Trust in science and scientists among secondary school students in two out-of-school learning activities

ABSTRACT

Research on science outreach activities is often located in the interface between science communication and science education. The transferability of aims and objectives of one research field to the other offers great potential. The widely recognized aim of ‘trust in science’ in science communication is still less discussed in science education. However, when teaching emotive scientific topics such as climate change, vaccines or genetic engineering, students’ trust in science is of great importance. This paper presents a study of two interventions (N_PartI = 443; N_PartII = 333), to (1) assess the level of trust in science among secondary school students, and (2) to investigate the impact of outreach activities on the development of trust in science. Results showed that the mean level of trust in science among secondary school students is similar to the level among university students. We found a trust-enhancing effect of the interventions exclusively for students with a low prior level of trust (low-trustors). Furthermore, results indicated that high levels of trust in science can support learning in science outreach activities. These findings are particularly important when considering that increasing students’ level of trust in science appears to be especially important for low-trustors in order to prevent negative social tendencies.

KEYWORDS:

• Informal education
• public awareness
• science outreach

Introduction

The Corona virus – a hoax or hyped-up scaremongering? 5G radiation responsible for Corona epidemic and microchips in vaccinations? Misinformation and conspiracy theories have been part of the public discourse not only since the global Corona pandemic. Misinformation partly even originates from dissident scientists or pretend to be based on scientific principles to gain convincing legitimacy (e.g. Díez Arroyo, 2013). But how can learners and laypeople distinguish between misinformation and scientific evidence? In a science-based society, with its highly evolved division of cognitive effort, the individual remains a layperson in most areas of knowledge (Bromme & Thomm, 2016). It is important for each individual to be able to judge what is based on scientific evidence, and what is not (Hendriks et al., 2016b). Therefore, one main goal of science instruction should not be to achieve blind trust in scientific information or scientists per se. Rather, the aim should be to increase trust in the process of scientific knowledge acquisition. To trust the process of science, one must first understand what defines authentic science (Lombrozo et al., 2008; Miller, 2004). Therefore, science education should give insights into the fundamental principles of scientific knowledge, practice and reasoning (Gräber et al., 2002). Students should understand the nature of science, scientific terms and concepts, and the meaning of science’s interaction with society – summarized under the term scientific literacy. The OECD PISA Framework defined scientific literacy as ‘the ability to engage with science-related issues, and with the ideas of science, as a reflective citizen’ (OECD, 2017, p. 22). Scientific literacy often demands a critical and reflective approach to knowledge and information. However, the goal is not to generally question everything but to enable students to engage in reasoned discourse about science and technology based on fundamental knowledge about and trust in science. Bromme (2020) defined this critical trust as ‘informed trust’ and emphasized that not only basic knowledge, but above all knowledge about science itself as relevant factors. For active participation in society, neither blind trust nor a fundamental questioning of every scientific finding is useful. Students, therefore, should learn to distinguish when to question emerging and uncertain scientific contexts and, in turn, when and why to trust in science (Bryce & Fraser, 2014; Fensham, 2014). ‘Doubting everything or believing everything are two equally accommodating solutions, either of which saves us from reflection’ (Henri Poincarés, La Science et l’Hypothèse, quoted by Allchin (2014)).

Definition of trust (in science)

Trust is described as a kind of assumption about others. When people (the trustors) are dependent on persons or institutions (the trustees), and when they are willing to accept the risks associated with this dependance, they trust these persons or institutions (Blöbaum, 2016; Bromme & Thomm, 2016; PytlikZillig & Kimbrough, 2016). This dependency describes the willingness to be vulnerable to another person or institution (Mayer et al., 1995). When this theory is applied to science and scientists, it means that the good that the trustee (the scientist) provides to the trustor (the layperson) is ‘knowledge’, and the risk to the trustor is his vulnerability to a lack of truth or validity of that knowledge (Hendriks et al., 2016b). People depend on the knowledge of experts when it comes to developing a personal opinion on science-based topics and making decisions about them (Hendriks et al., 2016b). In terms of science, trust can be described as a perception of scientists as credible, likely to tell the truth and share the public’s interest (National Academies of Sciences, Engineering, and Medicine, 2015).

