Which Epistemic Processes Occur When Pre-Service Teachers Reflect on Practitioners’ Misconceptions?

Abstract

Reflection can help pre-service teachers in practical phases to process anecdotal evidence (and possible misconceptions contained therein) appropriately for their professional development. In an exploratory approach, we analyze the think-aloud protocols of eleven pre-service teachers for primary education who reflect on anecdotal evidence. First, we examine whether and to what extent they identify and address misconceptions. Then, we use the AIR model of epistemic cognition to investigate which epistemic processes occur when pre-service teachers reflect on anecdotal evidence. Finally, through a qualitative comparison of three protocols, we explore possible connections between the use of epistemic processes and the processing of misconceptions. While all participants noticed a misconception from the field of educational sciences and dealt with it to varying degrees of proficiency, the subject-specific didactic misconception was hardly addressed despite causing irritation, and the content-specific misconception remained unnoticed. We were able to differentiate epistemic processes in the subcategories of noticing, reasoning, conclusion, and alternative actions as well as application of epistemic ideals. Pre-service teachers who recognized a misconception and elaborated on it with scientific knowledge seem to use epistemic processes differently, which is more concretely related to the given lesson plan. We discuss the results with regard to possible implications for university teacher training and further research desiderata.

1. Introduction

Practical phases are a crucial part of university teacher education, in which pre-service teachers can acquire relevant experiential knowledge about learning and instruction in school (Resch et al., 2024). For their professionalization process, it is essential to reflect on these experiences (Berliner, 2004) as it can help bridge the gap between theory and practice (Beauchamp, 2015; Weber et al., 2023). Reflection that is based on scientific knowledge may have different positive effects on the professionalization process of pre-service teachers: they may deepen their understanding of pedagogical situations and develop elaborate ideas for pedagogical action (Visscher & Coe, 2013), the uncertainty of pedagogical situations may be reduced (Hinzke et al., 2021), and pedagogical misconceptions may be recognized and processed (Steins et al., 2022). All these aspects touch on the issue of combining theory and practice in teacher training, which continues to be a major problem for many prospective teachers in practical phases, e.g., by working through the theory-practice shock by adapting to given teaching practice and rejecting an acquisition of theoretical reflective knowledge (Fischer et al., 2018).

The handling of anecdotal evidence seems particularly important to us, given that pre-service teachers regard anecdotal evidence as an important source of knowledge (Menz et al., 2021a). If pre-service teachers adopt the practices they observe without question, less valid knowledge is applied in practice (Gruber et al., 2006), which may have a negative impact on the quality of teaching (Rochnia et al., 2023). An incorrect handling of information can lead to misconceptions or distorted mental models (Richter & Maier, 2017). Simply adopting standardized procedures does not do justice to the complexity of teaching (Loughran, 2019). Practitioners may also put outdated or inaccurate knowledge into action, which is then passed on to pre-service teachers. This leads to the risk that using anecdotal evidence might transfer or even generate misconceptions. Therefore, pre-service teachers should be critical with anecdotal evidence as a source of knowledge.

Reflection can help to process anecdotal evidence adequately, the outcome of which can then be used for one’s own professionalization. Although there is a lot of research on pre-service teachers’ reflection, what exactly pre-service teachers need to do in this reflection to recognize misconceptions in anecdotal evidence has not yet been investigated. To approach this question, we draw on the AIR model of epistemic cognition (Chinn et al., 2014) and analyze occurring epistemic processes and the processing of given misconceptions in think-aloud data of pre-service teachers (who reflect on a lesson planned by a novice practitioner). In what follows, we outline the state of research on pre-service teachers’ reflection and our underlying theoretical approach.

1.1. Reflection in Practical Phases of Teacher Education

Reflecting on pedagogical situations and classroom practices can initiate the evidence-based thinking and action of pre-service teachers (Schellenbach-Zell et al., 2023; Schneider Kavanagh et al., 2020). As a form of pedagogical reasoning, reflection is “the thinking that underpins informed professional practice” (Loughran, 2019, p. 526), so there is a clear focus on the knowledge base underlying the pedagogical action. Reflection is systematic thinking with a specific aim, similar to problem solving (Dewey, 1933). In the context of teaching, it incorporates “processes of examining, understanding, analyzing, and evaluating certain objects (e.g., practice, experience, or information) with the goal of thinking more clearly, altering strategies, and improving action plans” (Hartmann et al., 2021, p. 1). We therefore assume a process model of reflection.

