Introduction

Even though falling vaccination rates have been a concern for public health authorities and health professionals around the world for quite some time (Dube et al. 2014), routine vaccination rates took a dire turn during the COVID-19 pandemic, with declines in both vaccination rates (see Cunniff et al. 2023 for a review) and overall vaccine confidence (de Figueiredo et al. 2022). For example, Estonia, where the current empirical study is conducted, has witnessed a steep decline in people’s beliefs in the importance, safety, effectiveness, and compatibility of vaccines in recent years (de Figueiredo et al. 2022). Such trends indicate that vaccine hesitancy, i.e., the number of individuals who “delay in acceptance or refusal of vaccination despite availability of vaccination services” (MacDonald, SAGE Working Group on Vaccine Hesitancy 2015, p. 4161), has been growing around the world. However, as the term vaccine hesitancy has been inconsistently used (cf. Browne, 2018; Bussink-Voorend et al. 2022), it is important to clarify that, in line with previous scholarship (cf. Bedford et al. 2018; Browne, 2018), we understand vaccine hesitancy/scepticism as an attitude or belief that contributes to behaviours involving the delay or decline of some or all recommended vaccines.

The WHO has declared that vaccine hesitancy, sometimes also referred to as vaccine scepticism (Kate et al. 2021), is a complex phenomenon (WHO SAGE report, 2014) that has become an increasingly important barrier to optimal vaccine coverage rates (Browne, 2018). In fact, vaccine hesitancy was named one of the top ten threats to global health by the World Health Organization (WHO) even before the COVID-19 pandemic struck the world (Ten threats…, 2019).

Prior research on vaccine hesitancy (see Bussink-Voorend et al. 2022 for a systematic review) indicates that individuals cannot be grouped exclusively as vaccine acceptant or hesitant. Rather, it is important to distinguish between people wholly opposed to vaccination (e.g., antivaxxers) and individuals who either have limited or inaccurate health information or who have genuine concerns and questions about any given vaccine, its safety, and the extent to which it is being deployed in their interests before accepting it (MacDonald, 2015). At the same time, however, one of the definitions of vaccine hesitancy views the phenomenon as a psychological state of indecisiveness that people may experience when making a vaccination decision (Bussink-Voorend et al. 2022). Therefore, it is crucial to consider people’s propensity to get involved in vaccination issues (e.g., collecting and evaluating information, weighing up pros and cons), as vaccine hesitancy may be displayed either by people who are ignorant or indifferent to vaccination issues and may simply forget to vaccinate, as well as by people who are strongly committed to making their own decisions regarding vaccination and who are willing to seek detailed information in order to make an informed choice (Peretti‐Watel et al. 2019). Considering the above, vaccine hesitancy should not be considered a static state (Browne, 2018); instead, vaccine decision-making should be seen as a journey with ups and downs, and changes over time with different influences and nudges along the path that sometimes prompt hesitancy and sometimes nudge a positive intention to vaccinate.

Previous research (Peretti-Watel et al. 2019) highlights that vaccine scepticism can range from strong antivaccination sentiments to merely having doubts about vaccines, leading to considerable differences in immunisation decisions. Thus, distinguishing between the two stances is important as the underlying reasons for vaccine scepticism may differ. Hence, as suggested by Brunson and Sobo (2017), providers, policymakers, and the media (Stephenson et al. 2018) should aim to recognise the complex nature of individual vaccination decisions, rather than frame the discussions in terms of being pro- or anti-vaccine.

Thus, even though previous research has viewed lower levels of education (Lazarus et al. 2021; Gottlieb, 2016) and lower income (Nguyen et al. 2022), as well as mistrust in governments and science (Shakeel et al., 2022) to be the main drivers of vaccine hesitancy, the topic is much more complex than that. In fact, previous research (Schmid et al. 2017) on influenza vaccine hesitancy, for instance, has identified more than 70 factors, including different cultural, psychological, political, economic, spiritual, and cognitive factors (Troiano and Nardi, 2021) that influence vaccine hesitancy. For example, adherence to complementary and alternative medicine has been found to be one of the strongest correlates of vaccine hesitancy (Frawley et al. 2017). Hence, as argued by Mesch and Schwirian (2019), confidence in the safety and effectiveness of the vaccine and the public health care system in general forms an essential part of vaccine hesitancy. Furthermore, according to Hornsey, Harris, and Fielding (2018), vaccine refusers strongly value independence and agency with respect to their own health and have personality traits associated with non-conformism, that is, they consider themselves independent thinkers who can reject consensus views. In short, individuals with vaccine hesitancy tend to trust themselves as experts and rely upon their own judgement in evaluating relevant information related to vaccines.

Although understanding population-specific drivers of vaccine uptake, and thereby identifying and measuring barriers to vaccine acceptance, has been considered crucial (Tuckerman et al. 2022), research on vaccination refusal has so far tended to take an etic perspective. That is, they have been mainly quantitatively oriented and aimed at taking objective outsiders’ perspectives in their empirical studies for assessing vaccine hesitancy. However, the instruments used in these studies have been highly heterogeneous, indicating the challenges scholars face when operationalising vaccine hesitancy (Bussink-Voorend et al. 2022). The number of studies that have aimed to provide an emic perspective on the topic, that is, tried to capture the beliefs and understandings from insider’s viewpoints, has been rather limited. In short, there is still a lack of qualitative studies involving individuals who have exhibited either strong antivaccination sentiments or who have had doubts about vaccines (e.g., Kate et al. 2021; Helps et al. 2018). This is also the reason why scholars (Dube et al. 2014) have urged future research to focus on studying the beliefs and perspectives of people with vaccine hesitancy themselves.