Trust is a complex construct that includes affective and cognitive dimensions (Dunn & Schweitzer, 2005; Mayer et al., 1995). Thus, trust is strongly influenced by emotional perceptions and can itself have a major impact on people’s perception, especially in emotionally charged scientific topics (Dunn & Schweitzer, 2005; Romano, 2003). Examples of such topics in recent societal discourse are genetic engineering (Broughton & Nadelson, 2012), climate change (Dunlap & McCright, 2011), vaccines (Keelan et al., 2010), or restrictions due to the Covid-19 pandemic (Plohl & Musil, 2021). Higher trust in science and scientists is likely to increase acceptance in these areas (Nadelson & Hardy, 2015; Sturgis et al., 2021). According to this definition, trust in science cannot be separated from trust in the scientists themselves and is therefore considered together in this paper as trust in science and scientists.

Factors influencing trust in science

Various authors have already stated that publics’ trust in science is useful and necessary for the functioning of a science-based society (Arimoto & Sato, 2012; Hendriks et al., 2016b; Sztompka, 2007). To gain insight into publics’ trust in science, some studies have already assessed the level of public trust in science in general (e.g. Besley, 2014; Wissenschaft im Dialog, 2019). For example, the Wellcome Global Monitor was conducted in 144 countries, making it the largest study to date examining attitudes toward science. The calculated trust index shows medium trust for more than half of EU citizens (61%), high trust for 25% and low trust in science for 11% (Gallup, 2019). Moreover, some scientists have tried to identify factors that influence this complex construct (e.g. Fiske & Dupree, 2014; Hendriks et al., 2016b). Findings suggest, for example, that people with privileged socio-demographic identities trust science more than those with marginalized identities (American Academy of Arts and Sciences, 2018; Funk et al., 2015; Gauchat, 2012). In addition, Nadelson et al. (2014) found a positive correlation of trust in science with the number of years studying science. The vast majority of the public recognizes science as an important tool for understanding the world and acknowledges that science is necessary to make effective decisions in their lives (Fischhoff, 2014) – they trust in science and scientists (Besley, 2014; Wissenschaft im Dialog, 2019). Nevertheless, there are some people who hold beliefs that contradict the best available scientific knowledge (Funk et al., 2015; Leiserowitz et al., 2019). Existing research identified religious beliefs (Nadelson et al., 2014), conservative political tendencies (Gauchat, 2012; Myers et al., 2017), ideological motivations (Lewandowsky et al., 2013) and poor scientific literacy (Lombrozo et al., 2008; Miller, 2004) as possible causes for the lack of trust some have in science. Furthermore, there is evidence that a low level of trust in science is a good predictor of people believing in conspiracy theories and paranormal beliefs (Fasce & Picó, 2019; Irwin et al., 2016; Requarth, 2017) as well as for science denialism (Lewandowsky et al., 2013; Nadelson & Hardy, 2015; Omer et al., 2009). These unwarranted beliefs (Fasce & Picó, 2019) have risen in prevalence with dangerous risks for society (Chigwedere et al., 2008; Gangarosa et al., 1998; Johnson et al., 2018). Enhancing public trust in science is therefore very important to ensure the functioning of a society based on science and technology.

