
Radio	−0.123 **	0.001	676
Newspapers	−0.090 *	0.021	653
ONPs	−0.093 *	0.016	675
SNS	0.008	0.831	663
International organisations (e.g., WHO)	−0.143 **	0	681

* Correlation is significant at the 0.05 level (2-tailed). ** Correlation is significant at the 0.01 level (2-tailed).

). The correlations reveal that lower income correlates with higher trust in friends and relatives, whereas higher income correlates with higher trust in all information sources, except SNS and colleagues. The correlations demonstrate that income as well has an impact on the rally ‘round the flag’ effect, and not only regarding government institutions, but also traditional and digital media.

Analysing the data on trust in information sources depending on the language used in the family, we found that Russian-speaking respondents had less confidence than the Latvian-speaking respondents in all sources of information during a pandemic (

Table 9. Trust in information sources of Latvian and Russian-speaking respondents (in %).

% Who Trust	Latvian-Speaking Respondents	Russian-Speaking Respondents
Friends and relatives	51	39
Colleagues	47	35
Healthcare professionals	88	81
Latvian government institutions	57	40
Television	58	35
Radio	58	33
Newspapers	52	32
ONPs	42	31
SNS	27	24
International organisations (e.g., WHO)	73	63

). Russian-speaking respondents show lower trust in government, politicians, the media, as well as in colleagues, friends, and relatives. However, regarding the most trustworthy sources of information, Russian- and Latvian-speaking respondents alike trust healthcare professionals and the WHO the most.

Using the Kruskal-Wallis H test (

Table 10. Correlation analysis of trust in information sources and language used in the family.

Kruskal-Wallis Test	Chi-Square	df	Asymp. Sig.
Friends and relatives	12.885	2	0.002
Colleagues	14.79	2	0.001
Healthcare professionals	14.496	2	0.001
Latvian government institutions	27.9	2	0
TV	53.651	2	0
Radio	59.188	2	0
Newspapers	43.879	2	0
ONPs	10.708	2	0.005
SNS	0.157	2	0.925
International organisations (e.g., WHO)	25.956	2	0

) for the data analysis, we obtained comparable results, finding a statistically significant correlation in the group of Russian-speaking respondents in almost all information sources included in the survey, except SNS.

Yet, the comparatively lower levels of trust of Russian-speaking population in all information sources and its impact on the rally ‘round the flag’ effect should be interpreted with caution. In Latvia, long-term differences in society according to ethnicity and the language used in the family influence the climate of trust and distrust. Data from a longitudinal study (2010–2020) (SKDS/Mediju nams 2019, 2020) show that Russian-speaking population expresses lower trust in all important social institutions in Latvia. Of particular importance are data on the low trust of Russian-speaking respondents in the media. They also show the linguistically divided information environment in Latvia (Rožukalne 2020), where the residents of one country receive information from and trust different media with a large part of Russian-speaking population relying on content provided by Russian TV channels controlled by Russia’s government (LF 2020).

However, our data show that the trust of the Russian-speaking population in government institutions as information sources in September 2020 was higher if compared to trust to government in general in December 2019 and December 2020 (

Table 11. Trust in institutions in 2019 and 2020 by the language used in family (in %).

	Latvian			Russian
	12.2019	12.2020	dif.	12.2019	12.2020	dif.
Cabinet of Ministers	25	29	4	16	14	−1
Television	66	59	−8	47	47	0
Radio	65	63	−2	47	50	3
Printed press	53	50	−4	42	40	−2
ONPs	41	40	−1	34	34	0
Health care (hospitals, clinics, doctors’ surgeries, etc.)	66	71	5	48	52	3

Source: (SKDS/Mediju nams 2019, 2020).

). Therefore, during the first wave of the pandemic, we observe the rally ‘round the flag’ effect regarding government institutions among both the Latvian and the Russian-speaking respondents, despite the usually lower levels of trust in the government.

We observe a similar trend regarding healthcare professionals: the level of trust in them as information sources is higher (Table 1) than general trust in them in December 2019 and December 2020 (Table 11) both in the Latvian and the Russian-speaking parts of the population. In both cases, healthcare professionals were the most trusted sources of information at the beginning of crisis.

Regarding media, the situation is different. The level of trust in TV and radio as information sources in September 2020 (Table 1) was lower than general trust in them in December 2019 and December 2020 (Table 11). We observe this trend in both Latvian- and the Russian-speaking parts of the population. However, newspapers receive almost the same level of trust throughout 2019 and 2020 in the Latvian-speaking part of the population, whereas they receive lower trust at the beginning of crisis among the Russian-speaking population. There are no significant differences in the levels of trust regarding ONPs: Latvian and Russian-speaking populations show similar trust in ONPs throughout 2019 and 2020.

Thus, we observe that the rally ‘round the flag’ effect has taken place regarding government institutions and healthcare professionals in both language groups, but the effect has not spilled over to the media, as Bækgaard et al. (2020) would predict. We observe the contrary with certain media, such as TV and radio, in both language groups. A note on the characteristics of Latvia’s media structure can shed light on the background of this finding. Our survey questions do not mention particular media names and do not distinguish between public service broadcasters and commercial media, instead listing only main types of media. Most of Latvia’s commercial radio and TV stations offer highly commercialized and entertaining content (Rožukalne 2013) which is difficult to associate with professional journalism that seeks to reveal the truth and thus inspires trust. Another key factor influencing trust is the linguistically divided audience. Latvian-speaking respondents do not trust Russian-language media, while Russian-speaking audiences have less trust in Latvian-language media (LF 2017, 2018, 2020).