For the above reasons, we have decided to adopt Q methodology, a method designed for accessing individuals’ subjective perspectives (Watts and Stenner, 2012), to explore subjective understandings among vaccine-hesitant individuals: people who have refused or postponed vaccinating themselves or their children despite the availability of vaccination services. Although Q methodology has been quite actively used in recent years to study different health-related phenomena (cf. Churruca et al. 2021 for a scoping review), only a limited number of studies have made use of Q methodology to explore immunisation decisions (Harvey et al. 2015; Patty et al. 2017). As Q methodology enables participants to express what they feel and think in a systematic manner (Ellingsen et al. 2014), we decided to use this approach to explore the subjective viewpoints among Estonians (N = 29) holding a range of vaccine-sceptic views. Considering that the coverage of Estonian children with vaccinations has dropped sharply after the COVID-19 pandemic and is already at the border of 70 percent, in contrast to the WHO-recommended 95 percent according to Estonian Health Fund (2024) (almost 25,000 children…2024), we were especially interested in inviting vaccine-hesitant parents who have had to make a decision about immunisation not only for themselves but also for their child(ren) to participate in the study.

Like Zhang and Monnat (2024), we believe that understanding the diversity of vaccine hesitancy within the specific cultural context is crucial before proposing population-specific solutions for vaccine uptake. Planning and executing a successful and efficient human factor-centred vaccination campaign is complicated, as the target group is not a homogeneous entity. Rather, it is a group shaped by a multitude of contexts, experiences, and desires, various levels of health literacy, values, and expectations. Hence, a top-down, one-size-fits-all approach cannot and has not worked. In fact, according to the Cochrane review (Cooper et al. 2021), no single or specific face-to-face intervention has been found that could positively influence vaccine hesitancy.

Method, sample, and analysis

General overview of Q methodology

Q methodology is a method suitable both for exploring the range and diversity of individual perspectives (e.g., views, opinions) as well as constructing broad categories of the subject matter and discovering patterns within and between these categories (Shinebourne, 2009).

All Q methodological studies have two key characteristics: (1) data collection in the form of Q sorts; and (2) subsequent intercorrelation and by-person factor analysis of those Q sorts (Watts and Stenner, 2012). Q sorts are a collection of items that are sorted by a participant according to a subjective measure, such as “agreement/disagreement”. Through sorting the items, the participant provides a model of their viewpoint. Although the method has been quite actively used in recent years to study different health-related phenomena (cf. Churruca et al. 2021 for a scoping review), only a few Q methodological studies have been carried out to explore immunisation decisions (Harvey et al. 2015; Patty et al. 2017).

We decided to make use of Q methodology for three reasons. Firstly, during the COVID-19 pandemic, heated discussions on the topic of vaccination started all over the world, leading to polarised views on the topic (Altman, Miner et al. 2023). Q methodology has been viewed as an ideal method for studying complex topics that tend to elicit strong opinions (Farrimond, 2017) and ethically fraught issues (Churruca et al. 2021). However, the method has been praised for its less-confrontational nature, as the study participants are asked to respond to pre-established statements rather than direct questions (Butler et al. 2014). Hence, we believed that Q methodology would enable the participants to sort and express what they think and feel in a systematic manner (Ellingsen et al. 2014). In addition, Q methodology has been valued for its ability to bring out marginalised viewpoints, i.e. viewpoints which are oftentimes overlooked or ignored (Brown, 2006, p. 376).

Secondly, we wanted to give privilege to the perspectives of vaccine-hesitant people (Alexander et al. 2018). A factor in Q methodology identifies a group of persons who have rank-ordered the provided items in a very similar fashion or, in other words, a group of persons who share a similar perspective, viewpoint, or attitude about the topic at hand (Watts and Stenner, 2012). At the same time, the forced-choice answering style enables researchers to identify opposing opinions (Jueng et al. 2017). Thus, the method enables us to propose a typology of vaccine-hesitant individuals and analyse the differences between the segments (i.e., types) accordingly. Thirdly, we were intrigued by the success of applying Q methodology to study different health-related phenomena and believed in the value of the methodology for studying vaccine hesitancy, a topic that is often met with diverse viewpoints. According to Stenner et al. (2003), it is important to reiterate that, in a Q study, the aim is not to offer generalisations based on population statistics, but to strategically sample extant diversity in viewpoints within the population.

Q set design

According to Watts and Stenner (2012) the first step in the Q methodology study is the formulation of a concourse, i.e. an overall field of shared knowledge and meaning on a given topic from which it is possible to extract an identifiable universe of statements for a specific context. Thus, in order to identify and capture the main statements and topics of discussion related to vaccine hesitancy, we relied upon different information sources. For building our Q-set, i.e., the statement set sampled from the concourse, we made use of both academic and popular media articles; alternative media websites; social media groups dedicated to alternative medicine; reports from applied research projects; as well as vaccination awareness campaign materials.