Measurement of trust in science

It is important to note that, first, a distinction between trust in scientists as persons and trust in science per se is not possible due to the high emotional dependence of the construct trust and, second, that the data collected always relate only to the selected research field and the transferability of the results must be carefully examined. Nevertheless, different instruments exist that already measure the multifaceted construct ‘trust in science’ in different fields of research (Anderson et al., 2012; Brossard & Lewenstein, 2010; Farias et al., 2013; Hall et al., 2002; Hartman et al., 2017). One of the most frequently used instruments is the Trust in Science and Scientists Inventory (TSSI) by Nadelson et al. (2014): A 21-item trust scale to measure generic (not domain-specific) trust in science and scientists. The TSSI was used so far to determine the influence of political attitudes and religious beliefs on trust in science (Blankenship & Stewart, 2019; Slater et al., 2019) or to identify factors that influence the development of conspiracy theories and science denialism (e.g. Nadelson & Hardy, 2015). Currently, there is much research on the relationship between trust in science and the acceptance of Covid-19 protective measures (e.g. Plohl & Musil, 2021; Szczuka et al., 2020). The instrument’s use was located in the fields of social psychology and science communication and the target group so far has been the general public (recruited from various contexts, age of the participants 18–65). The level of trust in science among school students and the relationship between trust in science and learning success or demographic data in science education have so far received little attention. In the school context, the construct trust has primarily been measured as interpersonal social relations (Forsyth et al., 2011; Liou & Daly, 2014; Murphy-Graham & Lample, 2014; Ream et al., 2014; Watson et al., 2019) or in the form of organizational trust in schools (Černe et al., 2014; Smetana et al., 2016). However, it has rarely been investigated as trust in science.

Design and methods

Based on the theoretical framework presented above, this research focused on investigating trust in the science of secondary school students. For this purpose, we developed two science-based outreach activities in which the students investigated ecological changes in the Baltic Sea and their social impacts using authentical and up-to-date scientific methods. Thus, this paper presents a study of two parts to investigate the influences of these outreach activities on the development of trust in science of secondary school students. The first activity took place in school, the second in an out-of-school student laboratory to ensure the independence of results from the learning location. For measuring trust in science and scientists of secondary school students, we first adapted the existing instrument TSSI. Using a simple experimental research design, we investigated (1) students’ level of trust in science and possible correlations with demographic variables. Furthermore, we examined (2) the impact of the outreach activities on the development of trust and (3) the correlation between learning success and trust in science. Therefore, we assessed in the pre-test demographic data as well as individual interest in biology and chemistry and the level of trust in science and scientists. The post-test analyzed the effectiveness of the outreach activities regarding their learning success and the level of trust in science and scientists.

Our studies were carried out as part of the KiSOC-project, which aims to investigate success factors for understandable and motivating science communication. A key aspect of the KiSOC is the close collaboration between Leibniz Institutes and universities, which gives this project a special interdisciplinary character. Marine ecology researchers, science educators and psychologists worked together in a co-design to develop and evaluate the outreach activities that were carried out in the Kieler Forschungswerkstatt (student laboratory of the Kiel University and Leibniz Institute for Science and Mathematics Education).

Description of the outreach activities

The two outreach activities underlying this research were similar in structure, overall topic, objectives, target group and evaluation design but differed in duration, learning location and complexity of the topic (Table 1). Because the investigations were identical in terms of design and procedure, they show a high degree of comparability. In each case, scientists of marine ecology, media psychology and education developed the interventions together in a co-design. Climatic and anthropogenic changes in marine ecosystems served as subject matter and hands-on experiments, as well as an interactive computer simulation, were used as learning methods. The outreach activities were designed for secondary level school students of Grade 10–13 (which is comparable to high school grade such as Grade 9–12 in the US). The 20-minute tests were administered at the beginning and end of each activity.

Table 1. Overview of the two interventions.

