Vaccination conspiracy beliefs among social science & humanities and STEM educated people-An analysis of the mediation paths.

Understanding vaccine hesitancy is becoming increasingly important, especially after the global outbreak of COVID-19. The main goal of this study was to explore the differences in vaccination conspiracy beliefs between people with a university degree coming from different scientific fields-Social Sciences & Humanities (SH) and Science, Technology, Engineering and Mathematics (STEM). The study was conducted on an online convenience sample of respondents with college and university degrees in Croatia (N = 577). The results revealed that respondents educated in SH proved to be more prone to vaccination conspiracy beliefs. The indirect effect through science literacy was confirmed, while this was not the case for the indirect effects through health beliefs (natural immunity beliefs) and trust in the healthcare system. However, all three variables were important direct predictors of vaccination conspiracy beliefs. Female gender and religiosity were positively correlated with vaccination conspiracy beliefs, while age was not a statistically significant predictor. The authors concluded by emphasizing the necessity of the more theoretically elaborated approaches to the study of the educational and other socio-demographic differences in vaccine hesitancy.

Introduction

It has become commonplace to start a research paper that deals with vaccine hesitancy with a paradox that stems from the tension between the well-known and scientifically-proven benefits of vaccination and the rising tide of vaccine hesitancy. The paradox has led to an extensive corpus of research and various models that aimed to explain and systematize the main factors that lie behind vaccine hesitancy. For instance, according to the so-called 3Cs model proposed by the WHO EURO Vaccine Communications Working Group in 2011 and further elaborated by the SAGE Working Group on Vaccine Hesitancy and MacDonald [1], the factors can be divided into three main groups: (a) confidence (trust in vaccine safety and effectiveness, trust in healthcare providers and policy-makers), (b) complacency (health beliefs, perception of vaccine-preventable diseases as risk-free, thus making vaccination unnecessary), and (c) convenience (availability, affordability, accessibility, science and health literacy, etc.). Research confirmed the importance of the elements from the 3Cs model [e.g. 2, 3]. Given the focus of the current study and the multitude of research studies, we will only outline the main findings of selected elements from the 3Cs model pertaining to the institutional trust (confidence), postmodern health beliefs (complacency), and science literacy (convenience—“ability to understand”).

When it comes to the institutional trust, a great number of research studies established a link between trust in social institutions and the healthcare system on one side, and vaccine hesitancy on the other, especially in incident/critical situations. For instance, Lewis and Speers [4] linked the confidence crisis in the MMR vaccine in the UK with the lack of confidence in the executive power at the time, and Mesch and Schwirian [5] found that willingness to vaccinate against H1N1 in the USA during the 2008/2009 pandemic was negatively linked to confidence in the government’s ability to address this health crisis. Orr et al. [6] studying the example of the 2013 Polio outbreak in Israel, found that two of the five most important reasons to resist vaccination were related to trust in the government (general mistrust and the conviction that the government had obtained too many vaccines and wanted to get rid of them by vaccinating children). Similar conclusions about the importance of trust were put forward in other studies [7-Soc Sci Med. 2019 ">11].

Complacency also contributes to low vaccine acceptance [2]. With regard to the so-called postmodern health beliefs (preference for natural immunity, hesitancy with regard to the use of pharmaceuticals, etc.), Repalust et al. [12] found that the use of complementary and alternative medicine (CAM) was positively associated with vaccine hesitancy in Croatia. Similar correlation was established in the research of Pavić and Milanović [13], where postmodern health beliefs were associated with CAM use and suspicion in conventional medicine in Croatia, whereas other research also confirms the link between such type of health beliefs and vaccine hesitancy [Soc Sci Med. 2018 ">14-18]. On the other hand, as shown by Lorini et al. [19], the impact of health literacy, as a specific type of science literacy, on vaccine hesitancy is not clear, given that reliable studies which dealt with the topic were not numerous and that the studies used different measures and brought about mutually opposing results that depend on age, country and vaccine types.

Generally speaking, the relationship between educational level and vaccine hesitancy is non-consistent and thus non-conclusive, even though it seems that most of the research results confirmed negative correlation between educational level and vaccine hesitancy. For example, such results were obtained by Ritvo et al. [20] on a sample of the Canadian general population and Peretti-Wattel et al. [21] in a study conducted in France. On the other hand, Börjesson and Enander [22] determined that in Sweden it was the less educated who were more likely to be vaccinated, while Casiday et al. [23] did not establish the correlation between acceptance of MMR vaccine and education. Numerous other studies can be cited that shows a positive [24, 25] or negative [6, 26, 27] correlation between educational level and pro-vaccination attitudes and behaviors. However, such studies explore educational level in general, probably tacitly assuming that science, health or vaccine literacy/knowledge was the mediating mechanism.

In the current study, we wanted to go beyond just exploring the level of education itself and its correlation to vaccine hesitancy. Bearing in mind the possible epistemic differences between the fields of science, we aimed to examine whether differences in the fields of education (i.e., scientific fields) might influence vaccine hesitancy. In other words, our expectation was that the academic background of individuals educated in social sciences and humanities (hereinafter: SH), that are also known as “soft” and cognitively non-restrictive sciences, differed from the academic background of the individuals educated in natural science, technology, engineering and mathematics (hereinafter: STEM) fields, usually defined as “hard”, cognitively restrictive scientific fields [28]. Consequently, there should be different levels of vaccine hesitancy among individuals with a degree in these fields. Namely, our basic contention was that individuals educated in SH fields might be somewhat more skeptical, that is, more inclined to relativize and point out uncertainties relating to vaccines than individuals educated in STEM fields. Such differences can be the result of the different academic socializations and/or the individual characteristics which are important when choosing a field of study. Additionally, it is our contention that such differences will be mediated by some of the indicators coming from the 3Cs model (trust in the healthcare system, specific health beliefs, and science literacy). The basis for this assumption lies in the concept of socio-cognitive differentiation of scientific fields and their cultures, that is, in the epistemic differences between them, which we will elaborate in the next section of the paper. In addition, up to this moment, there are only few studies which dealt with this research topic, mainly confirming that vaccine hesitancy is higher among SH graduates [29-31]. However, the factors behind the established differences are completely unexplored.

Therefore, we proceed with a summary of possible epistemic differences between social sciences and humanities on one side, and science, technology, engineering and mathematics on the other. Then we link these differences to the aforementioned 3Cs factors influencing vaccine hesitancy. In short, we aim to explore selected 3Cs indicators as possible mediating mechanisms arising from the epistemic differences between the scientific fields.

Epistemic differences and vaccine hesitancy

When it comes to fields of science, there have been many disputes in relation to (1) importance and legitimacy of knowledge in different fields, (2) methodological and epistemic standards and their possible unification, and (3) objectivity of the scientific knowledge in general. Some of these differences between the scientific fields can be grounded in a concept called „epistemic cultures”[32] or related concepts that indicate the socio-cognitive differentiation of sciences, that is, scientific fields [33], Becker’s academic tribes and territories [34], Biglan’s distinction of soft and hard scientific fields [28, 35], etc. These concepts, mostly conceived in the 1970s and 1980s and developed later, deal with differences in cognitive “objects”, “styles”, “structures” and “cultures”. If we try to make a rough cross-section of these theoretical approaches, it could be argued that STEM fields are characterized by a more rigid cognitive style, routine research and protocolisation, epistemological realism, an emphasis on the objectivity and superiority of scientific methods and practices, and an understanding of scientific knowledge as a cumulative endeavor. On the other hand, SH fields are characterized by theoretical pluralism, discursiveness, less rigid cognitive style, more open and uncertain findings of research, that is, epistemological relativism and greater criticism of scientific methods and practices.

To illustrate the aforementioned differences, a study conducted on a sample of Croatian scientists in natural and social sciences speaks in support of the differences in the understanding of scientific objectivity between the different fields [36]. In this study it was established that natural scientists were firmly convinced of the objectivity of their disciplines, and that they had high confidence in scientific methods and procedures, wherein they emphasized reproducibility and measurement. In contrast, the author of the study noted that relativism was more common among social scientists, who more often pointed out that subjectivity was somewhat inevitable, showed greater skepticism about the omnipotence of research rules and methods, are more often suspected of the notion that objectivity can be achieved, or even in principle deny the possibility of objectivity [36]. In other words, the author summarized that the inclination to positivism and the idea of the fully objective science can be more often found among researchers coming from natural sciences. On the other hand, different ideas and cultural values were more often present in social sciences and humanities. By definition, these sciences explore the human being and human societies, thus encouraging self-awareness, self-exploration and self-expression that reject the exclusive domination of the medical science (and natural sciences in general) when it comes to delineating what is and what is not a legitimate health issue. On the other hand, STEM educated people are socialized into an epistemic culture that promotes the more objective epistemic standards. For instance, biomedicine-trained individuals embrace the experimental epistemic culture as the „gold standard“, thus often rejecting the social sciences methods, such as surveys and qualitative methods, as insufficiently rigorous [37].