All three authors engaged in developing the initial statement set, that is, we redefined, refined and integrated them into a Q set consisting of 56 statements. As our aim was to explore subjective understandings among vaccine-hesitant individuals, we set out to cover a spectrum of statements related to vaccines and vaccinations: overall opinions about vaccination and healthcare (e.g., “In my opinion, all the vaccines described in the national state-funded immunisation programme of Estonia should to be done”; “One should vaccinate as little as possible and as much as necessary”); trust in the medical system and society (e.g., “I trust the guidelines issued by the state on the topic of vaccinations”, “I ask alternative medicine specialists for advice on the topic of vaccination”); media consumption (e.g., “It is not possible for me to find objective information about vaccines” or “Media covers the topic of vaccination in a biased manner”); myths and widespread misinformation (e.g., “Vaccines can cause autism”; ”Only pharmaceutical companies benefit from vaccines”); and personal values (“Vaccination is a personal decision”).

Following the examples of others (e.g., Balloo et al. 2018), we took special care to formulate the statement set in various ways. For instance, some of the statements were formulated positively, some negatively; some directly and some indirectly, to invite our participants to think thoroughly about their choices and limit socially desirable responses. For example, descriptive statements helped to reorient the nature of the issue and thereby reduce social desirability bias, encouraging participants to speak without judgement (Johnson and Van de Vijver, 2003). We also took special care to encourage some of the participants, who believed their vaccine sceptic views are not accepted in wider Estonian society, to participate in the study.

To ensure the applicability and comprehensibility of the statement set, a pilot study with two participants was conducted. Following the pilot testing, some adjustments were made, such as revising the wording of some statements. The statements were originally developed in Estonian, and a translation into Russian was made by the first author so that the statement set would be available for participants whose mother tongue was Russian (Table 1).

Table 1 Q-set of statements.
Full size table

Participants

As Q-methodological studies tend to explore diverse viewpoints within a particular set of people (Harvey et al. 2015), we decided to follow the principles of purposive sampling. Our participants were recruited from vaccine-hesitant target groups, i.e., all our participants had been hesitant about vaccinations and had delayed or refused one or more vaccines either for themselves or for their children that are otherwise covered by the national immunisation programme of Estonia. Considering that the number of parents who have doubts about vaccine decisions impacting their children has been rising, according to the WHO (Ten threats…, 2019), we also wanted to include individuals who have made vaccination decisions about their children. Some of our participants identified themselves as antivaxxers, some as sceptics, while others considered themselves to be neutral or “regular Estonian people” (The interviewee FE3), despite having postponed or refused to vaccinate themselves and/or their children either at the time of the COVID-19 pandemic or prior to that. We wanted to include participants with such a range of viewpoints because previous research (Peretti-Watel et al. 2019) has found that vaccine scepticism ranges from strong antivaccination viewpoints to having doubts about vaccination, leading to a variation in vaccination decisions. Furthermore, Peretti-Watel et al. (2015) have argued that it is crucial to differentiate between vaccine hesitancy and vaccine rejection as the underlying reasons for such views may differ. Hence, we believed that including individuals with both viewpoints as participants in our study enables us to avoid making too hasty generalisations on the topic.

We recruited the participants via social media, mainly from Facebook communities dedicated to alternative medicine topics, as well as through personal private connections. We sent invites to individuals who had been publicly visible spokespersons from the field of healthcare and the anti-vaccination movement, as well as to individuals who classified themselves as average Estonians who had not had some vaccines for various reasons but did not identify themselves as vaccine-hesitant. We did not have problems finding such individuals, as many of our contacts knew someone who had doubts about vaccines or had expressed anti-vaccine sentiments in the context of their family or workplace. We also made use of a snowballing recruitment technique (Dieteren et al. 2023) and invited our participants to suggest and recruit additional participants to partake in the study.

Q methodology studies tend to have relatively small samples, as enhancing sample size does not provide any advantage in analysis (Zabala et al. 2018). Our final sample comprised 29 individuals. At that point, we felt that we had managed to cover the full range of different views, as data saturation began to occur (Dieteren et al. 2023), which is typical of all qualitative sampling processes (Kumar et al. 2023).

Also, similar to previous scholarship (Nurmi and Jaakola 2023; Beavis et al. 2022) where the authors have been interested in capturing vaccine-hesitant parents’ views, most of our participants were female (N = 22) and only seven were male. This discrepancy is understandable, as females often carry more responsibility for health-related decisions in the family, like taking children to medical check-ups.

The highest number of participants in our study were between 30 and 44 years of age (N = 12). Nine respondents were 45–54 years old, four participants were between 55 and 64 years of age, three were over 65 years old, and two belonged to the 18–29 age range. Most of the participants (N = 23) were parents who had made vaccination decisions for their children, while six participants had made vaccination decisions only for themselves.

The participants were located all over Estonia, coming from both larger and smaller cities, as well as rural areas. As Estonia has a considerable Russian-speaking population, the native language of seven of our participants was Russian. Most of the participants (N = 20) had higher education (ranging from BA to MA and PhD degrees); seven participants had completed vocational secondary education, and two individuals had secondary education. In short, our sample demonstrates what Hornsey et al. (2018) have also noted—vaccine hesitancy tends to permeate different social strata.