	Outreach activity I	Outreach activity II
Location	School	Student laboratory
Thematic issue	Ocean acidification and impacts on calcifying organisms	Four anthropogenic changes of the Baltic Sea and impacts on ecosystem and societal level
Duration	1.5 hours	5 hours
Participants	N = 443	N = 333
	Grade 10–13	Grade 10–13
	Mean age: 17.58	Mean age: 17.02
Learning methods	Experiments or simulation	1^st Experiments + 2^nd Simulation or
		1^st Simulation + 2^nd Experiments

The first intervention was a 90-minute intervention at school and focused on the comparison of the two methods. The purpose of this intervention was to convey the process and the effects of ocean acidification on a global and local level. An introductory lecture provided an overview of the topic of ocean acidification and highlighted the resulting problems for marine ecosystems. Students were randomly assigned and worked on the same content either with experiments or with a simulation. Both methods conveyed the same content: pH-value, reasons for and extent of the increase in global atmospheric carbon dioxide (CO₂) concentration, equilibrium reaction of CO₂ in seawater and the effect of acidification on calcifying organisms. In both treatments, students used the same research questions to investigate the process and effects of ocean acidification. The activities were supported by a script, in which the results of the simulation or experimental work had to be summarized by leading questions. The methods were conducted in small groups with a supervisor each who was available for questions and organized the structural procedure (for example, he explained the materials, he provided assistance in answering the questions in the script, he paid attention to the avoidance of typical mistakes in performing the experiment or showed the students the most important functions of the simulation). The supervisors were the same for the whole intervention.

The second intervention was a full-day student lab day on the topic ‘Future of the Baltic Sea’ and focused on the combination of two learning methods. The purpose of this intervention was to convey the processes of major environmental impacts of the Baltic Sea (warming, eutrophication, acidification and salinity changes) and its effects on representative organisms of the ecosystem (gammarids, bladderwrack, and epiphytes) as well as the resulting challenges for the whole system (water quality, fishing, tourism). An introductory lecture provided information on the causes and processes of global changes in the oceans and gave an insight into the resulting effects on marine ecosystems. Subsequently, students were randomly assigned and worked on the same content with a combination of either first experiments and second simulation or the other way around. Both methods conveyed the same content. The experiments each conveyed a single process of change (warming, acidification, eutrophication or salinity changes) with the effects on one organism (gammarids, bladderwrack or epiphytes) and the resulting ecosystem and societal changes (water quality, fishing or tourism). Thus, the students learned about the different changes and their effects as well as their interactions one after the other. In the simulation, it was possible to simulate all changes simultaneously and to observe the effects on all three organisms together, taking into account their interactions. Furthermore, the adjusted parameters represented directly possible effects on the entire system (water quality, fishing, and tourism). To avoid disorientation and cognitive overload, especially by working on the simulation, students had to work on a supporting script in which they had to summarize the results of their work by the same guiding questions. A supervisor was available for each group for technical as well as content-related questions and he supported the group discussions. The supervisors were all undergraduate university science students and were briefed on the consistent process prior to conducting the interventions.

Evaluation methods

Demographic data

Since there is already much research on the correlations of trust to classic variables of demographic data (socio-economic status, education level, etc.), we assume that these findings can be applied to students as well. We, therefore, examined here only demographic data relevant to schools like age, gender, marks in science subjects (biology, chemistry and physics), chosen subject focus and individual interest in biology and chemistry. Nevertheless, we continue to call them demographic data.

To gain a general impression of students’ achievement in the science subjects, we calculated a mean value from the marks of the three science subjects. At German schools, it is typical that secondary school students focus on a thematic area (e.g. biology, chemistry, art, languages, sports, economics, etc.). We divided the subject focus into two groups: science-related (biology, chemistry, physics, general science) or non-science-related (art, languages, sports, economics, etc.) subject focus. For measuring individual interest, we adapted the PISA instrument ‘pleasure and interest in science’ (Frey et al., 2009), each for interest in biology and chemistry. The students had to assess five statements each using a 4-point rating scale (1 = completely disagree; 4 = completely agree). For the calculations, the two scales were combined since we were interested in general interest in this area, independent of the specific subject.