The differences in vaccine hesitancy might be partly explained by a different role that trust and criticism play in the STEM and SH epistemic cultures. Social sciences and humanities in general are characterized by epistemic cultures with high levels of context-dependence and with paying close attention to complexity and uncertainty. In consequence, it can be assumed that the people with an SH degree would be more prone to criticism in relation to scientific objectivity, and therefore less willing to accept a position of a layman who accepts the scientific results without a critical stance. When it comes to the issue of trust in social institutions, as Kuhn notes [38], social sciences find themselves in an ambivalent position as related to the social elites. On the one hand, social sciences share some metaphysical assumptions, such as the assumption of man’s unsocial nature that must be socialized and tamed in order to produce a good society. On the other hand, social sciences reflect critically on current social processes and acknowledge the discrepancy between the social “facts” and the ideal and promised society, wherein particular social institutions and social elites can be heavily criticized for not delivering social services in the expected way.

Research goal and hypotheses

The main goal of the current study was to connect the aforementioned differences between the „hard”and „soft”sciences with vaccine hesitancy. To be more precise, the hitherto research of vaccine hesitancy explored the impact of educational level only in general. However, the above-mentioned considerations lead us to believe that the difference in epistemic cultures might exert a significant impact on vaccine hesitancy, in particular vaccination conspiracy beliefs. Therefore, our goal was to fill in the gaps of the research so far by determining possible differences between science fields/fields of study when it comes to vaccine hesitancy, and to explore proximate cause of such differences. With this in mind, our aim was to explore possible differences in vaccination conspiracy beliefs as a specific type of vaccine hesitancy between university educated people graduating in different science fields. Consequently, our hypotheses were stated as follows:

H1. Respondents who graduated in STEM fields have lower vaccination conspiracy beliefs in comparison to the respondents who graduated in SH fields.

H2. Trust in healthcare system, natural immunity beliefs and science literacy are the mediators of the differences in vaccination conspiracy beliefs described in H1.

Our H2 represents a parallel multiple mediation model which can be visualized by means of the following conceptual diagram (Fig 1). Here we emphasize that in order to preserve visual clarity we omitted control variables (gender, age, and religiosity) from the figure. In other words, it is important to note that the mediation from H2 is estimated while simultaneously controlling the above-mentioned demographic variables. Consequently, the possible mediating influences cannot be attributed to the confounding effects of the demographic variables since they are controlled for in the statistical model.

Fig 1

Conceptual diagram of the parallel mediation tested in the study.

Methods and instruments

A convenience sample consisted of college and university educated people in Croatia (N = 577), and an online questionnaire was used as the data-collection tool. The participants were recruited with snowball method. The initial sample size was 624, but 47 respondents (7.53%) had missing values on some of the study variables. We conduced binary logistic regression with the missing value as the criterion variable and did not find any significant relationship with other study variables. That is why we proceeded with the complete case analysis, which reduced our sample from 624 to 577 respondents. The data were collected in June 2019 within a pilot study of vaccine hesitancy conducted by the authors. Namely, only the answers of the respondents from the pilot study who graduated from a university were used in the current study. The sample comprised 345 women and 232 men (59.79% and 40.21%, respectively). There were 315 SH and 262 STEM graduates (54.59% and 45.41%).

As for the delineation of SH and STEM fields, the official classification by the Ministry of Science and Education of the Republic of Croatia was used [39].

Informed consent was obtained by means of a written online form. Participants could indicate that they have read the description of the study, were over the age of 18, and that they agreed to participate. Participants did not receive any compensation for completing the survey, Members of the Ethics Committee of the Institute for Social Research Zagreb agreed that the project proposal (Protocol 1112/2017) was in accordance with ethical principles and rules of conduct highlighted in the ISRZ Code of Ethics, as well as the rules of conduct of research in social sciences in general.

As for the measurement scales, vaccination conspiracy beliefs were measured by the scale constructed by Shapiro et al. [40], which comprises seven items. All items were answered on a 5-point scale ranging from 1 = completely disagree to 5 = completely agree. The descriptive statistics for the scale items are presented in Table 1. The results from the items were added in order to get the total summary score (M = 15.11, SD = 8.44; MSH = 16.27; SDSH = 8.83; MSTEM = 13.72; SDSTEM = 7.74; t = 3.70: p = 0.00) to be used in the subsequent analyses with higher values representing higher belief in vaccination conspiracies. Internal consistency of the scale as the measure of reliability (Cronbach’s alpha) was 0.967.

Table 1

Descriptive statistics for the vaccine conspiracy beliefs scale.

Item	M (SD)	Corrected Item-Total Correlation	Skewness	Kurtosis
Vaccine safety data is often fabricated	2.38 (1.32)	.78	.70	-.66
Immunizing children is harmful and this fact is covered up.	1.92 (1.28)	.91	1.29	.42
Pharmaceutical companies cover up the dangers of vaccines.	2.53 (1.37)	.87	.55	-.95
People are deceived about vaccine efficacy.	2.04 (1.30)	.91	1.08	-.08
Vaccine efficacy data is often fabricated.	2.05 (1.29)	.89	1.09	.00
People are deceived about vaccine safety.	2.24 (1.34)	.93	.80	-.65
The government is trying to cover up the link between vaccines and autism.	1.96 (1.31)	.89	.10	.08

A three-item scale of natural immunity beliefs was taken from the Martin and Petrie [41], Vaccination Attitudes Examination (VAX) scale. Those items were answered on a five-point scale ranging from 1 = completely disagree to 5 = completely agree, as well. These three items were also summed in order to get the total summary score (M = 6.60, SD = 3.63; MSH = 6.92; SDSH = 3.83; MSTEM = 6.23; SDSTEM = 3.34; t = 2.32; p = 0.02) since the internal consistency, as measured by Cronbach’s alpha, was high (0.93). Higher values represent higher belief in natural immunity. The descriptive statistics for the items are shown in Table 2.

Table 2

Descriptive statistics for the natural immunity beliefs scale.

Item	M (SD)	Corrected Item-Total Correlation	Skewness	Kurtosis
Natural immunity lasts longer than a vaccination	2.32 (1.37)	.83	.70	-.66
Natural exposure to viruses and germs gives the safest protection	2.26 (1.28)	.87	1.29	.42
Being exposed to diseases naturally is safer for the immune system than being exposed through vaccination	2.03 (1.23)	.87	.55	-.95

The trust in healthcare system was measured by using a revised version of the nine-item scale constructed by Shea et al. [42], To be more specific, we left out the item about the equal treatment of patients of all races and ethnicities given that the Croatian society is largely mono-ethnic and mono-racial. All the items were answered on a five-point scale ranging from 1 = completely disagree to 5 = completely agree. Even though Shea et al. [42] confirmed that the scale can be divided into two sub-scales (value congruence and competence), in the current study this scale proved to be one-dimensional with a high level of reliability (Cronbach’s alpha was equal to 0.90). The descriptive statistics for the items is shown in Table 3, whereas the total summary score was also used in subsequent analyses (M = 23.82, SD = 6.37; MSH = 23.28; SDSH = 6.34; MSTEM = 24.46; SDSTEM = 6.35; t = -2.22; p = 0.03) with higher values representing higher trust in healthcare system.

Table 3

Descriptive statistics for the trust in healthcare system scale.

Item	M (SD)	Corrected Item-Total Correlation	Skewness	Kurtosis
The Health Care System does its best to make patients’ health better	2.74 (1.13)	.67	.17	-.86
The Health Care System covers up its mistakes*	3.57 (1.00)	.65	-.56	-.07
Patients receive high quality medical care from the Health Care System	2.98 (0.97)	.68	-.31	-.59
The Health Care System makes too many mistakes*	3.15 (0.97)	.70	.04	-.38
The Health Care System puts making money above patients’ needs*	3.15 (1.09)	.70	-.01	-.73
The Health Care System gives excellent medical care	2.58 (0.97)	.68	.02	-.68
The Health Care System lies to make money*	2.52 (1.09)	.73	.52	-.23
The Health Care System experiments on patients without them knowing*	2.10 (1.17)	.60	.84	-.22

* Reversely scored items.

