Introduction

Out-of-school labs (OSLs) are non-formal learning settings that try to provide students with authentic insights into scientific practices in order to foster their interest in and knowledge about scientific ways of thinking and working (e.g., Scharfenberg & Bogner, 2014). Consequently, OSL research has focused on investigating the effectiveness of OSLs for students’ motivational and cognitive learning outcomes, such as their interest (e.g., Priemer & Pawek, 2014) or achievement (e.g., Itzek-Greulich et al., 2015) in science. However, by engaging students in authentic learning activities that try to emulate the work of scientists (e.g., Glowinski & Bayrhuber, 2011), it is likely that OSLs may also have an impact on students’ beliefs about the nature of knowledge and knowing. According to Kitchener (2002) as well as Greene et al. (2016), we refer to these beliefs about knowledge as epistemic beliefs.

Interventions on students’ change of epistemic beliefs usually aim to counter a view of knowledge as being simple, certain, and provided by authority, and, instead, to enable “individuals to perceive greater complexity and tentativeness of knowledge and a more active role for themselves as knowers” (Muis et al., 2016a, p. 333). Epistemic beliefs recognizing the complexity, tentativeness, and personal construction of knowledge have traditionally been characterized as sophisticated, whereas the perception of knowledge as being simple, certain, and provided by authority has been labeled as a naïve view towards knowledge (e.g., Hofer & Pintrich, 1997). In the present paper, we adopt this traditional distinction between naïve and sophisticated epistemic beliefs.

One could assume that authentic learning activities in OLSs, which try to simulate the ways how scientists gather new knowledge, may help students to experience the complexity, variability, and tentativeness of scientific knowledge and knowing. Authentically contextualized learning activities are assumed to demonstrate to students the real-world meaning, relevance, and functionality of the target learning content (e.g., Lepper, 1988) in order to prevent the acquisition of inert knowledge that cannot actually be applied in real-world contexts (e.g., Herrington & Oliver, 2000). OSLs usually aim for “disciplinary authenticity” (Shaffer & Resnick, 1999). According to this kind of authentic learning, students should learn by doing, for instance, solving problems or tackling questions, in the same way as professionals in a particular discipline do. This is also how Lombardi and Oblinger (2007) define an authentic learning environment, namely as one that resembles the social structure and culture of a discipline by simulating the ways of working and thinking within a discipline. Thus, it is likely that the intended disciplinary authenticity of OSLs should provide students with authentic experiences of the inherent features of scientific practices and scientific knowledge. Against this background, some studies have begun to examine the effects of OSLs—as authentically contextualized learning settings—on students’ epistemic beliefs in different scientific disciplines, namely historical science (Mierwald et al., 2018), biomedical science (Kapitza, 2020), and the natural sciences more generally (Schiefer et al., 2017, 2020). This focus on natural sciences is consistent with intervention studies in general, which aim to foster a change of students’ epistemic beliefs. As the narrative literature review conducted by Muis et al. (2016a) exemplifies, out of 22 included studies that investigated the change of learners’ epistemic beliefs, only four studies referred to the social sciences and humanities, and, thus, to domains other than STEM.

As the findings of a recent study conducted by Urhahne and Kremer (2023) emphasize, epistemic beliefs are highly domain-specific and differ between the natural, mathematical, social, and linguistic sciences. The results show that pre-service teachers (i.e., university students) hold comparable epistemic beliefs about the natural and social sciences that are more sophisticated than their (again comparable) beliefs about the mathematical and linguistic sciences. These findings partly contrast with numerous previous studies showing that school and university students’ epistemic beliefs about the social sciences and humanities, especially psychology, are more sophisticated than those about the natural sciences and mathematics (e.g., Estes et al., 2003; Hofer, 2000; Muis et al., 2016b; Rosman et al., 2020). Taken together, these results suggest that students’ epistemic beliefs about the linguistic sciences are less sophisticated than those about the natural sciences, while students’ epistemic beliefs about the social sciences, especially psychology, are more or comparably sophisticated than those about the natural sciences. However, findings by Palmer and Marra (2004) suggest that—depending on the domain-specific expertise level—students’ shift to highly sophisticated epistemic beliefs may be more difficult in the social sciences and humanities than in the natural sciences. These findings point to the need to find effective and domain-specific ways to support students’ development of sophisticated epistemic beliefs in the social sciences and humanities. One promising way may relate to authentic learning activities in OSLs, as they may help students to get a better understanding of how scientists gather new knowledge. Against this background, the present two studies tested the impact of authentic learning activities in an OSL on students’ epistemic beliefs about the social sciences and humanities, namely the educational sciences (study A) and linguistics (study B).

It should be mentioned that it has been questioned whether a fixed and domain-general characterization of certain beliefs about knowledge as being either naïve or sophisticated is sensible and actually holds true in different domains and learning contexts. For instance, the findings of a study conducted by Bråten et al. (2008) suggest that supposedly sophisticated beliefs are not necessarily related to other learning outcomes (e.g., text comprehension) and that epistemic beliefs which have traditionally been labeled as naïve can be advantageous in certain learning contexts. Against this background, the present two studies further analyze how students’ epistemic beliefs are associated with other outcome variables that have been focused on in previous OSL research, namely students’ perceived authenticity (studies A and B), situational interest (studies A and B), and knowledge acquisition (study A only).

In summary, the present studies pursue two research objectives: firstly, to examine the effect of authentic learning activities in OSLs on students’ epistemic beliefs about the educational sciences and linguistics and secondly, to investigate the relation between students’ epistemic beliefs and other outcome variables, namely their perceived authenticity, situational interest, and knowledge acquisition. In the following sections, we first describe why authentic learning activities in OSLs may be a promising way for fostering sophisticated epistemic beliefs in students (as background of our first research objective) and then describe findings of previous research on the relation between learners’ epistemic beliefs and further motivational and cognitive outcome variables (as background of our second research objective).

Please note that the dataset of study A has been used for investigating two further research questions (see Nachtigall et al., 2020; Nachtigall & Rummel, 2021). This prior work does not include any analyses of students’ epistemic beliefs. The unique contribution of the present paper lies in the findings from two different studies on (1) the effects of authentic learning activities implemented in an OSL for social sciences and humanities on students’ development of epistemic beliefs and (2) the association between learners’ epistemic beliefs and further learning-related constructs.

Theoretical background

Impact of authentic learning on epistemic beliefs

Intervention studies that, so far, have aimed to affect learners’ development of sophisticated epistemic beliefs can—according to Kienhues et al. (2016)—be distinguished between studies that either directly or indirectly confront learners with the complexity, variability, and tentativeness of knowledge. The direct interventions usually use certain texts (e.g., refutation texts or multiple conflicting texts) that confront learners with divergent information, controversial knowledge, and conflicting perspectives in order to affect their epistemic beliefs (e.g., Kienhues et al., 2008). The indirect interventions try to engage learners in constructivist learning activities to demonstrate the processes of knowledge building and, thereby, the personal construction, tentativeness, and complexity of knowledge (e.g., Muis & Duffy, 2013). Authentic learning activities in OSLs offer opportunities for both approaches.