Learning success

We developed and validated a new questionnaire because no established test existed for these extra-curricular topics of marine ecology. The tests were developed on the basis of existing knowledge tests in chemistry (Höft et al., 2019), the questions each represented the different content areas of the two interventions. Most of the items were multiple-choice, but open-ended questions also were included to get a better insight into the students’ learning success of multilevel knowledge. In Outreach-Activity I, students could score up to 15 points; in Outreach-Activity II, students could reach 27 points. We calculated mean values for the analysis.

Trust in science and scientists

We measured trust in science and scientists using the Trust in Science and Scientists Inventory (TSSI) by Nadelson et al. (2014). The original scale included 21 items and has proven to be highly reliable and valid for use with undergraduate science students. Through an iterative process of design, the authors ensured sufficient validity of the items through the support of science faculty members and expert feedback. Nadelson et al. (2014) conducted one field test with 75 undergraduate college students enrolled in an introductory geoscience course an another field test with 301 undergraduates who were enrolled in a range of science courses, which resulted in an internal consistency of Cronbach’s alpha = .86. We first had to adapt and test the instrument in terms of usability for secondary school students before being used in our studies. The 21 items developed by Nadelson et al. (2014) have previously been used without content structuring, and there has not yet been statistical evidence of interrelation between the items. Previous studies already used only selected items of the instrument but neither provided a statistical nor a content explanation for their selection (Blankenship & Stewart, 2019; Kingsley et al., 2017; MacDonald et al., 2020). Our intention was to represent the 21-item variable set of the TSSI by a few independent factors with statistical and content-related justification. On the basis of 368 data sets of the first survey of Nadelson et al. (2014) we analyzed the structure of the instrument using explorative factor analysis. Both the Bartlett-test (Chi-square (210) = 2633.17, p < .001) and the Kaiser-Meyer-Olkin Measure of Sampling Adequacy (KMO = .902) indicated that the variables were suitable for factor analysis. After the main component analysis with the varimax rotation, we can present a two-factor solution. The items for both factors demonstrated sufficient internal consistency to warrant merging them into a scale each (Factor 1: α = .70; Factor 2: α = .82). We decided to use Factor 2 in order to investigate the general trust in scientists and science and to enable more practical use of the instrument on students. For practical reasons, we call this scale trust in science or just trust in the following (Table 2).

Table 2. Items loading on Factor 2 with corresponding factor load.

Variable		Load
10	We should trust that scientists are being honest in their work.	.903
11	We should trust that scientists are being ethical in their work.	.840
9	We should trust the work of scientists.	.751
7	I trust that the work of scientists to make life better for people.	.395
12	Scientific theories are trustworthy.	.344

The translation of the items into the German language was done independently by three experts. In addition, three teachers validated the instrument to ensure that it is suitable for students. They evaluated the translated items in terms of wording and complexity in order to adapt them to the knowledge and experience of secondary school students. We piloted the adapted instrument with a group of N = 44 students from three different classes of Grade 13 (mean age = 18.76 (SD = 0.91)). The students carried out the interventions as planned and were asked to discuss incomprehensible aspects of the questionnaires in an open plenary session. This collaborative validation enabled us to eliminate ambiguities in the formulation and technical terminology. We could find sufficiently high internal consistency (Cronbach’s α pre-test = .744; Cronbach’s α post-test = .865) for the new 5-item scale.

Participants

Results

Only data sets with no missing data were considered for calculations. Therefore, N differs in some cases. The following calculations show mean value comparisons and correlations (Table 3).

Table 3. Overview of the participants of the two interventions. All data are based on the students’ own statements in the pre-test. Marks are reported on a scale of 1 = very good to 6 = inadequate. Interest was assessed on a four-level Likert scale (1 = low interest to 4 = high interest).