All participants volunteered to participate in the study and gave their written consent. We provided the participants with an overview of the aims of the study, introduced them to the method, and explained how the findings would be presented. All findings are anonymised, ensuring that no participant is traceable from the data presented.

As we were interested in exploring individuals’ subjective opinions on the topic of vaccines, no sensitive personal health or special types of personal data that could help identify the data subject were collected. Q methodology statement sorting, as well as accompanying interviews, were conducted either during face-to-face meetings or over Zoom, depending on the participants’ preferences. Both methods are equally popular among scholars (Dieteren et al. 2023). In Zoom, the meeting procedure was the same: the interviewee showed the cards through the screen and manually placed them on the paper-printed large grid, which was simultaneously visible to the interviewee.

Procedure

The participants were instructed to rank the statements in accordance with their personal views. First, the 56 statement cards were asked to read and sorted into three categories: statements “you agree with”, statements “you disagree with”, and statements “you are unsure of”. Through sorting the items, participants provided us with a model of their viewpoint on the issue under study (Stenner and Stainton Rogers, 1998). After the initial sorting, participants were introduced to the grid and encouraged to rank the cards using a scale of −5 to +5, with -5 meaning “most disagree with/important” and +5 meaning “most agree with/less important”, or in the category “0” if they were unsure whether they agreed or disagreed. There were two places for +5 and for −5; four places for +4 and –4; five for +3 and −3; six for +2 and −2; seven for +1 and −1 and eight for 0, on the grid, meaning the sorting of the cards was done in a forced manner by using a pre-determined grid. This procedure required participants to rank each item in relation to each other (Stenner et al. 2003), although there is no right or wrong sorting, as the individual sorts according to their personal point of view. Participants were instructed to adjust the cards on the grid until they were satisfied with the result.

We also conducted post-sorting interviews, asking open-ended questions and gathered comments from the participants to gain a deeper understanding of their opinions and beliefs about the covered statement. For instance, at times we requested additional examples or stories to obtain a more nuanced understanding of the participants’ views. The entire process took ~80–120 min. Most participants admitted that it was a rare opportunity to openly share personal thoughts about the topic of vaccination on such a comprehensive scale. They were eager to have a longer discussion, so sometimes a short break was taken in between the tasks: 1. sorting into three pillars and 2. ranking on the grid. The collected qualitative data were used in factor interpretation, but it is not used for in-depth analysis in the current article.

Analysis and interpretation

The main objective of Q analysis is to establish portions of shared meanings (i.e., factors), which identify a group of people who have sorted items in a very similar way (Watts and Stenner, 2012). The by-person factor analysis of collected Q sorts (i.e., ranking of the statements) was conducted using dedicated free software, Ken-Q Analysis 2.0.1. (Banasick, 2023).

We began the analysis with the calculation of the correlation matrix. In Q methodology, correlation provides a “measure of the nature and extent of the relationship between any two Q sorts and hence a measure of their similarity or otherwise” (Watts and Stenner, 2012, p. 97). Factors were then extracted from this matrix using the centroid factor extraction, as suggested by Watts and Stenner (2012). Q methodology uses centroid factor analysis to extract factors, which is a way of defining “centres of gravity embedded in a correlation matrix” (Brown, 1980, p. 40). According to Brown (1980, p. 40), “a centroid refers to a kind of grand average of the relationships between all the sorts, because they are represented by their correlation coefficients”.

We followed the guidance by Watts and Stenner (2012, p. 111) and calculated a factor’s eigenvalue by summing the squared factor loadings of all Q-sorts on that factor. We considered this important as it enabled us to provide “a clear indication of the strength and potential explanatory power of an extracted factor” (Watts and Stenner, 2012, p. 111). We used the Varimax rotation method in our statistical analysis as this rotation technique maximises the variance of squared loadings within each factor, helping to simplify their interpretation (Akhtar-Danesh, 2017).

A factor score, i.e., an estimate of the factor’s viewpoint, was then prepared “via a weighted average of all the individual Q-sorts that load significantly on the factor and that factor alone” (Watts and Stenner, 2012, p. 142). Following the advice by Watts and Stenner (2012, p. 143), we then converted the total weighted score for each item into a standard (or z) score, as this enabled us to make cross-factor comparisons (Stenner et al. 2003).

We began with factor interpretation to outline five factor arrays for our study. It appeared that the factors 3–5 describe only a very small percentage variance (factor 5 eigenvalue is almost 0%), i.e., the study participants shared only a very small common part in their values (Watts and Stenner, 2012). Since the eigenvalue is still <1, there is no justification for including factors 3–5, as an extracted factor with an eigenvalue of <1.00 accounts for less study variance than a single Q sort (Watts and Stenner, 2012). The same applies to 3- and 4-factor study results.

However, 2-factor study results described 43% of variance, which according to Watts and Stenner (2012) is a strong result, as anything in the region of 35–40% or above would ordinarily be considered a sound solution. Based on the common factors, two factors were extracted and rotated.

The factor weights were flagged manually; therefore Q-sorting’s nos. 5 and 18 were left out (confounding sorts), i.e., their significant factor loading was higher than the calculated limit (0.34) in both factors. Significant factor loading was calculated manually (Watts and Stenner, 2012).