Specifically, in a history-related OSL project, Mierwald et al. (2018) implemented a direct intervention and provided secondary school students with multiple documents containing conflicting information from authentic primary and secondary sources. They compared studying these authentic learning materials with the use of didactically reduced and non-authentic textbook-like materials without conflicting information. Their results demonstrated that the authentic materials had an impact on students’ development of more sophisticated history-related epistemic beliefs, while students who learned with the textbook did not change their beliefs during the OSL visit. In contrast, the extracurricular science intervention developed by Schiefer et al., (2017, 2020) can be described as mostly indirect. It focused on engaging elementary school students, who participated in an extracurricular enrichment program at a children academy (which is comparable to an OSL), in authentic inquiry-based and hands-on activities. The activities tried to emulate processes of scientific inquiry and to situate the students into the role of scientists, and thereby aimed to foster their epistemic beliefs about the natural sciences. Schiefer et al., (2017, 2020) combined this indirect intervention with more direct elements, as they also used conflicting evidence and implemented a guided group discussion to help students to reflect and think about knowledge and knowing. Their findings showed a positive effect of the intervention on students’ development of epistemic beliefs when it was conducted by university researchers and compared to a differently treated control group (see Schiefer et al., 2020), but no effect when it was conducted by trained course instructors from the children academy and compared to a waitlist control group (see Schiefer et al., 2017). Kapitza (2020) also combined an indirect intervention with a more direct approach in order to foster secondary school students’ epistemic beliefs related to biomedical science in an OSL. The indirect intervention, implemented in two conditions, included authentic learning activities in an OSL which, for instance, asked students to situate themselves into the role of physicians and to treat two fictitious patients in an online simulation. After this indirect intervention, students in one of the two conditions additionally received a more direct intervention at school. The direct approach provided them with concept cartoons showing one person that expresses a rather sophisticated view related to biomedical knowledge and a second person that represents rather naïve beliefs. Students were asked to discuss the epistemic beliefs presented in the concept cartoons and to reflect on their own beliefs in small groups. The results of Kapitza’s (2020) study revealed no differences between the two conditions regarding students’ change of epistemic beliefs.

Given these inconsistent findings of a small number of studies, the potential of authentic learning activities implemented in OSLs for changing students’ epistemic beliefs is unclear. Thus, there is a need to further investigate whether authentic learning activities in OSLs that try to mimic scientific ways of thinking and working and, thereby, demonstrate processes of knowledge building in certain scientific disciplines have an impact on students’ development of sophisticated epistemic beliefs. The present two studies address this research gap and contribute to previous research by investigating the effects of authentic learning activities implemented in an OSL on students’ change of epistemic beliefs related to social sciences and humanities.

Interplay between epistemic beliefs, motivation, and learning

The goal of promoting the development of sophisticated epistemic beliefs underlies the assumption that beliefs in the complexity, tentativeness, and personal construction of knowledge are particularly beneficial for learning. Previous research indeed has demonstrated that students’ epistemic beliefs are related to both their learning from texts (for a review, see Bråten et al., 2011) and various motivational constructs (Chen & Barger, 2016; Guo et al., 2022). In a nutshell, ample empirical evidence suggests that epistemic beliefs traditionally labeled as sophisticated are usually related to increased cross-text comprehension and integration (Bråten et al., 2011) as well as to increased self-efficacy, mastery-goal orientation, interest in and enjoyment while learning, and perceived usefulness of learning in certain domains (Chen & Barger, 2016; Guo et al., 2022). But there seemingly exists an exception with regard to students’ epistemic beliefs about the source of knowledge: Findings from Bråten et al. (2008) show that “naïve” epistemic beliefs about the source of knowledge, namely the belief that knowledge is transmitted from authorities and experts, are related to better text comprehension than “sophisticated” epistemic beliefs.

Looking at the four studies that have investigated effects of OSLs on students’ epistemic beliefs (i.e., Kapitza, 2020; Mierwald et al., 2018; Schiefer et al., 2017, 2020), two of these studies also examined relations to other variables. The study conducted by Kapitza (2020) demonstrated a relation between students’ sophisticated epistemic beliefs and their increased interest in biomedical science but no association with students’ knowledge acquisition. Specifically, students who demonstrated a profile of highly sophisticated epistemic beliefs prior to the intervention differed significantly from students with a less sophisticated profile in their reported interest in biomedicine after the intervention, but not in their performance on a knowledge posttest. Schiefer et al. (2020) investigated the correlations between students’ epistemic beliefs and two motivational constructs, namely epistemic curiosity and investigative interest. Their findings show positive correlations between students’ epistemic beliefs about the development and justification of knowledge and the two motivational constructs at the beginning of the intervention, but only a positive correlation between students’ epistemic beliefs about the justification of knowledge and their epistemic curiosity at the end of the intervention. These positive correlations suggest that the more sophisticated certain epistemic beliefs of students are, the higher their epistemic curiosity (and investigative interest).

However, research on the effectiveness of OSLs has particularly been interested in examining the potential of OSLs for fostering students’ knowledge (e.g., Scharfenberg et al., 2007) about and situational interest (e.g., Neher-Asylbekov & Wagner, 2023; Priemer & Pawek, 2014) in scientific concepts as well as scientific ways of thinking and working. Situational interest is defined as a psychological state of engaging with certain content, and as an affective reaction that is first triggered and then maintained by stimuli of the learning environment (Hidi & Renninger, 2006). It is assumed that OSLs foster students’ situational interest and knowledge through the provision of authentic insights into scientific endeavors (Euler, 2004; Scharfenberg & Bogner, 2014). Consequently, research on OSLs has additionally focused on investigating students’ perceived authenticity of the learning setting (e.g., Glowinski & Bayrhuber, 2011).

Given the findings of previous research on students’ epistemic beliefs more generally and on the effectiveness of OSL interventions for the development of epistemic beliefs more specifically, it is likely to assume that students’ epistemic beliefs are linked to motivational constructs, such as their situational interest. With respect to associations between epistemic beliefs and cognitive outcomes, such as knowledge acquisition, the findings of previous research are less clear and consistent. Research on epistemic beliefs more generally has focused on learning from texts and it is unclear whether these findings can be transferred to learning in OSLs. Research on the development of epistemic beliefs in OSLs more specifically has not yet focused on investigating the relations between students’ epistemic beliefs and their knowledge acquisition. The study from Kapitza (2020) is the only exception and suggests no such relation. How students’ epistemic beliefs are linked to their perceived authenticity of the learning setting has not yet been investigated. However, given that our first research objective builds on the assumption that authentic learning activities promote students’ development of sophisticated epistemic beliefs, it is likely to assume that the authenticity of the learning activity perceived by students relates to their epistemic beliefs. Similarly, the model of authenticity in learning and teaching contexts (see Fig. 1) developed by Betz et al. (2016) includes the assumption that the effects of an authentically designed learning setting on various motivational, cognitive, or behavioral learning outcomes are always mediated through learners’ perceived authenticity. Therefore, for the two studies presented here, it can be assumed that students’ perceived authenticity of the learning activity relates to their epistemic beliefs.