	Outreach activity I	Outreach activity II
Number of participants	443	367
Average age in years	17.58 (SD = 1.41)	17.02 (SD = 1.20)
Gender	57 % female and 43 % male	54 % female and 46 % male
Chosen courses (science-related or non-science-related)	45.1% science courses; 54.9% non-science courses	54.8% science courses; 45.2% non-science courses
Average marks science subjects	2.55 (SD = 0.84)	3.0 (SD = 1.07);
Mean interest in biology	2.91 (SD = 0.72)	2.95 (SD = 0.65)
Mean interest in chemistry	2.31 (SD = 0.79)	2.47 (SD = 0.68)

Students’ level of trust in science

A sufficiently high internal consistency could be determined for both studies (Cronbach’s α Study I = .742; Cronbach’s α Study II = .740). The results of the pre-test showed that students had a medium to high trust in science (M_StudyI = 3.51 (SD = 0.60); M_StudyII = 3.48 (SD = 0.63); on a five-point Likert scale). The independent variables of the school-related demographic information did not affect the level of trust in science significantly. Thus, no difference in the level of trust could be found in terms of gender or chosen subject focus. Furthermore, there was no statistically significant correlation between trust and the age of the students nor with their grades. We could only find a weak correlation for trust in science and marks in science subjects as well as for trust in science and individual interest in biology and chemistry (Appendix 1–3).

Effects of the outreach activities on trust in science

The mean value comparison of the pre- and post-test results for trust in science and scientists showed no statistically significant differences for both studies (Table 4). Additional analysis of Repeated Measures ANOVAs showed that neither was there a change in trust depending on gender, subject focus, age, grade, nor in science marks for both studies.

Table 4. Mean value comparison of the pre- and post-test results for trust in science.

Intervention	Pre-test	Post-test	df	T	p
	Mean (SD)	Mean (SD)
I	3.52 (0.60)	3.51 (0.64)	416	.190	.849
II	3.48 (0.63)	3.51 (0.66)	332	−1.036	.301

We further dichotomized high and low levels of trust in science and scientists. These levels were operationalized as one standard deviation above (high-trustors) or below (low-trustors) the mean of the pre-test scores (+1SD/−1SD). The statistical division of the groups resulted for Intervention I in 63 high-trustors and 55 low-trustors and for Intervention II in 61 high-trustors and 39 low-trustors (Table 5). An analysis of variance with repeated measurements showed that the change of the trust level is related to the pre-level of trust (F_{Study I}(1,110) = 21.543, p < .001, partial η² = .164, f = 0.44; F_{Study II}(1,97) = 39.894, p < .0001, partial η² = .291, f = 0.64). Thus, low-trustors were able to gain significantly more trust in science through outreach activities compared to high-trustors (Figures 1 and 2). According to Cohen (1988) the effect size f each corresponds to a strong effect.

Figure 1. Change of mean trust in science through the outreach activity according to different pre-levels of trust in science for Intervention I (High-trustors N = 63; Low-trustors N = 55).

Figure 2. Change of mean trust in science through the outreach activity according to different pre-levels of trust in science for Intervention II (High-trustors N = 60; Low-trustors N  = 39).

Table 5. Descriptive information of the mean trust level of high-/low-trustors in pre-/post-test.

		Pre-test	Post-test
Intervention I	High-trustors (N = 63)	4.37 (0.21)	4.14 (0.44)
	Low-trustors (N = 55)	2.43 (0.44)	2.61 (0.65)
Intervention II	High-trustors (N = 60)	4.32 (0.22)	4.04 (0.63)
	Low-trustors (N = 39)	2.27 (0.42)	2.87 (0.7)

Correlation between learning success and trust in science

The level of trust in science had an influence on the level of learning success of the outreach activity (F(1, 438) = 12.647, p < .001, f = 0.16). However, only 2.6% of the variation in learning success was explained by the level of trust in science. The effect strength f = 0.16 represents, according to (Cohen, 1988), only a low effect. In both interventions, students with a high level of trust in science learned significantly more through the outreach activities than students with a low level of trust in science (Table 6). The effect sizes according to (Cohen, 1988) showed a medium effect for Intervention I and a strong effect for Intervention II.

Table 6. Learning success depending on the pre-test trust level: Mean value comparison of knowledge success between high-trustors (+1SD) and low-trustors (−1SD).