In the current study the items from the so-called Oxford scale of science literacy [43, 44] were used. The motivation for using a general science literacy scale instead of constructing a specific vaccine knowledge scale lies in our judgement that the specific scale would have to be composed of items that are themselves too „contaminated”with attitudes towards vaccines. For instance, the scale developed and applied by Cvjetkovic et al. [29] contained items about the connection between vaccines and autism and about the dangers of applying multiple vaccines at the same time. These items, although relevant, are strongly intertwined with vaccine attitudes and can be thought of more as a measure of vaccine attitudes than of knowledge. In our view, the Oxford scale measures „non-problematic”items and therefore, in spite of its general nature, is a more valid measure of science knowledge in our case. The Oxford scale arose as a joint effort of several researchers [43, 44] used in the report Science and Engineering Indicators prepared by the U.S. National Science Board. Some of the questions were also used in several Eurobarometer surveys [45, 46]. The scale consisted of 15 questions that probe into knowledge about general science facts, and the respondents have to mark them as „true“, „untrue“, „don’t know”answer. Both incorrect and „don’t know”answers were merged into one category, while correct answers comprised the other category making this variable dichotomous. The frequencies of answers on individual items are given in Table 4. The results were summed in order to obtain the total score (M = 12.30, SD = 2.42; MSH = 11.65; SDSH = 2.60; MSTEM = 13.00; SDSTEM = 2.02; t = -6.82; p = 0.00), with higher values indicating higher science literacy. Internal consistency of the variable (Kuder-Richardson—KR20—coefficient) was reasonably high (0.70).

Table 4

Descriptive statistics for the science literacy scale.

Item	Correct answer (%)	Incorrect answer, don’t know answer (%)
Antibiotics kill viruses as well as bacteria	87.69	12.31
The Sun goes around the Earth	83.54	16.46
The center of the Earth is very hot	90.81	9.19
The oxygen we breathe comes from plants	88.21	11.79
The earliest human beings lived at the same time as the dinosaurs	85.44	14.56
By consuming genetically modified fruit we alter our genes	72.77	27.73
All radioactivity is man-made	85.96	14.04
It is the mother’s genes that decide whether the baby is a boy or a girl	75.91	24.09
More than half of human genes are identical to those of mice	50.26	49.74
Electrons are smaller than atoms	85.44	14.56
Lasers work by focusing sound waves	66.38	33.62
It takes one month for the Earth to go around the Sun	90.47	9.53
Radioactive milk can be made safe by boiling it	88.39	11.61
The continents on which we live have been moving their location for millions of years and will continue to move in the future	93.93	6.07
Human beings, as we know them today, developed from earlier species of animals	81.63	18.37

Finally, sociodemographic variables like age, gender, religiosity were also collected. Age was measured by asking participants’ year of birth and then re-coding it into the age measured in years. Gender was measured as female and male, while religious identification was measured from 1-non religious to 6- very religious. Average age was 39.01 years (min = 20; max = 73; SD = 9.25), while the average result on the 1–6 religiosity scale was 3.03 (SD = 1.61).

All the instruments were translated into Croatian, and then back-translated in order to check the validity of the translation. Appropriate changes were made accordingly.

Results

With the purpose of gaining an initial insight into the bivariate relations between the study variables, in Table 5 we present the intercorrelation matrix. Among other things, it can be observed that the higher level of vaccine conspiracy beliefs can be found among women (r = 0.18; females = 16.33, males = 13.29, Cohen’s d = 0.37), more religious people (r = 0.36), people with higher beliefs in natural immunity (r = 0.81), people with lower trust in healthcare system (r = —0.66), and people with lower scientific literacy (r = —0.48). With respect to the field of study (H1), participants with degrees in social sciences and humanities are more likely to have vaccine conspiracy beliefs (r = 0.18; SH = 16.27, STEM = 13.72, Cohen’s d = 0.31).

Table 5

Intercorrelation matrix (Pearson’s correlation).

Variable	Gender	Age	Religious identification	Field of study	Natural immunity beliefs	Trust in healthcare	Scientific literacy	Vaccine conspiracy
Gender	1	- .08*	. 21**	- .17**	.17**	- .12**	- .26**	.18**
Age	- .08*	1	.15**	- .03	- .07	. 05	.06	- .04
Religious identification	. 21**	.15**	1	- .02	.31**	- .21**	- .32**	.36**
Field of study	- .17**	- .03	- .02	1	- .10**	.09*	.27**	- .15**
Natural immunity beliefs	.17**	- .07	.31**	- .10**	1	- .55**	- .45**	. 81**
Trust in healthcare	- .12**	. 05	- .21**	.09*	- .55**	1	.31**	- .66**
Scientific literacy	- .26**	.06	- .32**	.27**	- .45**	.31**	1	- .48**
Vaccine conspiracy	.18**	- .04	.36**	- .15**	. 81**	- .66**	- .48**	1

Notes:

*p < .05,

**p < .01;

Gender: 0—Male, 1—Female; Field of study: 1—SH, 2—STEM.

In order to test total, direct and indirect (mediation) effects of the field of study on vaccine conspiracy beliefs (H2), we used PROCESS, a mediation and moderation SPSS macro written by Hayes [47]. Given that PROCESS is a regression-based model, first we present the regression analysis with conspiracy beliefs as the criterion variable (Table 6).

Table 6

Linear regression analysis with the result on the vaccine conspiracy beliefs scale as the criterion variable.

Variable	B	SE	t	p	LLCI	ULCI
Intercept	17.90	1.90	9.41	.00	14.16	21.64
Gender	- .10	.39	- .26	.79	- .87	.66
Age (in years)	.03	.02	1.27	.20	- .02	.06
Religious identification	.50	.12	4.05	.00	.26	.74
Field of study	- .70	- .38	- 1.83	.07	- 1.45	.05
Natural immunity beliefs	1.34	.07	20.71	.00	1.21	1.47
Trust in healthcare system	- .40	.03	- 11.07	.00	- .45	- .31
Scientific literacy	- .33	.09	- 3.69	.00	- .50	- .15

Notes: R2 = 0.74; F = 202.42; p = 0.00; Gender: 0—Male, 1—Female; Field of study: 1—SH, 2—STEM; LLCI—lower limit of the confidence intervals; ULCI—upper limit of the confidence intervals.

In comparison to SH graduates STEM graduates score 0.70 points lower on the vaccine conspiracy beliefs scale. A one-unit increase in the natural immunity beliefs on average raises the results on the vaccine conspiracy belief scale by 1.34 points when all other variables are held constant. A one-unit increase in trust in the healthcare system on average lowers the results on the vaccine conspiracy belief scale by 0.40 points. Finally, a one unit increase in science literacy decreases vaccine conspiracy belief by 0.33 points.

As for the demographic variables (gender, age, and religious identification), we can observe that religious identification is a statistically significant predictor, whereas age and gender are not significant predictors when all other variables are entered into equation. Here we can once more emphasize that there is a bivariate correlation between gender and vaccine conspiracy beliefs, with women scoring 3.04 points higher than men (Cohen’s d = 0.37), but this difference disappears in the multivariate analysis. We can also note that a one-unit increase in religious identification raises the result on the vaccine conspiracy belief scale by 0.50 points when the other variables are held constant.

After the full regression model, we proceeded with the mediation analysis. In Table 7, total, direct and indirect effects of field of study are presented, with age, gender and religiosity entered as covariates. The covariates were used in order to check for the possible spurious effects of our predictors. As suggested by Hayes [47], bootstrapping was used as a method of obtaining confidence intervals of the estimates of the indirect effects and the pairwise differences between indirect effects.

Table 7

Vaccine conspiracy beliefs—Mediation analysis.

Total effect
Coeff.	SE	t	LLCI	ULCI
-2.18	.66	- 3.30**	- 3.47	- 0.88
Direct effect
Coeff.	SE	t	LLCI	ULCI
- .70	.38	- 1.84	- 1.45	.05
Indirect effects
	Coeff	BootSE	BootLLCI	BootULCI
Natural immunity beliefs	- .71	.38	- 1.48	.01
Trust in the healthcare system	- .37	.21	- .78	.02
Science literacy	- .39	.12	- .63	- .18

Note:

*p < .05,

**p < .01;

LLCI—lower limit of the confidence intervals; ULCI—upper limit of the confidence intervals.