Fig. 1
figure 1

Simplified model of authenticity in learning and teaching contexts adapted from Betz et al. (2016)

Full size image

The present studies

The present two studies aim to investigate whether students’ epistemic beliefs about educational sciences and linguistics can be affected by authentically contextualized learning activities in an OSL (research question 1), and how students’ epistemic beliefs relate to their perceived authenticity of the learning activity, their situational interest in the contents of the OSL projects, and their knowledge acquisition (research question 2). For this purpose, we present two (quasi-)experimental studies that investigated the effect of authentic learning activities on students’ epistemic beliefs in an OSL for educational sciences (study A) and linguistics (study B).

Both studies varied the authenticity level of the learning activity (i.e., the method that students used to work on a given task) and implemented two conditions in an OSL: an authentic learning activity emulating processes of scientific inquiry (A+ condition) and a less or even non-authentic learning activity (A− condition).

Specifically, study A compared the following two learning approaches: problem-solving prior to instruction (A+ condition) and instruction followed by problem-solving (A− condition). The learning approach implemented in the A+ condition is often called productive failure (e.g., Kapur & Bielaczyc, 2012). Productive failure (PF) asks students to collaboratively generate different solutions to a complex problem before receiving instruction on canonical solutions. It is described as an authentic learning activity (Cho et al., 2015; Nachtigall & Rummel, 2021), as it reflects processes of scientific inquiry, such as independently exploring solutions to a novel and complex problem, and situates learners in the role of a scientist (see Kapur & Toh, 2015). The approach implemented in the A− condition is often referred to as direct instruction (e.g., Kapur, 2012) and can be described as less authentic (see Nachtigall & Rummel, 2021), as learners only have to apply the procedure they were instructed on to correctly solve a problem. In contrast to PF, the direct instruction (DI) approach does not require students to explore different solutions on their own and, due to the initial instruction, the problem-solving task lacks novelty and complexity. Thus, in DI approaches, scientific inquiry appears as a pure application of clear instructions and, thus, as a “simple, algorithm procedure” (Hodson, 1999, p. 784). Consequently, it is likely that the DI approach promotes beliefs about knowledge being highly structured, stable, and objective, while the PF approach emphasizes the personal construction and complexity of knowledge.

Study B compared the following two learning activities: analyzing the usage of language by applying the descriptive (lat. describere‚ “describe”) linguistic method (A+ condition) and the prescriptive (lat. praescribere, “prescribe”) method (A− condition). According to contemporary views of grammatical language analysis, it is the task of linguists to unbiasedly describe the different forms of language use and, thus, to work descriptively (Pittner, 2016; Wunderli, 2014). The language should be described from a documentary-analytical point of view without pursuing the goal of language guidance. Hence, all language usages should be treated equally. Nowadays, linguists work with the descriptive method (Kilian et al., 2016; Pittner, 2016; Wunderli, 2014). Consequently, asking A+ students to use the descriptive method can be classified as an authentic learning activity that emulates an actual and current scientific practice within the field of linguistics. Instead, asking A− students to apply the prescriptive method can be classified as a less authentic activity, as the prescriptive view of language is considered obsolete and contradictory to the modern orientation of linguistics (Klein, 2004). The prescriptive approach prescribes a certain use of language and explicitly determines the correctness and incorrectness of certain usages of language. Applying this method requires to evaluate linguistic forms based on specific literary or dialectal models. One goal of this method is to create a prestigious high-level language through language-directing measures (Bußmann, 2008). Consequently, it is likely that the prescriptive method promotes conceptions of knowledge being highly objective, absolute, and inflexible, while the descriptive method recognizes that knowledge (about the correct usage of language) is negotiated and tentative.

Method

Participants

One hundred fifty-two secondary school students (age: M = 16.11 SD = 0.90; 65% girls) from seven educational science and/or social studies courses (10th grade) of six different schools participated with written parental consent in study A. Specifically, five educational science courses, one social studies course, and one mixed course with 11 students from a social studies course and 10 students from an educational science course participated in the study. The seven courses were randomly assigned to the A+ (i.e., three classes; n = 80) and A− condition (i.e., four classes; n = 72) as a whole.

In study B, 150 secondary school students (age: M = 16.96 SD = 0.90; 56% girls) from eight German language courses (10th–12th grade) of five schools participated with written parental consent. In every course, students were randomly assigned to the experimental groups (A+ : n = 75; A− : n = 75).

Research context and study design

Both studies took place in an interdisciplinary OSL for social sciences and humanities at a large German university. The OSL is located on the university campus and consists of a small reference library and three spacious and modern classrooms with moveable desks, a projector, and a large screen, and, if needed (which was the case in both studies), laptops for all students. Thus, the OSL can be characterized as an authentic learning environment because of its research-related location (i.e., university campus) and equipment (e.g., library and laptops). In addition, both OSL projects were led by real/authentic researchers, namely junior researchers in the field of educational psychology (study A) or German linguistics (study B). While study A was conducted in an OSL project for educational sciences, study B was conducted in a project for linguistics. Specifically, the OSL project for educational sciences aimed at communicating certain features of social science research methods to students and focused on the topic of evaluating causal versus correlative evidence. The OSL project for linguistics aimed at communicating certain linguistic methods (i.e., descriptive and prescriptive method) that can be used to analyze grammatical constructions. The project focused on analyzing double perfect constructions as their grammatical status has not yet been fully clarified.

Both investigations compared two experimental conditions: an authentic learning activity (A+) and a less authentic learning activity (A−). As OSLs are usually visited by whole classes, the classes were randomly assigned to the two conditions in study A. Thus, study A implemented a quasi-experimental design. In contrast to study A, the interventions in study B allowed for random assignment of individual students to the two conditions per class. Thus, study B implemented an experimental design by assigning one half of each class to the A+ condition and the other half to the A− condition.

Procedure and learning materials

Study A: educational science

In study A, students first received a brief introduction to different social science research methods (e.g., survey, observation, and experiment) and on the typical research process within the social sciences (i.e., generating a research question, planning a study, collecting and analyzing data). Afterwards, they experienced two successive learning phases: a problem-solving and an instruction phase (see Fig. 2). To simulate certain features of scientific inquiry (e.g., developing and falsifying conjectures and experiencing failure and uncertainty), A+ students started with the problem-solving phase that asked them to independently work on a complex and novel problem in small groups before they received instruction on the canonical solution. Students from the A− condition started with the instruction phase before they had to apply the instructed method for collaboratively solving the same problem as their counterparts in the A+ condition.