Intervention	High trust	Low trust	df	t	p	d
	Mean scores (SD)	Mean scores (SD)
I	9.3 (2.44)	8.0 (3.77)	116	2.253	.026	0.42
II	13.89 (4.50)	10.49 (3.52)	98	3.998	< .001	0.82

Discussion

Secondary school students had a rather high level of trust in science

The mean value of trust in the science of secondary school students is very similar to the mean value of university students (Koehler & Pennycook, 2019; Nadelson et al., 2014; Nadelson & Hardy, 2015). Even though the latter is a group that has chosen to study science and could therefore have been expected to have a higher level of trust in science. This suggests that trust in science might be developed to a certain degree during school education, at least in this cohort of secondary schools. The non-existent correlations of age or grades with a trust may indicate that trust may have been developed prior to entering secondary school. However, this would need to be verified, especially with the inclusion of other socio-economic variables and perhaps also with other parts of the instrument. It would further be interesting for research to look longitudinally to examine changes during the school career and cross-sectionally through all levels of the school system to see if there is a development of trust during the school career (comparable to Kuhn et al. (2000) and Hofer and Pintrich (1997) who describe such development of epistemic sophistication). It would also be interesting to investigate which factors have a promoting effect on the development of trust during the school years.

The correlation between individual interest, understanding of science and trust might indicate that not only the understanding of science practice is important to promote trust in science but also the interest in it. This seems to be especially relevant for educators and should be considered for the design of outreach activities. However, these considerations are very limited because the correlation between interest and trust is very weak. Further research to better understand this correlation is therefore required.

Trust-enhancing effect of outreach activities only for low-trustors verifiable

The fact that we did not find general changes in the trust level due to the outreach activities is consistent with recent research (Kingsley et al., 2017; Ocobock & Hawley, 2020). However, we could identify a positive influence of the outreach activities on low-trustors. Working with authentic, up-to-date methods and content may have been beneficial here. Science-based outreach activities, therefore, might be useful with the intention of reducing negative societal tendencies, which are supported by a low level of trust (science denialism, conspiracy theories, etc.). It is questionable whether such a high level of trust in science, as was observed among the high-trustors in both interventions, requires further enhancement. According to Bromme (2020), it is not so much absolute trust as purely informed trust that should be promoted. The handling of authentic scientific methods, the contact with scientists and the working with real data may lead to a more differentiated view of science. Perhaps the lower trust level reflects a rejection of universal statements. This finding highlights the importance of further, more differentiated measurement instruments for the construct trust. The existing instruments do not allow to distinguish between blind and informed trust. Of course, these observations only apply to our interventions, which are designed very closely to science. Further science-based outreach activities (e.g. in interdisciplinary student laboratories) should be evaluated to test whether the results are reproducible.

Trust in science positively influences learning success

Although the learning success could only be predicted to a limited extent by general trust in science, the dichotomous subdivision of the construct showed that a high level of trust seemed to facilitate learning of science-related topics in outreach activities. This assumption is also supported by the fact that, at least to a small extent, a high level of trust correlates with good marks in science subjects. We could not find similar research results in this area yet, but evidence from other fields of research allows comparable conclusions. For example, research in the field of business and economics shows that interpersonal and organizational trust has a positive impact on organizational learning (Attiq et al., 2017; Jiang & Chen, 2017). Research on education and development indicates a positive influence of interpersonal trust between co-learners on the learning process (Anwar & Greer, 2012; Brücknerová & Novotný, 2017; Klijn et al., 2010). In addition, Durkin and Shafto (2016) identified epistemic trust as a mutable factor that influences learning in an academic setting. This means that the trustworthiness of informants (teachers, professors, etc.) positive influences the learning success (Koenig & Harris, 2005; Lee & Kim, 2016; Pasquini et al., 2007; Schlesinger et al., 2016). Thus, trust in science should already be encouraged in school to support the learning of scientific principles and processes. A high level of trust might encourage learning and motivation for science activities rather than questioning the trustworthiness of science.