We can observe that the total effect coefficient amounts to—2.18, meaning that STEM graduates score 2.18 point lower on the vaccine conspiracy beliefs scale when compared to SH graduates, and when both direct and indirect effects are accounted for. The coefficient for total effect can be decomposed into the direct effect (- 0.70) and the indirect (mediation) effect (-1.45), with the latter including indirect effects of natural immunity beliefs (- 0.71), trust in healthcare system (- 0.37) and science literacy (- 0.39). The direct effect is not statistically significant, while only the indirect effect of scientific literacy proved to be significant by means of the bootstrapping analysis. In the remaining two cases, the bootstrapping analysis of indirect (mediating) effects revealed that zero effect is included into the confidence intervals. All bootstrapping confidence intervals of the pairwise differences between mediation effects also contained zero difference, thus not being able to confirm the existence of the pairwise differences between the indirect effects.

Discussion

In pursuance of the research goal, in this paper we proposed two hypotheses. The first one (H1) was related to the differences in vaccination conspiracy beliefs between SH and STEM fields, and the second one (H2) to the possible mediating mechanism of such differences.

To summarize the findings pertaining to H1, even though both groups mostly largely reject vaccination conspiracy beliefs, our findings confirm that the people with an SH degree are more prone to vaccination conspiracy beliefs than the people with a STEM degree, thus confirming H1. The bivariate difference equals 2.55 points, it falls to 0.70 when demographic variables (gender, age, and religiosity) are accounted for, but it also rises to 2.18 when both direct and indirect effects are included. Bearing in mind the lack of similar studies, we cannot directly compare our results with other research. We can confirm that our results are in concordance with the results obtained by Cvjetkovic et al. [29], who determined that the law students of the Belgrade University (Serbia) had substantially more negative vaccine attitudes than the engineering, and especially medical students, even while controlling for vaccine knowledge. However, since Cvjetkovic et al. did not conduct a full mediation analysis, their results cannot be directly compared to our results with regard to the indirect effects. Likewise, a study conducted by Šálek et al. [31] on a sample of medical and teacher education students in the Czech Republic found differences in the positive vaccine perception rate between those two groups. Vaccine acceptance rate was lower among teacher education students (72%) than among medical students (92%). Also teacher education students more often reported negative experiences with vaccination and were more often alternative medicine followers.

On the other hand, the mediation analysis conducted within our study confirmed the mediation effect of scientific literacy, while the mediation effect of natural immunity beliefs and the trust in healthcare system were not reliably confirmed, thus only partially confirming our second hypothesis (H2). Notwithstanding that fact, we emphasize that all mediation factors are important direct predictors of vaccine conspiracy beliefs. The data employed in the study just did not confirm the mediation function between the field of study and vaccination conspiracy beliefs in case of the two mediators. Another word of caution is also warranted here. As mentioned before, a parallel multiple mediation model with control variables was tested in our study. This means that mediation paths were tested with simultaneously controlling the effects of gender, religiosity and age. In additional analyses we confirmed that all mediation paths were statistically significant when gender was not employed as a control variable, while in that case indirect effects of natural immunity beliefs, trust in the healthcare system and science literacy amount to—0.87,—0.43 and—0.42, respectively. Consequently, the mediation paths can be explained by the disproportionate presence of women in STEM fields and the correlations between gender on one side, and vaccine conspiracy beliefs and the mediating factors on the other side, as visible from the intercorrelation matrix. In other words, all three mediators are indeed significant, but in cases of two of them (trust in healthcare system and natural immunity beliefs) this is the case only because women on average happen to have lower trust and higher beliefs in natural immunity, as well as higher vaccination conspiracy beliefs, while at the same time being disproportionally more present in social sciences and humanities.

The direct impact of the natural immunity preference on vaccine conspiracy beliefs is consistent with the results of the study of vaccine hesitancy in Croatia [12], which showed that the use of complementary and alternative medicine (CAM) was positively connected to vaccine hesitancy and vaccine refusal, as well as with a research conducted on a sample of Croatian population [13] that showed a negative connection between „postmodern”views on health (natural immunity included) and trust in conventional medicine as opposed to the trust in CAM. Positive attitudes towards CAM and its use partially follow from the general question of the legitimacy of the medical science authority. Namely, as Sointu [48] summarized, CAM users defy the biomedical definition of health by arguing for holistic health that empowers them by providing a sense of agency, control and meaning. Therefore, CAM produces a new selfhood and subjectivity that is directed towards exploring the inner-self. However, in the current study we could not reliably confirm that such differences represent a mediating factor of the established connection between the field of study and vaccination conspiracy beliefs. Namely, although the mediation effect was present in the hypothesized direction, its statistical significance could not be demonstrated by the bootstrapping analysis. Nevertheless, given the theoretical plausibility of such differences in epistemic culture and the fact that the bootstrapping interval included zero only by a small margin, we think that new studies which will employ random sampling are warranted.

Similarly, we also could not reliably confirm the mediation effect of the trust in the healthcare system with our study data. The possible explanation might be that the aforementioned criticism penetrated all fields of study and that it does not explain the differences in vaccine hesitancy. Perhaps we should not overstate the differences/gap between natural sciences and social sciences/humanities, since objectivity is nevertheless an important scientific ideal in (all) sciences [49]. Especially in the social sciences, methods and precision of natural sciences continue to be an inspiration to be emulated for some of the scientists from this field, but who are still aware of a distinctiveness of the social sciences. Conversely, in their qualitative study of science and non-science (i.e. humanities) tertiary educated people, Quinn et al. [50] established that even people who accepted the scientific method of evidence-based medicine tended to exhibit various „habits of mind“, such as the mistrust of authority, open-mindedness, skepticism, and rationality/belief systems that led them to engage in the acceptance and use of CAM. Thus, as Quin et al. concluded, scientific literacy can go hand in hand with belief systems that are prone to potential distrust towards science and scientists. In other words, a high level of science knowledge is not always translated into support for science [51], that is, individuals can use knowledge in different ways, according to their pre-existing interests and motivations [52], or educational epistemic culture. Additionally, it should be emphasized that within the scientific fields there are considerable differences among the various disciplines, and that there are various paradigms and methodological approaches within the same discipline. However, despite the fact that our mediation analysis did not confirm the impact of this factor, we think that the idea should be tested in future studies conducted on different samples, given that the bootstrapping interval from our study in this case as well comprised zero only by a small margin.

We posited that the aforementioned epistemic differences and the obvious assumption that people with a degree in SH should be less familiar with the knowledge coming from natural sciences, biomedical sciences included, could also explain the differences in vaccine hesitancy. In their meta-analysis of the connection between science literacy and science attitudes, after controlling for several potential confounding variables, Allum et al. [53] found a „small but consistent”correlation that gave some support to the deficit model of the relationship between science knowledge and attitudes. A study in Croatia on a representative sample also indicated a significant but weak relationship between the level of scientific literacy and (positive) attitudes toward science [54]. This is consistent with the results of the current study, in that science literacy is a statistically significant predictor of vaccine conspiracy beliefs, as well as the mediator of the SH-STEM differences in vaccine hesitancy. In other words, the results of the current study confirm the above mentioned “small, but consistent” relationship, given a partial mediation of the differences in the vaccine conspiracy beliefs between the STEM and SH fields through science literacy.

Aside from the discussion pertaining to H1 and H2, we shall also present a short discussion of the results with regard to other sociodemographic variables. We should, however, bear in mind that our study was conducted on a sample of higher-educated individuals, thus not being fully comparable to the results obtained in studies on general population research samples.

To start with, we have not confirmed the relationship between age and vaccine conspiracy beliefs. However, this relationship is highly varied and far from being conclusive. For instance, our results differ from the results obtained by Repalust et al. [12] since in their study the younger age groups were less vaccine acceptant in comparison to the older age groups. However, other research can be cited that determined a positive [55], negative [56], or non-existent [57] relation between age and various measures of vaccine hesitancy. These differences might be the result of age differences in relation to education, health beliefs, trust in social institutions etc. However, to our best knowledge, age differences in vaccination attitudes so far have not been comprehensively theoretically explained.