Fig. 2
figure 2

Procedure of study A

Full size image

During the problem-solving phase in both conditions, students were provided with information (i.e., design and hypothesis) on a published correlation study which investigated the relation between exposure to violent media and aggressive behavior in children and adolescents. The task asked students to think about possible results of the study and to discuss whether or not the study design allowed for testing a causal hypothesis. During the instruction phase in both conditions, the same instructor (1) briefly repeated the steps of a typical research process within the social sciences and emphasized that mistakes and failure can be a constructive part in each of these, (2) recapitulated the problem-solving task in the A+ condition or introduced the same problem in the A− condition, (3) described, based on typical student solutions, different possible results (i.e., positive, negative, and no association) of a correlation study and their respective interpretations, (4) explained why correlational evidence cannot be interpreted as causal, (5) emphasized that misinterpreting correlative evidence as causal is a common mistake made by students, journalists, and even researchers, (6) presented the method of controlled experiments as an opportunity to test causal claims, and finally (7) summarized the main messages. After the second learning phase (i.e., instruction in the A+ condition and problem-solving in the A− condition), students in both conditions were asked to individually solve a problem that was isomorphic to the task of the collaborative problem-solving phase. A questionnaire administered after the introduction (i.e., pre-questionnaire) and after the practice phase (i.e., post-questionnaire) assessed students’ epistemic beliefs about the educational sciences. The post-questionnaire also measured students’ perceived authenticity and situational interest. At the end of the project, we administered a knowledge test. Figure 2 shows the procedure of study A.

Study B: linguistics

In study B, students first received a brief introduction to the German tense system (Reichenbach, 1947). Specifically, they received an overview of the different existing tenses and information on how these tenses are built and what functions they have. Afterwards, students started with the analysis of double perfect constructions. Double perfect constructions include sentences, such as the following one:

Er hat sich versteckt gehabt.

*He had himself hidden had.

Double perfect constructions have been inconsistently treated in research on linguistics, leading to the fact that they are featured by an unclear grammatical status. Specifically, it is unclear whether double perfect constructions constitute a substandard phenomenon that is redundant and rather used without reflection, or whether they fulfill functions that cannot be covered by the existing tenses and, thus, represent an extension of the tense system (Rödel, 2007; Zybatow & Weskott, 2018). Due to this still unclear grammatical status, double perfect constructions are an interesting topic for linguistic analyses and research and, thus, a particularly suitable topic for an OSL project for linguistics.

In the OSL project of study B, students were asked to write a short grammar article about double perfect constructions during the analyzing activity (see Fig. 3). At the beginning of the analyzing activity, A+ students received an instruction on the use of the descriptive linguistic method. That is, students were asked to write a value-free grammar article in which all grammatical phenomena are just observed and described objectively. They were explicitly asked to formulate no evaluative statements on what is “right” and “wrong” in the German language. The A− students were instead instructed to apply the prescriptive method, whereby they had to write a normative grammar article which aims to sensitize readers against an inappropriate use of language. Hence, A− students were asked to formulate explicit rules that represent what is considered “right” and “wrong” in the German language and a clear recommendation on whether double perfect constructions should be used or not. Students in both conditions received an exemplary article about another grammatical phenomenon that served as an orientation for writing their own grammar article about the phenomenon of double perfect constructions. Moreover, students in both conditions had to work on different tasks of which the results should serve as a base for the grammar article. These tasks consisted of transcribing (i.e., the systematic representation of spoken language in written form representing an essential part of the methodologies of linguistics), reading and summarizing sections of linguistic articles, and analyzing the tense of sentences with double perfect constructions.

Fig. 3
figure 3

Procedure of study B

Full size image

After the analyzing phase and, thus, after writing the article, students worked on the post-questionnaire which also included the assessment of students’ epistemic beliefs about linguistics. In the last phase, the reflection phase, the instructor chose randomly two articles that were written by the students, one of each group, and discussed them together with the whole class. Moreover, students received information on the experimental design of the study and the two different linguistic methods that they used during their OSL visit. Figure 3 shows the procedure of study B.

Measures

To assess students’ epistemic beliefs about the educational (and social) sciencesFootnote 1 in study A and about linguistics in study B, we used the CAEB (Stahl & Bromme, 2007). The CAEB (Connotative Aspects of Epistemic Beliefs) measures students’ evaluative associations with the features of knowledge within a certain discipline by the use of semantic differentials. These semantic differentials measure—based on previous research on epistemic beliefs and two validation studies (see Stahl & Bromme, 2007)—two dimensions of epistemic beliefs: texture and variability. The texture dimension (ten items) refers to beliefs about the structure and accuracy of knowledge. The variability dimension (seven items) refers to beliefs about the stability and dynamics of knowledge.

The CAEB initially intended to assess three dimensions of epistemic beliefs that have traditionally been distinguished, namely simplicity, source, and certainty of knowledge. The dimension simplicity of knowledge describes at its lower levels the naïve view of knowledge as being “discrete, concrete, [and] knowable facts” (Hofer & Pintrich, 1997, p. 120), and at its higher levels the sophisticated view of knowledge as being “relative, contingent, and contextual” (Hofer & Pintrich, 1997, p. 120). The CAEB used semantic differentials, such as sorted-unsorted or structured-unstructured, in order to measure learners’ beliefs about the simplicity of knowledge. Naïve beliefs about the source of knowledge relate to the view of knowledge as being externally provided by authority, while one with more developed beliefs recognizes the personal and social construction of knowledge (Hofer & Pintrich, 1997). Stahl and Bromme (2007) operationalized this dimension with adjective pairs, such as objective-subjective or discovered-negotiated. The lower levels of the dimension certainty of knowledge refer to the rather naïve view about knowledge as being fixed, absolute, and certain, while the higher levels point to a more sophisticated view of knowledge as being fluid, tentative, and evolving (Hofer & Pintrich, 1997). The CAEB used semantic differentials, such as absolute-relative or inflexible-flexible, in order to assess learners’ beliefs about the certainty of knowledge. However, these three dimensions could not empirically be supported in the validation studies of the CAEB. Instead, Stahl and Bromme (2007) identified the two dimensions of texture and variability. The texture dimension mainly unifies the two dimensions simplicity of knowledge and source of knowledge (see Table 1), while the variability dimension unifies most of the items that initially aimed to assess beliefs about the certainty of knowledge and one item of the dimension source of knowledge (see Table 1).