3.4. Summary of Results

Summing up the analysis of correlations between different variables, we confirm a positive correlation between our selected variables and trust in information sources, thus partly confirming the assumption of our RQ1. We observe that, during the first wave of the COVID-19 pandemic in Latvia from March to September 2020, higher perceived disease risk correlates with higher trust in all information sources, except relatives, friends, and colleagues. The most trustworthy sources were the healthcare professionals, followed by public information sources such as WHO and government institutions. Socio-demographic variables such as age, education, income, and language spoken in the family also show considerable correlations with trust. Younger people and people with higher level of education and higher income show higher trust in government institutions at the beginning of a crisis. We observe the rally ‘round the flag effect’ regarding government institutions both in Latvian- and Russian-speaking parts of population. Examining trust in traditional and digital media, we find that it correlates positively with perceptions of disease risk, education, and income. Age is not an important variable for trust regarding traditional and digital media, but it is an essential factor regarding SNS. Notwithstanding the language spoken in the family, trust in media is relatively low both in the Latvian and the Russian-speaking populations. Interpersonal trust seems to be relatively stable, and in most cases the correlations are statistically insignificant, except the cases when lower levels of education and lower income correlate with higher trust in friends and colleagues. Gender is the only measured variable which does not show significant correlations of trust neither with the government nor media information sources, except SNS.

Regarding our RQ2, we observe that with some of the variables—perceived disease risk, education, and income—trust in government institutions spills over to the traditional and digital media. Gender and age are variables where trust in government institutions does not spill over to the media. Examining language spoken in family, we find that trust in radio and TV as information sources is comparatively low in the measured period if compared with the pre-pandemic period in both language groups, whereas trust in government as an information source in both groups is comparatively high. Hence, even if trust in radio and TV correlate positively with the perceived disease risk, education, income, and language spoken in family, the initial levels of trust were already lower. Hence, one can observe a decline in trust during the crisis. Finally, we find that elevated trust in government institutions spills over to interpersonal trust and trust in SNS the least amount. Friends, relatives, and colleagues as well as SNS have relatively stable levels of trust which do not fluctuate notably during a crisis.

4. Conclusions

In times of crisis, people tend to rally ‘round the flag’, thus expressing their trust in and reliance on certain sources of information. Trust in government as well as in the media, which transmit the crisis communication, has been crucially important during the COVID-19 pandemic, mobilizing for collective action to overcome the crisis. At the same time, a space for disinformation and conspiracy theories widens if there is low trust in the government officials and/or the media which spread the government’s instructions, and people begin to look for alternative sources of information. A variety of factors influence trust, especially during a crisis. This study, based on data obtained in a nationally representative sociological survey in September 2020, indicates a positive correlation between trust in information sources and the perceived disease risk, confirming the importance of health risk perception in the rally ‘round the flag’ effect in Latvia. The correlation between trust in information sources and perceived disease risk was strong regarding all sources of information, except relatives, friends, and colleagues. Correlations between trust in sources and the perceived risk are complex, and other, complementary variables should be explored further. Our study demonstrates that such socio-demographic variables as age; education; income; and, in multi-ethnic societies, also language spoken in the family can be crucial factors contributing to the ‘rally round the flag’ effect. Gender, however, does not appear as an essential variable there. Our conclusions suggest that increase in trust in government institutions as well as other information sources, even in crisis situations, does not depend on any single element, but on various, potentially interacting variables.

In addition, our study reveals that elevated trust in government institutions at the beginning of crisis does not necessarily lead to elevated trust in other democratic institutions such as media. Our data show that at the beginning of the COVID-19 pandemic healthcare professionals and the government institutions were the most trusted sources of information and had higher levels of trust than both before the pandemic and during its later phases, whereas TV and radio were less trusted. It indicates that even if trust in the media correlates positively with other variables, such as, for instance, the perceived disease risk or education, the longitudinal aspect should also be considered to evaluate the rally ‘round the flag’ effect and its spill-over to the media. We see the discrepancies between the level of trust in government and the media as a result of a process where the public responded to the government performance during the crisis by rallying ‘round the flag,’ while the perception of the media remained along the lines of long-established attitudes.

In sum, we argue that trust in government and the media alongside a high perceived disease risk could potentially enhance the government’s chances for successful management of COVID-19 crisis. We found that the segment of population experiencing very low disease risk (the group of the unconcerned) tends to distrust all other information sources except friends, relatives, and colleagues. This finding suggests that these informal, interpersonal networks have potential as information channels through which the unconcerned may be approached to increase their perceived disease risk and thus also their compliance with the pandemic restrictions and other regulations and recommendations.

This study has certain limitations, providing venue for further research. As it employed quantitative data from an internet-based sociological survey, the respondents represent only the digitally active part of society. Further studies would be necessary to validate the research results among the representatives of society who are not active on the Internet. Importantly, this study does not evaluate causality; neither does it make correlations between the perceived disease risk and the measured socio-demographic variables, which provides space for further research. Additional complementary research methods, such as interviews and observations, can be employed to determine the impact of trust on the perceived disease risk, and vice-versa. Trust reflects the cultural characteristics of specific societies, but in this study, we have not included data describing societal values and culture in interaction with public institutions. In further research on changes in the level of trust, it is important to evaluate factors that characterize the culture of society and are not related to demographic indicators. Trust in disinformation and fake news may be one direction of research.