The factor arrays provided the basis for different factor interpretations. The overall aim of factor interpretation was to explain the viewpoint captured by the factor and shared by the significantly loading participants. We used the crib sheet method (Watts and Stenner, 2012) for interpretation. This method enabled us to organise statements by how they were ranked relative to each other in the context of each factor. Following Watts and Stenner (2012), the crib sheet includes four basic categories and lists the highest ranking in the Factor 1 array (+5) and the lowest (−5), as presented in Table 2. Every statement was examined to achieve holistic interpretations of the factors.

Table 2 Relative ranking of statements in factor 1 and factor 2; consensus statements.
Full size table

Results

As a result of our analysis, two factors emerged, explaining 43% of the variance in our study. Twenty-seven of the 29 Q sorts loaded significantly on one or the other of these two factors (Table 3). Thus, according to the subjective understandings about the topic of vaccination, the factors or opinion types are: 1. Mainstream Medicine Non-Trusting Vaccine Sceptic; 2. Mainstream Medicine Trusting Vaccine Hesitant.

Table 3 Rotated two-factor matrix with an indicating a defining sort.
Full size table

Each factor is presented using factor scores together with the interviews to interpret and describe the factors to determine what kind of opinions were represented by factor 1 or 2, what unites or distinguishes them. The factors are interpreted at face value.

Factor 1: Mainstream medicine non-trusting vaccine sceptic

Factor 1 has an eigenvalue of 7.5, explaining 26% of the study variance. Table 2 presents the Factor Arrays of the participants loading on Factor 1 in comparison to the participants loading on Factor 2 to measure distinguishing and consensus statements between the two factors. Statements and the reflective Factor Array score will be presented in brackets to illustrate the interpretation of the study.

According to Table 2, the highest and lowest ranked statements, marked by participants loading onto Factor 1, reflect their established view that vaccines are harmful (16: “I believe that vaccines damage the natural immune system”; +5). These views are either denied or ignored by the mainstream media (1: “Media covers the topic of vaccination in a biased manner”; +5) and frowned upon by the society at large (12: “Anti-vaccine attitudes are not frowned upon in Estonian society”; −5). Although most individuals associated with Factor 1 strongly disagree with the statement that “All the vaccines described in the national immunisation programme of Estonia (state-funded programme) should be done” (13: −5) and are hence against all vaccines, this stance is not universal. Some participants loading onto Factor 1 are specifically against infant vaccines and believe that immunisation is more harmful in childhood (19: +3), as children’s immune systems are not yet fully developed. Individuals loaded on Factor 1 tend to agree with the statement that “a baby who is breastfed does not additionally need to be immunised to be protected from infections” (41: +3). In fact, all the participants from both factor groups who expressed hesitancy towards infant vaccines admitted that they became interested in the topic of vaccines when they had their first child, as they wanted to provide the healthiest care possible for their offspring.

Participants loaded on Factor 1 do not trust vaccines and find them harmful. They strongly believe that vaccines damage the natural immune system (16: +5), but details in their answers vary. Some participants tend rather to agree with the view that healthy people do not need to be immunised (18: +1) because their immune system is strong enough to protect them against diseases. At the same time, they also tended to agree that people with chronic diseases (17: +1) as well as children (19: +3) should not be vaccinated because their immune system is weak.

Factor 1 representatives do not associate vaccines with progress in healthcare regarding immunisation development; rather, they hold the opposite view. They do not believe that people currently live longer thanks to vaccinations (29: −4) but are convinced that adopting a healthy lifestyle and maintaining hygiene helps to prevent infections more effectively than vaccines (31: +4). Participants associated with Factor 1 perceive vaccination to involve high risk to one’s health compared to individuals loading onto Factor 2. Factor 1 representatives tend to disagree that vaccinations help to prevent diseases (23: −2), especially compared to Factor 2 representatives, who have full trust in general prevention statements. Furthermore, they rather tend to believe that by getting vaccinated, they are risking their health (15: +3). Participants loaded on Factor 1 agree that vaccines can cause different health problems, such as allergies (40: −2) and autism (33: +2), as they tend to believe that vaccines contain toxic substances (26: +2). Even microchipping of people during vaccination is a more realistic scenario for them (28: −1), compared to Factor 2 representatives who find it impossible.

As the individuals represented by Factor 1 do not see the real impact and value in immunisation, they are rather convinced that vaccination is beneficial only to big and powerful pharmaceutical companies, i.e., big pharma (25: +2), and rather agree with a statement that vaccination is a practice through which the state controls people (27: +1).

Participants associated with Factor 1 had made up their minds about vaccination years before the COVID-19 pandemic, and were forced to make acute decisions about vaccines. Interviews with the participants loading onto Factor 1 revealed that the COVID-19 pandemic did not influence their views about vaccines; rather, it became apparent to them that society did not accept their views and lifestyle. All of them claimed to have experienced hostility from former friends and colleagues who did not share their views about vaccination. These negative personal experiences also illustrate why participants associated with Factor 1 were strongly against the statement that “anti-vaccine attitudes are not frowned upon in Estonian society” (12: −5). Such personal negative experiences acquired during the pandemic are also the reason why Factor 1 sorters tend to position themselves in opposition to Estonian society and have adopted an “us vs. them” mentality. This may also explain why Factor 1 sorters do not feel the need to contribute to the collective good, as they disagree with the statement that “It is important for me to contribute to a healthier society by getting vaccinated” (37: −4). They do not tend to share any prosocial concerns, as they believe that unvaccinated people cannot harm others (38: −4). Such beliefs stem from the deep distrust participants associated with Factor 1 have towards state-issued vaccination guidelines (49: −3) and public health authorities in general.