Table 1 Dimensions and semantic differentials of the CAEB (see Stahl & Bromme, 2007)
Full size table

In both studies, students were asked to rate the 17 adjective pairs on a 7-point semantic differential scale as shown in Fig. 4 and were provided with the following written instruction (translated from German to English): Now I ask you to evaluate the [educational sciences/ the linguistics] as science. By that I refer to your views related to the knowledge in the [educational sciences/ the linguistics] in general and not to your personal knowledge in this area. For this purpose, I provide you with a list of different adjective pairs. For instance, one can view knowledge as being “objective” to “subjective”. Please make a judgement for each adjective pair, even if some pairs may seem abstract to you.

Fig. 4
figure 4

CAEB items as administered in study A and B (*items that were reversed for analysis)

Full size image

In both studies, we administered the CAEB in a pre- and post-questionnaire. Specifically, study A assessed students’ epistemic beliefs after a brief introduction at the beginning of the OSL project using a pre-questionnaire, and after a practice phase at the end of the OSL project using a post-questionnaire. Study B administered the pre-questionnaire to measure students’ epistemic beliefs one week before students’ OSL visit at school and the post-questionnaire after the analyzing phase during the OSL project.

Prior to our further analyses, we recoded the direction of six adjective pairs such that “1” always indicates a rather simple and naïve view and “7” a rather sophisticated view about the nature of knowledge within the educational sciences or linguistics. In line with previous studies using the CAEB (e.g., Bientzle et al., 2014), we reverse coded the following items: dynamic-static, temporary-everlasting, flexible-inflexible, refutable-irrefutable, negotiated-discovered, and open-closed.

Based on a reliability analysis of the items, we excluded two texture items (i.e., superficial-profound and discovered-negotiated) and three variability items (i.e., irrefutable-refutable, everlasting-temporary, and stable-unstable) in study A, leading to satisfying internal consistencies of the remaining eight texture items (Cronbach’s α: pre-questionnaire = 0.72; post-questionnaire = 0.77) and rather unsatisfying but reasonable (due to the small number of items) internal consistencies of the remaining four variability items (Cronbach’s α: pre-questionnaire = 0.57; post-questionnaire = 0.59). In study B, we excluded one texture item (i.e., superficial-profound) and one variability item (i.e., irrefutable-refutable), leading to satisfying internal consistencies of the remaining nine texture items (Cronbach’s α: pre-questionnaire = 0.76; post-questionnaire = 0.78) and almost satisfying internal consistencies of the remaining six variability items (Cronbach’s α: pre-questionnaire = 0.65; post-questionnaire = 0.62).

To achieve our second research objective of exploring the relation between students’ epistemic beliefs and other learning-related variables, the post-questionnaires additionally assessed students’ perceived authenticity of the learning activity and their situational interest in the contents of the OSL projects at the end of students’ OSL visit in study A and study B. Study A, moreover, measured students’ knowledge about the contents of the project (i.e., causal versus correlational evidence) at the end of the experiment (see Fig. 2). To measure perceived authenticity of the learning activity, we used eight items (Cronbach’s α: study A = 0.86, study B = 0.81) adapted from Gulikers et al. (2006). Students replied to items such as “I think that the tasks in this project fit well with the work of real scientists” on a scale of 1 (strongly disagree) to 5 (strongly agree). To assess situational interest in the contents of the OSL project, we used the scale from Knogler et al. (2015) with each six items measuring triggered (e.g., To what extent was the engagement with the contents of this project exciting for you? Cronbach’s α: study A = 0.88, study B = 0.93) and maintained (e.g., To what extend would you like to learn more about certain topics of the project? Cronbach’s α: study A = 0.77, study B = 0.83) situational interest. Students replied on a five-point Likert scale ranging from 1 (not at all) to 5 (greatly). To measure knowledge about causal and correlational evidence in study A, we used eight items (total score: 0–21 point) that asked students to reproduce, apply, or transfer the concepts they were instructed on during the instruction phase. Two raters coded around 20% of the knowledge tests (i.e., n = 35) and reached a high agreement between their ratings (ICCabsolute = 0.93; 95% CI [0.85, 0.96]). For more information on the knowledge test, see Nachtigall et al. (2020).

Results

Research question 1

Study A: educational science

With respect to study A, we investigated whether the authenticity level of the learning activity (i.e., PF vs. DI) had an impact on students’ development of epistemic beliefs about educational (and social) sciences (RQ1). For this purpose, we conducted a mixed ANOVA (for each of the two dimensions of epistemic beliefs) with condition as between-subject factor, the time of measurement as within-subject factor, and students’ epistemic beliefs as dependent variable. With regard to students’ epistemic beliefs about the texture of knowledge within the educational sciences, the mixed ANOVA revealed no main effect of condition (F(1150) = 1.41, p = 0.24, ηp2 = 0.009), no interaction effect of condition and time of measurement (F(1150) = 2.51, p = 0.12, ηp2 = 0.016), but a significant medium-sized main effect of time of measurement (F(1150) = 14.68, p < 0.001, ηp2 = 0.089) on students’ development of epistemic beliefs. Specifically, both A+ students (pretest: M = 3.58, SD = 0.66; posttest: M = 3.70, SD = 0.77) and A− students (pretest: M = 3.36, SD = 0.81; posttest: M = 3.63, SD = 0.94) developed more sophisticated views about the texture of knowledge by the end of their OSL visit. As Fig. 5 shows, by the end of their OSL visit, students described the nature of knowledge within the educational sciences as being in particular more imprecise, vague, and unconfirmable than at the beginning of their OSL visit.

Fig. 5
figure 5

Polarity profiles of student responses (means) to the texture items in the A+ condition (left) and the A− condition (right) in study A

Full size image

Regarding students’ epistemic beliefs about the variability of knowledge, the mixed ANOVA revealed no main effect of condition (F(1150) = 0.60, p = 0.81, ηp2 = 0.000), no main effect of time of measurement (F(1150) = 0.03, p = 0.87, ηp2 = 0.000), and no interaction effect of condition and time of measurement (F(1150) = 0.05, p = 0.82, ηp2 = 0.000) on students’ development of epistemic beliefs. That is, neither A+ students (pretest: M = 4.85, SD = 0.77; posttest: M = 4.83, SD = 0.78) nor A− students (pretest: M = 4.81, SD = 0.99; posttest: M = 4.81, SD = 0.90) changed their views about the variability of knowledge within the educational (and social) sciences during their OSL visit (see Fig. 6).