With regard to the established positive connection between religiosity and vaccine conspiracy beliefs, the results of the current study are mainly consistent with the results obtained by other studies [12]. Possible reasons for the relationship between religiosity and negative vaccine attitudes might include a latent conflict between science and religiosity within the Croatian citizens’ value system or, more specifically, the framing of biomedical questions as religious issues, dominated by the idea that medical science should not meddle too much into „God’s work“. More specifically, religious people might object to the fact that human and animal tissues, such as cell lines from aborted fetuses and hydrolyzed gelatin obtained from animals, are being used in the production of some vaccines [58]. As for the Catholic Church, being the dominant religious community in the Croatian society, according to the document drafted in 2005 by the Pontifical Academy for Life, the use of vaccines with the aborted fetuses’ tissue constitutes „at least a mediate remote passive material cooperation to the abortion“. Catholics are thus obliged to push for morally acceptable alternatives in the vaccine production, even though the use of vaccines should be tolerated in order to save children’s lives. Such a situation presents Catholic parents with a „moral coercion”that must be resolved as soon as possible [59].

When it comes to gender, our results are in line with the majority of studies indicating that vaccine hesitancy is more likely to be found among women [17, 60], even though inverse findings can also be found [61]. In addition, the results of our study suggest an interesting finding that the gender composition of scientific fields can be a part of the explanation of their epistemic differences and its impact on vaccine hesitancy. These results are in line with preponderance of women in the CAM, both as users and practitioners [62], as well as with the positive relationship between CAM beliefs and practices and vaccine hesitancy [63].

Conclusion and limitations

The results of our preliminary study suggest that future research on vaccine hesitancy should differentiate not only between different education levels, but between education/science fields as well. Namely, the people with a degree in SH proved to be more prone to vaccination conspiracy beliefs, whereas this difference is probably mediated at least by the differences in science literacy. The reasons could probably be found in the different academic socialization and the background characteristics which are important when selecting a field of study. Future studies should also further explore the possible mediators of this difference and establish some more in-depth theoretical explanations for their link with epistemic differences between scientific fields. In other words, when it comes to education and other sociodemographic factors as well, more comprehensive research is warranted that will go beyond simple descriptions of sociodemographic differences in vaccine hesitancy, in order to draw more complete conclusions, as well as possible recommendations in terms of science/health education.

Even though we did not confirm that beliefs in natural immunity and trust in the healthcare system are mediating paths of the vaccination conspiracy beliefs differences between SH and STEM, these constructs proved to be strong predictors of vaccination conspiracy beliefs. In the context of the COVID-19 pandemics this finding suggests that vaccine hesitancy cannot be contested only by providing factual, scientific information on COVID-19 vaccines. Successful COVID-19 crisis management needs to include a trust-building communication, and to address lay beliefs about vaccine and health in general.

In our view, the main weakness of the current study represents the convenience sample employed, as well as the data-collection method (online survey) employed in the study. It can be reasonably assumed that some amount of self-selection among potential study participants might be present in the study. Therefore, we cannot exclude the possibility that some sort of bias is shared between the study variables that might influence their relationship. It might also be assumed that the variance of the measurements was somewhat restricted, that is, the less (or more) vaccine hesitant people might have been more inclined to participate in the survey which might have influenced our results and conclusions. As for the measurements employed in the study, more specific measures of vaccine knowledge might bring different results than general measures of science knowledge such as the Oxford scale used in the current study. In addition, given that our study was cross-sectional and the possible omitted variables bias might have existed, causal mechanisms cannot be fully supported by the data from our study.

(SAV)

Click here for additional data file.

(DOCX)

Click here for additional data file.

20 Oct 2021

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: https://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols.

We look forward to receiving your revised manuscript.

Kind regards,

Kristina Hood, Ph.D.

Academic Editor

PLOS ONE

Journal Requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

2. Please amend your manuscript to include your abstract after the title page.

3. Please include your full ethics statement in the ‘Methods’ section of your manuscript file. In your statement, please include the full name of the IRB or ethics committee who approved or waived your study, as well as whether or not you obtained informed written or verbal consent. If consent was waived for your study, please include this information in your statement as well.

4. Please include captions for your Supporting Information files at the end of your manuscript, and update any in-text citations to match accordingly. Please see our Supporting Information guidelines for more information: http://journals.plos.org/plosone/s/supporting-information.

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: Partly

Reviewer #3: Partly

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: I Don't Know

Reviewer #2: Yes

Reviewer #3: No

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: No

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: Thank you for the opportunity to review the manuscript entitled ‘Vaccination conspiracy beliefs: differences between persons educated in different scientific fields and their mediating mechanisms'.

The author(s) present a study in which they investigated vaccine conspiracy beliefs in regard to educational background – social science and humanities versus stem. It is an interesting take on the role of education in vaccine conspiracy beliefs, because education is in this kind of research mostly investigated in terms of educational levels and not field differences.

I found the manuscript to be well written, and it contributes with new insights. Nevertheless, I recommend some changes which can contribute to an even greater quality of it.

Title

I suggest renaming the title to emphasize your main finding and make it more interesting for potential readers, some examples are: ‘Vaccine conspiracy beliefs stronger in social sciences and humanities then in STEM – a correlational study’.

Abstract

Start with an introduction sentence.

Introduction

Page 4, lines 76-79, add references for describing the SH and STEM fields.

Throughout the manuscript replace ‘persons’, perhaps with ‘people’, although I am not sure this is the adequate term too.

Research goals, questions, and methods

Please insert a paragraph that in details describes the procedure of how and when the study was conducted - where was the survey shared, which months and year, how the instruments were translated to Croatian.

Check the journal guidelines – but I find the Descriptive tables should be better put in the Results section. In the Descriptive tables, you can write the SD on items in brackets and italic after the M (e.g., M(Sd) 2.32(1.37))

Replace ‘Reverse items’ with ‘Reversely scored items’ throughout the manuscript.

Results

Add the age range, apart from the M and Sd, as in minimum and maximum.

In Table 6, the significance of the standardized Beta regression coefficients should be written. Now only the statistical significance of B is written.

I would be much more interested to see the differences in the used instruments between the two educational groups then the skewness and kurtosis. So please add another table which will show the M and Sd of the used instruments separately for the SH and STEM subsamples. Even conducting an ANOVA would be appropriate.

Discussion

Start with a sentence that reminds the reader of the aim of the study.

At the end, I also think it is important to state the overall limitations of an online survey, your sample and methodology in general.

Highlights

I suggest rewriting the highlights as more comprehendible or fuller sentences, it is hard for me as a reader to follow it in this form.

Overall

Language and grammar require some editing (especially in the highlights section), so I suggest carefully reading through the manuscript and double checking or engaging a native speaker. I would also emphasize that the variables used here are not the only ones that are shown to be associated with vaccine conspiracy beliefs (another example from Croatia is DOI: 10.1080/08870446.2019.1673894)

Reviewer #2: This article is exploring the differences in vaccination conspiracy beliefs between people belonging to different scientific fields – Social Sciences & Humanities (SH) and Science, Technology, Engineering and Mathematics (STEM). The results revealed that those educated in SH were more prone to vaccination conspiracy beliefs, and this was mediated by health beliefs; lower trust in the healthcare system, and lower science literacy.

Although the authors deal with important questions for the society today, the rationale for this specific research question is not explored enough. The theoretical background need to be more explained, followed by the importance of differentiation between SH and STEM persons. Is there any specific recommendation coming from these findings (in terms of science education)?

Some issues remained unanswered: What kind of professions exactly, do the authors have in the groups under the umbrella of STEM and SH? What are the employment statuses of the respondents? Sometimes people work in completely different fields then those from which they graduated. It would be good to discuss the relation between employment-education status of the respondents and relation of those variables to the conspiracy belief.

Structure of the manuscript in some parts is not adequate for the academic paper. There is no difference between Introduction and Method. Research goals and questions should be placed in the Introduction part. Method part is missing; it should stand alone not incorporated in other parts. Instruments should be placed in the Method part. Sample information is not sufficient: we need to know when the survey was conducted, whether there was ethical approval from any institution, the length of the questionnaire. Also I would recommend reporting the sample characteristics in more details.

Additionally, descriptive statistics are usually part of results not method part. Endorsement of scientific knowledge and vaccine conspiracy belief should be mentioned as research question (because they are presented in the tables), and the total scores could be given separately for SH and STEM part

Tables are not according to APA standards.