Our findings reveal that participants associated with Factor 1 do not want to engage with the Estonian medical system on the topic of vaccines (53: “Estonian medicine is of good quality”; +19), especially compared to Factor 2 representatives who have full trust in the healthcare system and medical doctors. Factor 1 participants have not asked a medical doctor for advice on the topic (43: −3) and tend not to trust the advice given by the medical doctors they personally know regarding vaccines (45: −1). In fact, they are reluctant to show trust in the medical system in general (50: 0). At the same time, several participants associated with Factor 1 claim that they have no experience with the Estonian medical system, as they have not required any medical help. Some Russian-speaking participants also expressed that they believe the Estonian medical system to be less developed in comparison to the system in Russia, and tend to seek medical services there. The statement positioned at 0 reflects the low relevance of the mainstream medical system to the Factor 1 representatives’ lifestyle. This is likely related to the fact that the experiences participants associated with Factor 1 have had with the medical system have been perceived negatively (51: −1). During the interviews, some participants explained that at times they had not had necessary access to the medical system, while others had experienced negative attitudes expressed by doctors or had fallen victim to medication errors. Considering that the individuals associated with Factor 1 are highly critical of mainstream medicine, they have turned to alternative medicine in case of problems. In fact, they trust natural medicine and believe that it can help with various health problems (54: “Natural medicine provides help only in case of some health issues”; −1). Such trust in alternative options has thus become part of their lifestyle—some participants use alternative medicine services regularly, while other participants sorted on Factor 1 offer such products and services themselves.

Individuals associated with Factor 1 claim that they do not need advice on the topic of vaccines (47: 0), as they have conducted their own research and acquired expert knowledge to make informed decisions about vaccines. Factor 1 sorters perceive mainstream media to cover the topic of vaccination in a biased manner (1: +5), which they believe also drives hostility among the public. Although they admit that it is difficult to find objective information (4: “It is not possible to find objective information about vaccines”; 0), they have now found reliable sources: communities and resources, mainly from alternative media, from which to gather trustworthy and relevant information on the topic.

Factor 2: Mainstream medicine trusting vaccine-hesitant

Factor 2 has an eigenvalue of 4.8 and explains 17% of the study variance. Although the accompanying interviews revealed that their level of vaccine hesitancy and reasons differed, Factor 2 sorters tended to express their hesitancy towards one or several vaccines (e.g., COVID-19 vaccines) that were either recently included in the immunisation programme or less communicated to the public (Table 2).

Individuals loading significantly on Factor 2 expressed strong trust in the mainstream medical system and the efficacy of vaccines by ranking the statements “Estonian medicine is of good quality” (53: +5) and “Vaccinations help to prevent diseases” (23: +5) the highest. They also expressed trust in medical doctors by ranking the statement “I do not trust medical doctors” (50: −5) the lowest. Furthermore, the individuals loading onto Factor 2 are not prone to conspiratorial thinking; for example, they ranked the statement “Vaccination is a way to microchip people” (28: −5) the lowest.

Analysis of the Q sorting exercise revealed that individuals associated with Factor 2 agree that “one should vaccinate as little as possible and as much as necessary” (14: +4). They do not have established convictions on the topic; rather, each vaccination is considered on a case-by-case basis. Immunisation decisions, i.e., health-related issues, are very important to them, so they want to make as informed a decision as possible. Interviews with Factor 2 individuals indicate that they have many questions about vaccines and constantly search for information on the topic. However, they admitted that it is difficult to find objective information (4: −2). As individuals loading onto Factor 2 strive for a comprehensive overview, they find the information provided through official communication materials to be too general.

Participants associated with Factor 2 have significantly more respect and trust in the medical system and medical doctors compared to participants associated with Factor 1. For example, individuals sorted on Factor 2 declare that Estonian medicine is of good quality (53: +5) and strongly believe that vaccinations generally help to prevent diseases (23: +5) and that “people currently live longer, thanks to vaccinations” (29: +4). Overall, participants associated with Factor 2 agree that a healthy lifestyle and hygiene cannot be more effective in preventing infections than vaccines (31: −1) and have not excluded immunisation as an important part of healthcare progress. Such a belief is likely related to the fact that participants sorted to Factor 2 have had positive experiences with medical services (51: +4) and have full trust in medical doctors (50: −5). Hence, in comparison to individuals associated with Factor 1, the vaccine hesitancy of participants associated with Factor 2 is not related to a lack of trust in medicine, doctors, or public health authorities.

In comparison to individuals in Factor 1, Factor 2 sorters are also not prone to conspiracy thinking. For instance, they do not believe that vaccination is a way to microchip people (28: −5) and rather do not agree that only pharmaceutical companies benefit from vaccines (25: −3). Yet, like individuals associated with Factor 1, participants sorted on Factor 2 have doubts about vaccine side effects, e.g., they are uncertain about the role of vaccines in causing allergies (40: 0), or autism (33: 0). In short, they are not fully confident in the safety of vaccines.