Fig. 6
figure 6

Polarity profiles of student responses (means) to the variability items in the A+ condition (left) and the A− condition (right) in study A

Full size image

Study B: linguistics

To examine whether the authenticity level of the learning activity (i.e., descriptive vs. prescriptive) had an impact on students’ development of epistemic beliefs about linguistics (RQ1), we again conducted a mixed ANOVA (for each of the two dimensions of epistemic beliefs) with condition as between-subject factor, the time of measurement as within-subject factor, and students’ epistemic beliefs as dependent variable. With respect to students’ epistemic beliefs about the texture of knowledge within the linguistics, the mixed ANOVA revealed no main effect of condition (F(1146) = 1.22, p = 0.27, ηp2 = 0.008), no main effect of time of measurement (F(1146) = 0.13, p = 0.72, ηp2 = 0.001), and also no interaction effect of condition and time of measurement (F(1146) = 0.38, p = 0.54, ηp2 = 0.003). Specifically, neither A+ students (pretest: M = 3.47, SD = 0.83; posttest: M = 3.45, SD = 1.04) nor A− students (pretest: M = 3.55, SD = 0.86; posttest: M = 3.63, SD = 0.79) developed more sophisticated views about the texture of knowledge within the linguistics by the end of their OSL visit. As Fig. 7 shows, by the end of their OSL visit, students in both conditions perceived knowledge within the linguistics only as being slightly more unsorted and relative than at the beginning of their OSL visit.

Fig. 7
figure 7

Polarity profiles of student responses (means) to the texture items in the A+ condition (left) and the A− condition (right) in study B

Full size image

With regard to students’ epistemic beliefs about the variability of knowledge, the mixed ANOVA revealed no main effect of condition (F(1145) = 0.43, p = 0.52, ηp2 = 0.003), but a significant small-sized main effect of time of measurement (F(1145) = 5.36, p = 0.02, ηp2 = 0.036) and a significant small-sized interaction effect of condition and time of measurement (F(1145) = 7.41, p = 0.007, ηp2 = 0.049). That is, in contrast to A+ students (pretest: M = 4.28, SD = 0.91; posttest: M = 4.25, SD = 0.96), A− students (pretest: M = 4.16, SD = 0.97; posttest: M = 4.55, SD = 0.91) developed more sophisticated epistemic beliefs about the variability of knowledge within the linguistics during their OSL visit (see Fig. 8).

Fig. 8
figure 8

Polarity profiles of student responses (means) to the variability items in the A+ condition (left) and the A− condition (right) in study B

Full size image

Research question 2

Study A: educational science

To explore the relations between students’ epistemic beliefs about educational (and social) sciences and other variables (RQ2), we calculated correlations between students’ epistemic beliefs reported at the end of their OSL visit and their perceived authenticity of the learning activity, their situational interest in the contents of the OSL project, and their knowledge about the contents of the project.Footnote 2 The results of the correlational analyses, which are illustrated in Table 2, show significant negative correlations between students’ epistemic beliefs about the texture of knowledge and their perceived authenticity and triggered as well as maintained situational interest. Thus, the more sophisticated students’ epistemic beliefs about the texture of knowledge within the educational (and social) sciences, the lower their perceived authenticity of the learning activity and their situational interest in the contents of the project. Separated by condition, the correlations with perceived authenticity and triggered situational interest can only be found for students in the A− condition. In the A+ condition, students’ epistemic beliefs merely correlate with their maintained situational interest. An association between students’ epistemic beliefs about the texture of knowledge and their performance on the knowledge test is absent in both conditions.

Table 2 Correlations between students’ epistemic beliefs about knowledge within the social or educational sciences and other variables (Note: light gray: significant small correlation; dark gray: significant moderate correlation)
Full size table

For students’ epistemic beliefs about the variability of knowledge, the correlational analysis shows significant positive correlations with their perceived authenticity and their knowledge-test performance (see Table 2). That is, the more sophisticated students’ epistemic beliefs about the variability of knowledge within the educational (and social) sciences, the higher their perceived authenticity of the learning activity and their performance on the knowledge test. While the correlation with perceived authenticity only applies to A+ students, the association with the knowledge-test performance can be found in both conditions. Students’ epistemic beliefs about the variability of knowledge did not correlate with their triggered or maintained situational interest in neither of the two conditions.

Study B: linguistics

Regarding the interplay between students’ epistemic beliefs in linguistics and their perceived authenticity as well as their situational interestFootnote 3 (RQ2), the results of the correlational analysis only demonstrate a negative significant correlation between A+ students’ epistemic beliefs about the texture of knowledge and their perceived authenticity (see Table 3). Hence, the more sophisticated A+ students’ epistemic beliefs about the texture of knowledge within the linguistics, the lower their perceived authenticity of the learning activity. The analysis revealed no further significant correlations.

Table 3 Correlations between students’ epistemic beliefs about linguistics and other variables (Note. light gray, significant small correlation)
Full size table

Discussion

Non-formal learning settings, such as science museums, science centers, science outreach programs, or out-of-school labs (OSLs), often aim to situate students into the role of being a scientist and to offer students authentic insights into the ways how scientists think and work and, thus, how they gather new knowledge (e.g., Glowinski & Bayrhuber, 2011; Luehmann, 2009). Thereby, it is likely that these settings with their implemented authentic learning activities may affect learners’ epistemic beliefs. Consequently, research on the effectiveness of OSLs has begun to investigate the impact of OSLs on students’ development of domain-specific epistemic beliefs and has revealed inconsistent findings (e.g., Mierwald et al., 2018; Schiefer et al., 2017). Against this background and in light of the findings by Palmer and Marra (2004) suggesting that students’ development of highly sophisticated epistemic beliefs about the social sciences and humanities may be particularly difficult, the present two studies investigated the effectiveness of authentic learning activities for students’ development of epistemic beliefs in two different OSL projects for social sciences and humanities (RQ1). We further investigated whether students’ epistemic beliefs traditionally labeled as sophisticated indeed correlate with other variables that have been focused on by research on OSLs (RQ2), namely students’ perceived authenticity of the learning activity, their situational interest, and their knowledge acquisition (study A only).

Impact of authentic learning activities on epistemic beliefs

Our results with respect to RQ1 suggest, contrary to our expectations, that authentic learning activities in OSLs that try to emulate scientific practices do not have an impact on students’ development of sophisticated epistemic beliefs about educational sciences and linguistics. Specifically, in study A, students from both conditions and, thus, independent from the authenticity level of the learning activity developed more sophisticated epistemic beliefs about the texture of knowledge within the educational sciences by the end of their OSL visit. Regarding students’ epistemic beliefs about the variability of knowledge, the findings demonstrated no significant change. In study B, students in both conditions did not significantly change their epistemic beliefs about the texture of knowledge related to linguistics during their OSL visit, but A− students (i.e., students who participated in the less authentic learning activity) developed more sophisticated beliefs about the variability of knowledge.