I would also like to see empirical confirmation for the assumptions mentioned in the discussion: ...Additionally, a large portion of these differences can be explained by the mediating factors that are part of the 3C's model which, according to our claim, might be stemming from the differences in epistemic cultures, i.e. teaching patterns, cognitive approaches, research traditions, goals, behaviors, logic and methods. How can authors assume that? In the discussion part it would be better if the authors analyze results that they measured in the study.

Finally, what is the relation between Conspiracy beliefs vs. Vaccine intentions? What are the effects of conspiracy belief in the current pandemic? And how can the results of the study be connected to the situation in pandemic – covid-19 conspiracy theories. The connection between vaccine CT and Covid-19 CT should be explained.

Reviewer #3: The goal of the present study was to determine whether were differences in vaccine hesitancy and likelihood of accepting vaccine conspiracy theories in individuals educated in social sciences and humanities versus those educated in STEM. The research question is interesting and timely given concerns related to COVID-19 vaccine hesitancy. However, the rationale is underdeveloped, the method/results lack important details, and the discussion over-reaches. Below, I outline my concerns in more detail.

The introduction is underdeveloped. The authors mention the 3Cs at the beginning of the introduction and review some research on the “confidence” component, but no attention is given to convenience or complacency. They then state that they will examine the 3Cs in relation to those educated in STEM and social sciences/humanities. It is unclear how or why the 3Cs are being examined. More discussion and connection to the research questions would be helpful. The authors also outline differences between STEM and social sciences/humanities fields, but don’t clearly connect this to vaccine hesitancy. It would be helpful for the authors to more explicitly state their reasoning and provide hypotheses, if they had a priori hypotheses. Clearer rationale for the mediation analyses needs to be provided.

More information needs to be provided regarding participant recruitment. How were participants contacted? What was the response rate? Is the sample representative of the population in question? What range of degrees are represented in the sample (e.g., BS/BA, MS/MA, PhD, etc.)? Did participants receive compensation? Was there informed consent?

Were there any missing data, outliers, non-normality, etc.? If so, how were these issues addressed?

Why were age, gender, and religiosity included as covariates? Did results differ if covariates were excluded?

Before reporting the mediation analysis, it would be useful to report simple comparisons (t-tests) indicating whether STEM vs. SH participants differed in any of the mediators or outcome variables.

A clearer description of the mediation analysis(es) is necessary, noting how variables were coded (i.e., STEM vs. SH) and included in the model. I assume a parallel mediation model was estimated.

Remove reference to marginal significance (p = .06). If relying on p-values, an effect is significant or not. Also, discussion of full or partial mediation is inappropriate, given the correlational and cross-sectional nature of the study.

It is unclear to me what purpose the hierarchical regression analysis serves.

The claim in the discussion that the 3Cs mediated difference in education field and vaccine conspiracy attitudes is misleading as two of the three indirect effects were not significant. The discussion should be based on empirical findings from the study, not speculation of mechanisms unsupported by the data. For example, is there empirical evidence to suggest that SH individuals are more likely to endorse CAM than STEM individuals? If not, the authors need to temper their claims in the discussion.

Another limitation is the correlational, cross-sectional nature of the study. You cannot truly test mediation. Be sure to avoid causal language. Alternative explanations for the findings should also be considered. There are a number of unmeasured factors that may differ between SH and STEM individuals.

Minor Comments

I suggest reformatting the paper to include section headers (e.g., Method).

For all measures, indicate what direction of scores to help readers interpret the data. For example, do higher scores on the vaccine conspiracy attitudes scale indicate more or less vaccine hesitancy?

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: No

Reviewer #3: No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

Submitted filename: RECENZIJA Plos.docx

Click here for additional data file.

14 Dec 2021

Response to Editor/Reviewers

Dear Editor,

Dear Reviewers,

Thank you for giving us the opportunity to revise and resubmit this manuscript. We appreciate your suggestions and we have incorporated them into the revised manuscript. We hope that paper is significantly improved in this way. All our specific responses are highlighted in blue.

Reviewer #1: Thank you for the opportunity to review the manuscript entitled ‘Vaccination conspiracy beliefs: differences between persons educated in different scientific fields and their mediating mechanisms'.

The author(s) present a study in which they investigated vaccine conspiracy beliefs in regard to educational background – social science and humanities versus stem. It is an interesting take on the role of education in vaccine conspiracy beliefs, because education is in this kind of research mostly investigated in terms of educational levels and not field differences.

I found the manuscript to be well written, and it contributes with new insights. Nevertheless, I recommend some changes which can contribute to an even greater quality of it.

Title

I suggest renaming the title to emphasize your main finding and make it more interesting for potential readers, some examples are: ‘Vaccine conspiracy beliefs stronger in social sciences and humanities then in STEM – a correlational study’.

Reply: We changed a title to make it clearer, and to emhasize our findings. Thank you for the suggestion.

Abstract

Start with an introduction sentence.

Reply: Added.

Introduction

Page 4, lines 76-79, add references for describing the SH and STEM fields.

Reply: Done.

Throughout the manuscript replace ‘persons’, perhaps with ‘people’, although I am not sure this is the adequate term too.

Reply: Where appropriate (in case where we are talking about the persons who participated in the survey), we changed „persons“ into „participants“. In other cases, we changed „persons“ into „people“.

Research goals, questions, and methods

Please insert a paragraph that in details describes the procedure of how and when the study was conducted - where was the survey shared, which months and year, how the instruments were translated to Croatian.

Reply: we added the requested details, and also explained that the data were taken from the wider pilot study of vaccine hesitancy in Croatia.

Check the journal guidelines – but I find the Descriptive tables should be better put in the Results section.

Reply: We understand your point, but we kept it in the Methods (now: Methods and instruments) section since they describe the measures employed in the study, i.e. they are not connected to our research questions as such.

In the Descriptive tables, you can write the SD on items in brackets and italic after the M (e.g., M(Sd) 2.32(1.37))

Replace ‘Reverse items’ with ‘Reversely scored items’ throughout the manuscript.

Reply: Done.

Results

Add the age range, apart from the M and Sd, as in minimum and maximum.

Reply: Done.

In Table 6, the significance of the standardized Beta regression coefficients should be written. Now only the statistical significance of B is written.

Reply: In the revised version we did not conduct hierarchical regression analysis, since mediation analysis is done by PROCESS in a more precise way. Standardized coefficients are not part of the regression table any more.

I would be much more interested to see the differences in the used instruments between the two educational groups then the skewness and kurtosis. So please add another table which will show the M and Sd of the used instruments separately for the SH and STEM subsamples. Even conducting an ANOVA would be appropriate.

Reply: we see your point, and that is why we added the separate descriptive statistics for SH and STEM for each instrument. Additionally, we also added intercorrelational matrix, as well as effect sizes (Cohen's d) for nominal variables. All this, we hope, should provide a clearer insight into the variables relatioinships.

Discussion

Start with a sentence that reminds the reader of the aim of the study.

Reply: Done, we start with the brief summary of the results and their relation to our research questions.

At the end, I also think it is important to state the overall limitations of an online survey, your sample and methodology in general.

Reply: Done.

Highlights

I suggest rewriting the highlights as more comprehendible or fuller sentences, it is hard for me as a reader to follow it in this form.

Reply: Done

Overall

Language and grammar require some editing (especially in the highlights section), so I suggest carefully reading through the manuscript and double checking or engaging a native speaker. I would also emphasize that the variables used here are not the only ones that are shown to be associated with vaccine conspiracy beliefs (another example from Croatia is DOI: 10.1080/08870446.2019.1673894)

Reply: Done

Reviewer #2: This article is exploring the differences in vaccination conspiracy beliefs between people belonging to different scientific fields – Social Sciences & Humanities (SH) and Science, Technology, Engineering and Mathematics (STEM). The results revealed that those educated in SH were more prone to vaccination conspiracy beliefs, and this was mediated by health beliefs; lower trust in the healthcare system, and lower science literacy.

Although the authors deal with important questions for the society today, the rationale for this specific research question is not explored enough. The theoretical background need to be more explained, followed by the importance of differentiation between SH and STEM persons. Is there any specific recommendation coming from these findings (in terms of science education)?

Reply: We added an explanation of the importance of the differentiation between SH and STEM (86-90).

Some issues remained unanswered: What kind of professions exactly, do the authors have in the groups under the umbrella of STEM and SH?

Reply: Done. We added the following sentence - As for the delineation of SH and STEM fields, the official classification of Ministry of Science and Education of the Republic of Croatia was used.