Individuals associated with Factor 2 exhibit prosocial concerns and believe in the importance of collective responsibility. They tend to agree that unvaccinated people can harm others (38: +2) and thus also agree with the statement that “it is important for me to contribute to a healthier society by getting vaccinated” (37: +2). At the same time, interviews with the participants associated with Factor 2 indicate that their vaccine scepticism started or progressed during the COVID-19 pandemic when they began showing traits of antisystem sentiments. Some participants confessed that they felt their human rights were violated during the pandemic, while others claimed that the official communication by medical experts and public authorities was conflicting and threatening. They also viewed mainstream media coverage of the pandemic as one-sided and thus not trustworthy enough.

Areas of commonality amongst vaccine-hesitant groups

Although the analysis of the two factors was shaped by distinct viewpoints, there were also some areas of commonality between the views expressed, indicating considerable common ground amongst the vaccine-hesitant target group (Table 2). As shown in Table 2 negative Q-sort values (Q-SV) indicate disagreement with Q-statements and positive Q-SV indicate agreement with Q-statements. It is important to note that the stronger the agreement or disagreement with a statement, the more important the Q-statement was to the stakeholder group (Grimsrud et al. 2020).

Foremost, both factor solutions included a shared view that immunisation is a personal choice (36: +4). Hence, all individuals cherished individual freedom and their own autonomy in making vaccination decisions. All participants also claimed to take decision-making about immunisation very seriously. They expressed a vivid interest in health issues and claimed to have an increased need for health-related information. In fact, individuals associated with Factor 1 and Factor 2 agreed that they tend to double-check the claims they have heard or read about vaccination (F1 5: +3; F2 5: +1), indicating that our participants did not have blind trust in the information they read. At the same time, they tended to share an individualist epistemology by attributing the central role in immunisation decisions to themselves, both in obtaining knowledge on the topic and deciding what to trust. Both factor solutions also included a shared view that “media covers the topic of vaccinations in a biased manner” (F1 1: +5; F2 1: +3).

Both groups agreed that the harmfulness of vaccines tends not to be publicly discussed (F1 3: +4; F2 3: +2) and that the real side effects of vaccines are unknown (F1 22: +4; F2 22: +2). Interviews with the participants revealed that such views have grown stronger since the COVID-19 pandemic, when everyone who asked critical questions or expressed doubts about the vaccines was automatically labelled as antivaxxers. The damage done by such labelling is likely the reason why both factor solutions agreed that they have no interest in sharing their views on the topic, neither in mainstream media (F1 8: −3; F2 8: −4) nor on social media (F1 and F2 7: −2). Although all participants claimed that social media communities and groups have become safe havens for holding vaccine-sceptical discussions and it is easier to find different viewpoints expressed on social media, neither group believed that social media is a place for finding objective information about vaccinations (F1 2: −2; F2 2: −3). In fact, interviews revealed that they were rather critical of the content on social media, considering it to be full of emotional, biased, and non-factual information. This claim again illustrates their reflexive critical attitude characteristic of individualistic epistemology.

Discussion

Findings of our Q methodology study amongst vaccine hesitant individuals (N = 29) in Estonia, a country which has witnessed a steep decline in vaccination rates in recent years (de Figuereido et al. 2022), revealed two different perspectives on vaccination: 13 individuals associated with Factor 1, grouped as Mainstream Medicine Non-Trusting Vaccine Sceptics, and 14 participants associated with Factor 2, grouped as Mainstream Medicine Trusting Vaccine Hesitant.

Although the use of Q-methodology highlighted elements of diversity within the participants of our study, we would first like to emphasise the areas of consensus reflected in the degree of correlation between the factors. For example, like the findings of others (Helps et al. 2018; Kate et al. 2021), our participants exhibited an individualist epistemology, attributing a central role to the individual in obtaining knowledge and judging what is true, and having a sceptical attitude towards different information sources. Participants associated with both factors were eagerly seeking health information from various sources; hence, our results conflict with the findings of earlier studies (Gottlieb, 2016), which suggest that vaccine refusal is related to ignorance and lack of education. In fact, participants associated with Factor 2 claim that they need guidance on the topic, although they tend to hide their vaccine-sceptical views rather than approach experts for advice. Participants associated with both factors claimed that since the COVID-19 pandemic, there is a tendency to label everyone who has doubts or who asks critical questions about vaccines as antivaxxers. The media’s role in perpetuating age-old imaginaries (Stephenson et al. 2018) portraying “vaccine sceptics as gullible, ignorant and/or selfish” and framing “nonvaccinating” as a problem of individual’s wrong beliefs (Campeau, 2023) is also likely why all participants shared equally critical views about the mainstream media, which was believed to cover the topic of vaccines in a biased manner. Furthermore, our findings suggest that the polarised public debate on the topic of vaccines during the COVID-19 pandemic tended to drive especially those vaccine-hesitant individuals who loaded onto Factor 2 further into the margins.

Similar to previous research (cf. Wilson and Wiysonge, 2020), individuals associated with both factors considered social media a haven for vaccine-hesitant communities. However, neither group viewed social media platforms as places where objective information about vaccines could be found. Instead, all participants stated that there is too much non-factual and affective communication on social media, making it an unreliable source of information.