A potential reason for the findings of study A may relate to the instruction phase that was almost identical in both conditions and compared and contrasted typical conceptions of students about the adequacy of correlation studies for examining a causal hypothesis and the actual explanatory power of correlational evidence. The instruction discussed the differences between correlational and causal evidence and the common mistake of interpreting correlational evidence as causal. Thus, students in the A+ condition as well as students in the A− condition were confronted with somehow conflicting perspectives and the fact that data from educational science studies can be differently and sometimes incorrectly interpreted. Hence, the instruction alone may have constituted a direct intervention (see Kienhues et al., 2016) in both conditions and led to a change in A+ and A− students’ beliefs about the structure and accuracy (i.e., texture) of knowledge within the educational sciences.

In line with the previous possible explanation for the findings of study A, a reason for the findings of study B suggesting that A− students developed more sophisticated beliefs about the variability of knowledge within linguistics than A+ students may also relate to the learning materials. Specifically, the exemplary article in the A− condition focused on criticizing the linguistic change towards the incorrect use of a certain German grammatical phenomenon. Hence, although the article very clearly exemplified a prescriptive evaluation of an inappropriate use of language and, thus, argued for the invariability of correct language use, it also built on the observation that the usage of language has actually changed. That is, the exemplary article in the A − condition made the change of language as part of a critique on the incorrect usage of language very explicit. In contrast, the exemplary article in the A+ condition described a grammatical phenomenon in a non-evaluative manner and, thereby, only implicitly addressed changes in the usage of language. This might be the reason why—in contrast to our intentions—only A− and not A+ students changed their beliefs about the dynamics of knowledge within the linguistics.

The further findings of the present two studies showing that the authenticity of the learning activity did not have an effect on students’ epistemic beliefs about the variability of knowledge in study A and students’ development of epistemic beliefs about the texture of knowledge in study B could also be seen as a support for the findings by Palmer and Marra (2004). Their findings suggested that students’ development of highly sophisticated epistemic beliefs in the social sciences and humanities might be particularly difficult. Thus, it would be important to find powerful ways for fostering students’ development of highly sophisticated epistemic beliefs in the social sciences and humanities and, moreover, to investigate whether this development indeed depends on the domain and differs between the social sciences/humanities and the natural sciences.

Interplay between epistemic beliefs, perceived authenticity, situational interest, and knowledge

The findings with respect to our RQ2 suggest highly differential relations between students’ epistemic beliefs and other variables depending on the dimension to which students’ beliefs relate (texture vs. variability), the level of authenticity of the learning activity in which they were engaged (A+ vs. A−), and the domain to which their beliefs refer to (educational science/study A vs. linguistics/study B). Figure 9 gives an overview of these differential relations.

Fig. 9
figure 9

Overview of significant correlations between epistemic beliefs (EB) and other variables in both studies. Note. (1) Up arrow, positive correlation with EB; down, negative correlation with EB; (2) dark gray, moderate correlation with EB; light gray, small correlation with EB; (3) filled, A+ condition; unfilled, A− condition

Full size image

With regard to students’ perceived authenticity of the learning activity, our findings show significant negative correlations with students’ epistemic beliefs about the texture of knowledge in both the A+ condition of study B and the A− condition of study A, while their epistemic beliefs about the variability of knowledge correlate positively with students’ perceived authenticity in the A+ condition of study A. The findings of study A partly align with our initial intentions, as they suggest that, on the one hand, the higher the perceived authenticity of an actually highly authentic activity (in the A+ condition), the more sophisticated are students’ epistemic beliefs and, on the other hand, the higher the perceived authenticity of an actually barely authentic activity (in the A− condition), the more naïve are students’ epistemic beliefs. The correlation in the A+ condition is in line with the model of authenticity in teaching and learning contexts by Betz et al. (2016) assuming a positive relation between learners’ perceived authenticity of the learning activity and the learning outcomes of a learning activity that is intended to be authentic. In contrast, the correlation in study B suggests that the lower the perceived authenticity of an actually highly authentic activity (in the A+ condition), the more sophisticated are students’ epistemic beliefs. Or in other words, the higher the perceived authenticity, the more naïve the epistemic beliefs. This unexpected finding is surprising and might, at the first glance, support the findings by Bråten et al. (2008) showing that naïve epistemic beliefs can be superior to sophisticated beliefs in certain circumstances. However, the A+ activity in study B asked students to use linguistic knowledge in a non-objective, not absolute, and indefinite way in order to describe grammatical phenomena in a neutral and non-evaluative manner. Consequently, perceiving this activity as one that authentically emulates scientific practices within the linguistics should go along with sophisticated epistemic beliefs indicating that knowledge within the linguistics is seen as rather non-objective, not absolute, and indefinite.

These differential correlations between learners’ perceived authenticity and their epistemic beliefs further suggest the need for a particular interplay between intended authenticity and perceived authenticity. This interplay appears to be different when investigating the effects of authentic learning on epistemic beliefs than when investigating the effects on other variables, such as situational interest. Specifically, perceived authenticity is usually assumed to mediate the effects of the learning setting on certain outcome variables (see Fig. 1; Betz et al., 2016). Thus, it is hypothesized that only when an authentic learning setting is perceived as authentic will it have a positive effect on certain outcomes; and only when the less authentic or non-authentic learning setting is perceived as less authentic or non-authentic will it have no positive effect on the target outcome variable. With respect to epistemic beliefs, however, the interplay appears to be different. Based on the results of the present two studies, it can be assumed that the intended authenticity of the learning activity moderates the relationship between perceived authenticity and epistemic beliefs. It is likely that when learners perceive learning activities as authentic, they are probably willing to accept the resembled ways of thinking and working as valid scientific practices and to adopt corresponding beliefs about the nature of scientific knowledge. The direction of this relationship may be influenced by the intended authenticity of the learning activity. That is, if the intended authenticity of the learning activity is high, the relationship between learners’ perceived authenticity and their epistemic beliefs should be positive. Thus, the higher learners’ perceived authenticity of an actually authentic activity, the more sophisticated their epistemic beliefs (as suggested by the results of study A). However, if the intended authenticity of the learning activity is low, the relationship between perceived authenticity and epistemic beliefs should be negative. Thus, the more learners perceive a learning activity that is actually low in authenticity as authentic, the more naïve their epistemic beliefs (as suggested again by the results of study A).

A further assumption that follows from the discussion of the particular interplay between intended and perceived authenticity in the context of epistemic beliefs relates to learners’ prior epistemic beliefs. Specifically, one could assume that the interaction between intended authenticity and perceived authenticity is influenced by learners’ prior epistemic beliefs. That is, learners with more sophisticated epistemic beliefs are more likely to recognize the actual level of authenticity of the learning activity. In contrast, learners with naive epistemic beliefs are more likely to fail to recognize the actual level of authenticity of the learning activity. Results from post hoc correlational analyses partially support this assumption. Specifically, in study A, students’ prior epistemic beliefs about the texture of knowledge within the educational sciences are negatively correlated with their perceived authenticity of the actually inauthentic learning activity in the A− condition (r =  − 0.35, p = 0.002). Thus, the more naïve the students’ prior beliefs, the higher their perceived authenticity of the learning activity in the A− condition, and the more sophisticated their prior beliefs, the lower their perceived authenticity.