What are the employment statuses of the respondents? Sometimes people work in completely different fields then those from which they graduated. It would be good to discuss the relation between employment-education status of the respondents and relation of those variables to the conspiracy belief.

Reply: we completely agree with this, but unfortunately we did not measure employment status, i.e. we do not have this indicator available in the dataset.

Structure of the manuscript in some parts is not adequate for the academic paper. There is no difference between Introduction and Method. Research goals and questions should be placed in the Introduction part. Method part is missing; it should stand alone not incorporated in other parts. Instruments should be placed in the Method part.

Reply: We now clearly separated research goals and hypotheses from methods, and moved the methods into a separate section (now: Methods and instruments).

Sample information is not sufficient: we need to know when the survey was conducted, whether there was ethical approval from any institution, the length of the questionnaire. Also I would recommend reporting the sample characteristics in more details.

Reply: All the details are added now ()243-254).

Additionally, descriptive statistics are usually part of results not method part.

Reply: We agree, but in our case descriptive statistics that is presented is not connected to the research questions, but has a role of presenting the characteristics of the measurement instruments.

Endorsement of scientific knowledge and vaccine conspiracy belief should be mentioned as research question (because they are presented in the tables), and the total scores could be given separately for SH and STEM part

Reply: Done, we presented all descriptive statistics separately for SH and STEM, and also added intercorrelational matrix which sheds more light onto the relationships between the variables.

Tables are not according to APA standards.

I would also like to see empirical confirmation for the assumptions mentioned in the discussion: ...Additionally, a large portion of these differences can be explained by the mediating factors that are part of the 3C's model which, according to our claim, might be stemming from the differences in epistemic cultures, i.e. teaching patterns, cognitive approaches, research traditions, goals, behaviors, logic and methods. How can authors assume that? In the discussion part it would be better if the authors analyze results that they measured in the study.

Reply: we also agree with this, that is why in the revised version we emphasized that only science literacy was a statistically significant mediator. But we also pointed out to the fact that confidence intervals in the bootstrapping analyses of the remaining two mediators include zero effect by only a small margin, and that the coefficients go into the direction that is consistent with the assumed epsitemic differences between the science fields. In other words, we recommended that these mediators are worth researching in studies with other (random) samples.

Finally, what is the relation between Conspiracy beliefs vs. Vaccine intentions? What are the effects of conspiracy belief in the current pandemic? And how can the results of the study be connected to the situation in pandemic – covid-19 conspiracy theories. The connection between vaccine CT and Covid-19 CT should be explained.

Reply: We added the commentary of the aplicabiltiy of our results to the management of COVID-19 pandemic. In brief, we pointed out to the fact that vaccination conspiracy beliefs are strongly connected to health beliefs (importance of natural immunity) and to the trust in healthcare system (this can be seen from the intercorrelation matrix and from the results of regression analysis), i.e. that that vaccine hesitancy cannot be reduced solely by providing factual information.

Reviewer #3: The goal of the present study was to determine whether were differences in vaccine hesitancy and likelihood of accepting vaccine conspiracy theories in individuals educated in social sciences and humanities versus those educated in STEM. The research question is interesting and timely given concerns related to COVID-19 vaccine hesitancy. However, the rationale is underdeveloped, the method/results lack important details, and the discussion over-reaches. Below, I outline my concerns in more detail.

The introduction is underdeveloped. The authors mention the 3Cs at the beginning of the introduction and review some research on the “confidence” component, but no attention is given to convenience or complacency. They then state that they will examine the 3Cs in relation to those educated in STEM and social sciences/humanities. It is unclear how or why the 3Cs are being examined.

Reply: we agree, but 3Cs model was described only to point out that our mediators are not chosen at random, but that they are a part of a wider picture which can be systematized by 3Cs model. We added explanation which elements of 3Cs model we are using - institutional trust (confidence), postmodern health beliefs (complacency), and science literacy (convenience - “ability to understand”).

More discussion and connection to the research questions would be helpful. The authors also outline differences between STEM and social sciences/humanities fields, but don’t clearly connect this to vaccine hesitancy. It would be helpful for the authors to more explicitly state their reasoning and provide hypotheses, if they had a priori hypotheses. Clearer rationale for the mediation analyses needs to be provided.

Reply: we hope that we adressed these concerns in the revised version of the manuscript. We did transform or research questions into hypotheses, because we indeed posed hypotheses based on our reflection of epistemic differences between SH and STEM.

More information needs to be provided regarding participant recruitment. How were participants contacted? What was the response rate? Is the sample representative of the population in question? What range of degrees are represented in the sample (e.g., BS/BA, MS/MA, PhD, etc.)? Did participants receive compensation? Was there informed consent?

Were there any missing data, outliers, non-normality, etc.? If so, how were these issues addressed?

Reply: We added more information about sample and recruitment (243-254). There were not significant outliers, while non-normality should be successfully handled by PROCESS according to Hayes book (Introduction to Mediation, Moderation, and Conditional Process Analysis, 2018, New York: Guilford Press). As for missing values, about 7,53% of all cases did have missing values on at least one of the study variables. We conduced binary logistic regression with the missing value as the criterion variable and did not find any significant relationship with other study variables.That is why we proceeded with the complete case analysis, which reduced our sample from 624 do 577 respondents. The description is now added into the revised version of the manuscript.

Why were age, gender, and religiosity included as covariates? Did results differ if covariates were excluded?

Reply: These covariates were chosen because of the possible connection to vaccine hesitancy based on the previous studies. The results did differ with regards to gender, and this was an excellent point that we now added to the discussion of the results.

Before reporting the mediation analysis, it would be useful to report simple comparisons (t-tests) indicating whether STEM vs. SH participants differed in any of the mediators or outcome variables.

Reply: This is a very good point, and it was also suggested by other reviewers. We did this by adding the intercorrelation matrix and effect sizes (Cohen's d) of the nominal variables. Hopefully, this shed more light onto the relations between the study variables before the mediation analysis.

A clearer description of the mediation analysis(es) is necessary, noting how variables were coded (i.e., STEM vs. SH) and included in the model. I assume a parallel mediation model was estimated.

Reply: Yes, it as a multiple parallel mediation. We added a conceptual diagram that should make the entire procedure much clearer than in the initial version of the manuscript.

Remove reference to marginal significance (p = .06). If relying on p-values, an effect is significant or not. Also, discussion of full or partial mediation is inappropriate, given the correlational and cross-sectional nature of the study.

Reply: Yes, we avoided this reference. But having in mind that confidence intervals include zero effect by only a small margin, form the point of view of Bayesian statistics as well for theoretical reasons (all mediation coefficients point to the hypothesized direction), we just emphasized that the suggested mediating mechanisms are worth of researching in future studies with random samples.

It is unclear to me what purpose the hierarchical regression analysis serves.

Reply: PROCESS is a regression-based method and hierarchical regression is also used to study mediation. However, PROCESS is a strict method that gives confidence intervals for direct and indirect effects, while hierarchical regression is much more „impressionistic“ method, i.e. it does not include explict test of the effects. That why we followed your suggestion and removed the hierarchical regression. As noted PROCESS is regression-based, so we provided the overall regression that is a part of the PROCESS output. This is useful because it shows the direct effects of all predictors. The other part of the story (overall and indirect effects of the field of study) is presented in the mediation analysis. Hopefully, we made it all clearer in the revised manuscript.

The claim in the discussion that the 3Cs mediated difference in education field and vaccine conspiracy attitudes is misleading as two of the three indirect effects were not significant. The discussion should be based on empirical findings from the study, not speculation of mechanisms unsupported by the data. For example, is there empirical evidence to suggest that SH individuals are more likely to endorse CAM than STEM individuals? If not, the authors need to temper their claims in the discussion.

Reply: We agree, and we made it excplicit now what the study results showed (only one of the mediation paths is statistically significant), and what are our hypotheses.

Another limitation is the correlational, cross-sectional nature of the study. You cannot truly test mediation. Be sure to avoid causal language. Alternative explanations for the findings should also be considered. There are a number of unmeasured factors that may differ between SH and STEM individuals.

Reply: we fully agree, that is why we included demographic control variables. And we are also aware that the study is correlational, we added it as an additional limitation at the end of the paper.

Minor Comments

I suggest reformatting the paper to include section headers (e.g., Method).