At the same time, our study participants exhibited considerable differences in their views about the mainstream medical system and medical doctors. Participants associated with Factor 1 had only minimal and often negative experiences with mainstream medicine, which led them to develop full trust in alternative approaches to health. Hence, our findings indicate similar tendencies to earlier research (Soveri et al. 2020), which has shown how low trust in doctors and negative attitudes towards vaccines can lead participants to turn to alternative medicine. Conversely, participants associated with Factor 2 demonstrated complete trust in the medical system and doctors. Their experiences with the medical system had been mainly positive, and they believed the Estonian medical system to be of good quality. This suggests that individuals associated with Factor 2 could be viewed as a group prone to changing their views about immunisation. Therefore, as noted by others (Helps et al. 2018), it is very important that clinicians maintain an empathetic relationship with non-vaccinating individuals and show a willingness to answer their questions and concerns, as this could lead to them reconsidering their vaccination decisions.

Individuals associated with Factor 1, however, would be difficult to convince to change their views through public health communication, as Factor 1 sorters revealed several views commonly associated with anti-vaccine rhetoric. For example, they tended to agree with claims about the dangerous side effects of vaccines, such as causing allergies or autism, and exhibited signs of conspiracy mentality (Bruder et al. 2013), such as the belief that vaccines only profit big pharma, all of which are associated with conspiratorial thinking.

Similar to previous research (Kärki, 2022), our findings show that vaccine-hesitant individuals are not a homogenous group. Furthermore, as noted by others (e.g., Peretti-Watel et al. 2019), our results reveal that vaccine hesitancy can range from strong anti-vaccination sentiments to having doubts about vaccines. Our findings thus provide important insights into the complexities of vaccine hesitancy, which could be used when developing new community engagement solutions in vaccination uptake planning (Burgess et al. 2021). In short, there is an opportunity for policymakers to support the decision-making process of vaccine-hesitant individuals by establishing respectful, label-free communication that considers the different information needs and perspectives of vaccine-hesitant groups.

Our findings reveal that individuals associated with Factor 1 tend to acquire information mainly from alternative media and social media groups, where their opinions and beliefs about vaccines and the medical system in general are shared by like-minded participants of this refracted public (Abidin, 2021). Establishing trust by avoiding confrontations, as well as simple pro-and-contra framing by medical experts, policymakers, and media, should thus be the first crucial step in starting the debate: break the topic into a broader range of different questions and ensure a neutral discussion environment. One way of initiating such interactions could be through digital microintervention programmes, which have proven to be effective information activism tools (Klaassen et al. 2024).

Individuals associated with Factor 2, however, are actively seeking health-related information and are open to advice from medical professionals. Our discussions revealed that Factor 2 representatives are not satisfied with the overly general, poster-sized information typically provided during public campaigns. Hence, we argue that, in addition to campaign-based communication, it is important to build long-term national educational programmes to raise individuals’ health literacy (Sørensen et al. 2012) and enhance resilience on the topic of immunisation in the context of informational disorder. As previous studies (Brewer et al. 2017) have found that presumptive approaches might not work with vaccine-hesitant individuals, easing their doubts about vaccines can only be achieved through a non-judgmental and collaborative communication style by authorities and medical doctors. Furthermore, considering that vaccine-hesitant individuals are a heterogeneous group, it is crucial for healthcare authorities to apply more nuanced and individually focused communication approaches.

Conclusion

Vaccine hesitancy is a public health challenge in many countries, especially after the COVID-19 pandemic. To address this complex phenomenon, the vaccine-hesitant target group and socio-cultural context must be studied and analysed by scholars and policymakers. As a result of our Q methodology study with 29 vaccine-hesitant individuals from Estonia, two clear factors (or types) of vaccine hesitancy emerged: Factor 1: Mainstream Medicine Non-Trusting Vaccine Sceptic, and Factor 2: Mainstream Medicine Trusting Vaccine Hesitant.

Factor 1 represents individuals with strong vaccine-hesitant views that are unlikely to change. Despite heightened interest in vaccines and health-related topics, they exhibit no trust in medical doctors or the healthcare system. They claim to have gained expert knowledge on the topic by consulting various information sources, mainly alternative media. Individuals belonging to Factor 1 are prone to conspiratorial thinking and, in case of health concerns, turn to alternative medicine. During the COVID-19 pandemic, representatives of Factor 1 felt that their views were frowned upon, and they felt outcast from society.

Factor 2 representatives do not consider themselves vaccine sceptics, but rather “ordinary people” who have doubts about vaccines. Their vaccine-hesitant views are prone to change and can be directed towards one or some vaccines. They exhibit high levels of trust in medical doctors and the medical system in general. They are highly interested in the topic of vaccines and actively seek information from a wide range of channels, but admit that objective information on the topic is difficult to find. During the COVID-19 pandemic, representatives of Factor 2 failed to receive information that would ease their doubts about vaccines. The polarised nature of the public debate only intensified their hesitancy, but in fear of being labelled anti-vaxxers, they refrained from publicly voicing their concerns.

Our findings indicate that there is a need for policymakers to establish long-term national educational programmes to develop citizens’ health literacy. Additionally, promoting non-judgmental, collaborative communication between vaccine-hesitant individuals and medical doctors, policymakers, and the media is crucial to prevent polarisation.