With respect to students’ situational interest, the findings of study B show no significant correlations with students’ epistemic beliefs. The findings of study A, however, demonstrate that the more sophisticated students’ epistemic beliefs about the texture of knowledge within the educational sciences, the lower their triggered situational interest in the A− condition and the lower their maintained situational interest in the A+ condition. Although these significant negative correlations contrast with previous research pointing to positive relations between students’ sophisticated epistemic beliefs and motivational constructs (Chen & Barger, 2016; Guo et al., 2022; Kapitza, 2020; Schiefer et al., 2020), they may also point to the fact that when students become aware of the complexity of a learning domain, they may be discouraged from (re)engaging with it.

Regarding students’ knowledge acquisition, the results of the correlational analysis demonstrate a positive association between A+ and A− students’ epistemic beliefs about the variability of knowledge within the educational sciences and their performance on the knowledge test administered in study A. Although these findings are in line with previous research showing positive relations between learners’ epistemic beliefs and other cognitive learning outcomes (Bråten et al., 2011), it is surprising that the relationship that we found in study A only applies to students’ beliefs about the variability and not also about the texture of knowledge.

Taken together, our findings regarding the interplay between epistemic beliefs on the one hand and perceived authenticity, situational interest, and knowledge acquisition on the other hand point to the need to further investigate these differential and presumably also highly interdependent relations.

Limitations

One limitation relates to the quasi-experimental design of study A, whereby the two test groups may be not equivalent. To address this limitation and, thus, to account for the hierarchical structure of the data of study A, we additionally conducted multilevel regression analyses (see supplementary materials). The results of the multilevel regression analyses are in line with the findings of the mixed ANOVAs reported in this paper and reveal no significant effect of the authenticity level of the learning activity on neither of the two dimensions of students’ epistemic beliefs, but a significant effect of time of measurement on students’ epistemic beliefs about the texture of knowledge.

A further limitation may relate to the use of only one quantitative instrument (i.e., the CAEB) with a very specific and potentially rather abstract format (i.e., semantic differentials) for assessing students’ epistemic beliefs. As the CAEB is described as a promising instrument for research in OLSs (Kremer & Kapitza, 2020) and as a scale that can be easier and quicker administered than other instruments (Mason, 2016; Stahl & Bromme, 2007), we decided to use the CAEB in the present two studies. However, looking at the previous OSL studies that investigated effects on epistemic beliefs, it can be noted that they all used a Likert-scale questionnaire for measuring students’ epistemic beliefs. Specifically, Schiefer et al., (2017, 2020) and Kapitza (2020) administered the questionnaire developed by Conley et al. (2004), while Mierwald et al. (2018) adapted the Beliefs about History Questionnaire from Maggioni et al. (2004). Kapitza (2020) additionally used concept cartoons to qualitatively capture students’ beliefs. Therefore, one might assume that using a different instrument or combining the CAEB with other quantitative or qualitative instruments to measure students’ epistemic beliefs might have led to different results.

The fact that the present two studies partly demonstrated no effects of the experimental manipulations could be seen as a further limitation, because null effects are difficult to interpret. Nevertheless, we think it is of high importance to also publish studies that demonstrate no significant effects in order to encounter the publication bias prevalent within the social sciences (see Franco et al., 2014).

Conclusion

The findings of our two studies demonstrate that the level of authenticity of the learning activity does not seem to play a role in promoting students’ development of sophisticated epistemic beliefs about educational science and linguistics as two exemplary disciplines of the social sciences and humanities. Based on the discussion above, it can be assumed that learners’ epistemic beliefs are likely to be affected by the design of the learning materials. Consequently, it might be an interesting avenue for future research to investigate, through an experimental manipulation, whether direct interventions, such as authentic learning materials with conflicting information (as, for example, in the OSL study by Mierwald et al., 2018), are more effective than indirect interventions, such as authentic learning activities that mimic the way scientists gather new knowledge, as for instance in the two studies presented here.

Our findings from the correlational analyses, however, suggest that the authenticity of the learning activity does play a role in students’ epistemic beliefs. Specifically, as our results show, learners’ perceived authenticity of the learning activity is significantly related to their epistemic beliefs about both the texture and variability of knowledge, especially within the educational sciences. These associations appear to be highly differential, meaning that the presence and direction of these associations depend on the dimension of students’ epistemic beliefs, the level of authenticity of the learning activity, and the learning domain. The same applies to the correlations between epistemic beliefs on the one hand and students’ situational interest and knowledge acquisition on the other hand. Thus, further research is required to uncover and explain these associations, and especially how learners’ epistemic beliefs affect and are affected by other learning-related constructs.

Our findings contribute to research on both authentic learning in OSLs and the development and nature of domain-specific epistemic beliefs. In the context of research on authentic learning in OSLs, our studies emphasize that it might be particularly suitable and fruitful to focus on investigating how OSLs can promote sophisticated epistemic beliefs about scientific disciplines due to the following three reasons: (1) Authenticity in OSLs aims to emulate the work of scientists. As complexity, novelty, and uncertainty are inherent in scientific endeavors to develop new knowledge, an authentic reflection of these processes is likely to affect epistemic beliefs. Although the level of authenticity of the learning activity did not affect students’ epistemic beliefs in the present two studies, the level of authenticity of other design elements of the OSL setting, such as the materials that students use, may affect epistemic beliefs. (2) The role of authenticity in promoting sophisticated epistemic beliefs is further supported by our correlational analyses, as these point to relations between students’ perceived authenticity of the learning setting and their epistemic beliefs. (3) Moreover, as suggested by previous research and the findings of the present two studies, epistemic beliefs are related to motivational and cognitive learning outcomes that have been the focus of OSL research to date. Thus, in order to promote interest in and knowledge of science, it might be fruitful to also promote sophisticated epistemic beliefs. Beyond these reasons for investigating the effects of OSLs on epistemic beliefs, our findings provide interesting directions for further developing and testing a theory of the effects and mechanisms of authentic learning in OSLs. As our discussion suggests, the interplay between intended and perceived authenticity in the context of fostering epistemic beliefs seems to be different from other contexts that have been mainly investigated so far (e.g., fostering interest in science), and thus also seems to be inconsistent with the model of authenticity developed by Betz et al. (2016). Our discussion also highlights the need to investigate the role that learners’ prerequisites (i.e., their prior epistemic beliefs) play in authentic learning contexts, which is a component of Betz et al.’s (2016) model but has not yet been the focus of research on OSLs. In the context of research on epistemic beliefs, our studies provide two interesting avenues for further research: (1) comparing the effectiveness of direct versus indirect interventions and (2) investigating the direction of the relationship between epistemic beliefs and other learning-related constructs.