For all measures, indicate what direction of scores to help readers interpret the data. For example, do higher scores on the vaccine conspiracy attitudes scale indicate more or less vaccine hesitancy?

Reply: Added.

Submitted filename: Response to reviewers.docx

Click here for additional data file.

14 Jan 2022

26 Jan 2022

Dear Editor,

Thank you for giving us the opportunity to revise and resubmit new version of this manuscript. We appreciate the time and suggestions provided by reviewer 3 and by you and we have incorporated the suggested changes into the manuscript.

We have responded specifically to each suggestion below, beginning with your own.

- throughout: change “persons” to “people”

Response: Done

- throughout: be consistent in your use of tense. Often past and present tenses are mixed, even in the same paragraph (e.g. paragraph starting line 176, paragraph starting line 281), which is very confusing for the reader.

Response: Past tense is now consistently used throughout the paper.

- 45: “become a commonplace” is hardly used in English, change to “has become commonplace”

- 56: “Various research” change to “Research” and add some references.

Response: Done.

- 62 and 486: “and healthcare system” change to “and the healthcare system”

Response: Done.

- 76: “vaccine hesitance” change to “vaccine hesitancy”

Response: Done.

- 81: add comma after “[28]”

Response: Done.

- 101: I find the term “socialization” somewhat problematic as it seems to suggest that it is an exclusively external influence (basically, all “nurture”) without allowing the fact that certain personality traits make people more or less likely to pursue a “soft” or “hard” science. Ideally, one would want to carry out a study amongst final year secondary school students who are planning to go into either direction.

Response: We agree. The term „socialization“ is changed to „background“, since the latter term covers both the academic socialization and the preexisting differences among the students. Also we added sentence „Such differences can be the result of the different academic socializations and/or the individual characteristics which are important when choosing field of study.” Also in chapter Conclusion and limitations „The reasons could probably be found in the different academic socialization and the background characteristics which are important when selecting a field of study.“

- 112: “there only few” change to “there are only few”

Response: Done.

- 113: “differences” – unclear what the directions of the differences are.

Response: The direction is now explicitly stated („mainly confirming that vaccine hesitancy is higher among SH graduates“).

- 125-129: please rewrite the sentence as I am unsure what is being stated.

Response: Done.

- 139: “on the sample” change to “on a sample”

Response: Done.

- 151-155: another difference between SH and STEM is a difference in focus, with STEM more focused on groups and averages and SH on individuals.

Response: We agree, but that topic was not covered in the research that we refer to.

- 165: willing to accepted” change to “willing to accept”

Response: Done.

- 204: you have to start with the participant info otherwise the results in Tables 1 - 4 are uninterpretable.

Response: The participants' info is moved to the beginning of the chapter.

- as suggested by one of the Reviewers in the previous round, please add simple t-tests when splitting the data between groups, e.g. lines 209-210).

Response: Done.

- 210, 220, 233: “higher results” change to “higher values”

Response: Done.

- Table headings for Tables 1 to 4: “for THE… scale”

Response: Done.

- 237: please add a reference for the Oxford scale

Response: Done.

- 249: please add references for the surveys referred to.

Response: Done.

- 266: “consisted of the college”: delete “the”

Response: Done.

- 269: “conduced” change to “conducted”

Response: Done.

- 277: add a reference or link for the classification used

Response: Done.

- 303: delete “Andrew F.”

Response: Done.

- Tables 6 & 7: define LLCI and ULCI

Response: Explanations added.

- paragraph starting 308: why are the numbers (e.g. 1.36) different from those in the Table?

Response: It is corrected now.

- 315: “is statistically” change to “is a statistically”

Response: Done.

- 337: “significant”: where is this indicated in the Table?

Response: It is visible since the confidence interval includes zero (effect).

- 352: “with the other” change to “with other”

Response: Done.

- 372: “additional analyses”: where can the reader find these?

Response: We included them in the paper now (i.e. the magnitude of the indirect effects when gender is not a covariate). All analyses can be easily replicated by using our data and SPSS PROCESS syntax (process y=Vac_con /x=Stu_re /m=Lit Imm HCS /cov= Age Relig Gender /model=4.) Since PROCESS uses bootstraping results can be negligibly different from analysis to analysis, but the conclusions are always the same (we repeated all analyses several times).

- 382: delete “so”

Response: Done.

- 385-386: “sample of…”: please rephrase

Response: It is rephrased as: „as well as with a research conducted on a sample of Croatian population”.

- 428: “found…correlation”: change to either “found a ... correlation” or “found… correlations”

Response: Done.

- 430: “indicate” change to “indicates”

Response: Done.

- 432: “having in mind that” change to “in that”

Response: Done.

- 432: “the vaccine” change to “vaccine”

Response: Done.

- 455: “to much” change to “too much”

Response: Done.

- 456: “object the fact” change to “object to the fact”

Response: Done.

- 471: “as well with” change to “as well as with”

Response: Done.

- 494: delete “in it”

Response: Done.

Just as a suggestion, I think some of the research by Gordon Pennycook and colleagues will be of interest to your future research, e.g. their examination of a bias to accept “pseudo-profound bullshit”, which is linked with e.g. religious beliefs and analytical thinking (e.g. Pennycook & Rand, 2019; DOI: 10.1111/jopy.12476).

Response: Than your for the suggestion, we are going to use it in the continuation of our project.

Reviewer #3:

Reviewer #3: Overall, the authors have done a nice job addressing my initial concerns and suggestions. The manuscript is much stronger. I do have a few minor suggestions.

1. In the introduction, it would be helpful to explicitly connect field of study with the 3Cs to clearly articulate rationale.

Response: the following sentence is added: „. Additionally, it is our contention that such differences will be mediated by some of the indicators coming from the 3Cs model (trust in the healthcare system, specific health beliefs, and science literacy).”

2. There are still a few places in the manuscript where the authors refer to "causal" relationships. That cannot be determined from this study.

Response: We checked it, we use this word only in the following sentence: „In addition, given that our study was cross-sectional and the possible omitted variables bias might have existed, causal mechanisms cannot be fully supported by the data from our study.“. Following suggestions of the reviewer, in other parts of the paper we only refer to „correlations“, „connections“, etc. Please let us know if we are wrong, i.e. if we missed some of the causal implying language.

3. Provide an explicit rationale for the inclusion of covariates in the analyses. Also, as the some of the results were different when covariates were excluded, report the stats in text or supporting material.

Response: The following sentence is added: „. The covariates were used in order to check for the possible spurious effects of our predictors.”.

The stats (indirect effects) of the model when gender is not added as a covariate are now reported in the text.

4. It is important to acknowledge that differences between education and training in these fields may not be the reason for these findings. Rather, there may be unmeasured individual differences (e.g., personality, cognitive styles, etc.) that lead people into different educational fields and influence vaccination conspiracy beliefs.

Response: Yes, we agree. That is why we replaced the term „academic socialization“ with „academic background“, since the latter term implies both possible explanations. The following sentence is also added: „Such differences can be the result of the different academic socializations and/or the individual characteristics which are important when choosing a field of study.”(in Introduction) and „The reasons could probably be found in the different academic socialization and the background characteristics which are important when selecting a field of study.“ (in chapter Conclusion and limitations).

Submitted filename: Response to Reviewers and Editor.docx

Click here for additional data file.

16 Feb 2022

Vaccination conspiracy beliefs among social science & humanities and STEM educated people - an analysis of the mediation paths

PONE-D-21-19906R2

Dear Dr. Šuljok,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Steven Frisson

Academic Editor

PLOS ONE

Additional Editor Comments:

Please make the following minor corrections:

- line 50: comma after "MacDonald[30]"

- line 131: comma after [5] and before "Biglan's"

- line 208: I don't understand the use of "whereas" here as it is not contrastive. Change to "and"

- line 218: change "classification of Ministry" to "classification by the Ministry"

- lines 223-224: change "agreed that project" to "agreed that the project"

- line 404: change "that bootstrapping" to "that the bootstrapping"

- line 406: change "in healthcare system" to "in the healthcare system"

- line 413: change "aware of distinctiveness" to "aware of a distinctiveness"

Reviewers' comments:

N/A

21 Feb 2022

PONE-D-21-19906R2

Vaccination conspiracy beliefs among social science & humanities and STEM educated people - an analysis of the mediation paths

Dear Dr. Šuljok:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

If we can help with anything else, please email us at plosone@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. Steven Frisson

Academic Editor

PLOS ONE