Limitations

A theoretical limitation is that the presented results only apply to our interventions. Further research on students’ trust in science is thus required to test the replicability of the results. Furthermore, the selection of participants may have influenced the level of trust in science: The participants of our studies had a rather high educational level, as all of them aspired to the highest level of school graduation in Germany (Abitur). Supervisors may also have had an impact on the results. However, since all supervisors were students of natural sciences and were informed about a uniform procedure of the interventions, this effect can be assumed to be small. In addition, the students had quite good marks in science and a high level of interest in biology and chemistry, which may have influenced the high level of trust also. However, the studies conducted with the TSSI so far have also been conducted with highly educated participants (Kingsley et al., 2017; Plohl & Musil, 2021; Slater et al., 2019; Szczuka et al., 2020), so the limitation in terms of comparability of results is considered to be low. Students with a high level of trust in science had better marks in the science subjects. Therefore, the finding that a higher level of trust in science has a positive effect on the learning success of the outreach activity may be reasonably questioned. It may also be that the students with better marks generally had a learning advantage and therefore learned more through the intervention. Further research is required to examine this aspect in detail. The comparison of this study with the values of the university students of Nadelson et al. (2014) is only valid to a limited extent, since the items used varied (this study used only five among 21 items) as well as the demographic information and learning activities were different.

Implications

This paper presents an example of the great potential of transferring classical research interests in science communication to the research field of science education (Baram-Tsabari & Osborne, 2015). We conclude that students in our cohort of secondary schools had a fairly high level of trust in science. Accordingly, trust may have already been developed during the previous school years. For further research, it is therefore of interest to investigate which factors have a beneficial effect on the development of trust. In addition, we call for an investigation of the replicability of the results with other outreach activities. Even if the results do not point to a lack of trust in science, the further promotion of trust in science through authentic learning environments such as outreach activities in student laboratories is of great importance. To promote trust in science and scientists, we suggest that such activities be developed in a science-based approach and use authentic methods that should be explicitly presented to students. A co-design model with scientists, educators, and media experts is an effective way to achieve this. Increasing informed trust, especially among students with low levels of trust in science, can thus counteract negative social trends and avoid unwarranted beliefs.

Ethical statement

The authors confirm that all research meets ethical guidelines and adheres to the legal requirements of the study country (Germany). The Ministry of Education, Science and Culture of the State of Schleswig-Holstein (Germany) approved the research. In addition, all participants provided appropriate informed consent for the research.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Appendix

Appendix 1. Statistical values of t-tests for differences between the trust in science and gender of the participants (male = M and female = F) for both interventions (I + II).

Table

Variable	Mean M (SD)	Mean F (SD)	df	t	p
Gender Intervention I	3.53 (0.54)	3.49 (0.67)	436	0.771	.441
Gender Intervention II	3.44 (0.66)	3.53 (0.55)	355	−1.475	.141

Appendix

Appendix 2. Statistical values of t-tests for differences between the trust in science and the choice of subject focus (S = science-related or N-S = non-science-related) for both interventions (I + II).

Table

Variable	Mean S (SD)	Mean N-S (SD)	df	t	p
Subject focus Intervention I	3.50 (0.63)	3.52 (0.57)	432	−0.255	.799
Subject focus Intervention II	3.49 (0.69)	3.44 (0.56)	364	0.705	.481

Appendix

Appendix 3. Statistical values of correlations between trust in science and different socio-demographic variables for both studies (I + II), significant correlations are highlighted in bold.

Table

Variable	r	p
Age Study I	−.064	.182
Age Study II	−.089	.089
Grade Study I	−.068	.154
Grade Study II	−.084	.110
Marks Study I	.101	.039
Marks Study II	.116	.027
Interest Study I	.218	< .001
Interest Study II	.231	< .001