Weber et al. (2023) identify “three fundamental skills for reflection: the ability to notice and attend to important events in the classroom, reasoning about these events, and the proposition of alternative teaching strategies” (p. 2). They address two important terms of professional vision, specifically teacher noticing and reasoning. Professional vision is characterized by two competencies: (1) noticing, which means the ability to identify events “that are of importance for teaching and learning in classrooms” (Seidel & Stürmer, 2014, p. 742), and (2) (knowledge-based) reasoning, which encompasses the processes of description, explanation, and prediction of the identified event (Seidel & Stürmer, 2014). Noticing itself comprises various facets such as attending to a situation or describing it (Farrell et al., 2024; König et al., 2022). In addition to developing alternative courses of action, we consider it equally important to draw conclusions for the professional self, since it is important to not only implement epistemically informed action but also develop an epistemic stance (Buehl & Fives, 2016). Lohse-Bossenz et al. (2023) also point out that self-reference is an important distinguishing feature of reflection from lesson analysis. The authors were able to show that university students who also made self-references developed higher self-efficacy beliefs (in relation to classroom management) than those who carried out a pure lesson analysis.

For our research, we build on the work of Csanadi et al. (2021) and Schellenbach-Zell et al. (2023): when analyzing reflections, one can investigate the content level, which focuses on the knowledge that pre-service teachers apply when reflecting, or the process level, “that is characterized by an engagement in certain epistemic processes” (Csanadi et al., 2021, p. 149).

1.2. Content Level of Reflection (Dealing with Anecdotal Evidence)

In analyzing the content level of reflection, we focus on the knowledge of pre-service teachers. The professional knowledge of teachers is defined as consisting of the domains content knowledge (CK), pedagogical content knowledge (PCK), and pedagogical psychological knowledge (PPK) (Baumert & Kunter, 2006; Shulman, 1986; Voss et al., 2011). In order to develop an adequate mental model (Johnson-Laird, 1983), which is not distorted by false or biased information (Richter & Maier, 2017), a mindful integration of information from different sources is necessary (Rousseau & Gunia, 2016). In the sense of a moderate understanding of evidence, this includes theories and models in addition to empirical studies (R. Stark, 2017; Wilkes & Stark, 2023). A mindful integration may also include personal evidence (e.g., use of diagnostic tests; Rousseau & Gunia, 2016).

In practical phases, pre-service teachers are frequently confronted with a certain type of knowledge when engaging with practitioners: anecdotal evidence includes “both personal observations and narratives from others, that is, personal or vicarious experiences” (Menz et al., 2021b, p. 146). According to Menz et al. (2021a), “literature indicates that anecdotal evidence plays a vital role in (pre-service) teachers’ professional lives whereas empirical evidence and research seem to be less important” (Menz et al., 2021a, p. 481). Also, when it comes to gaining practical knowledge, university students perceive teachers as being more competent, benevolent, and integrity than researchers (Hendriks et al., 2021).

Anecdotal evidence, therefore, is an important source of knowledge for pre-service teachers. At the same time, it is particularly prone to containing misconceptions that are passed on through communication or observed classroom action. Menz et al. (2021b) examined sources of educational psychological misconceptions among pre-service teachers and found that they frequently referred to anecdotal evidence as a source. In addition, participants who refer more to anecdotical evidence have more misconceptions than university students who refer more to scientific evidence in Menz et al.’s (2021b) study. In case anecdotal evidence contains misconceptions, this may be an issue; the problem with misconceptions is that they are firmly embedded in mental models: information that does not fit into these models is often either not recognized or recognized but not integrated into the model—especially when there is no alternative explanation (Steins et al., 2022).

Research on teacher misconceptions has shown that refutations are effective and can deconstruct misconceptions (Menz et al., 2021a). Error-based approaches to evidence-based problem solving and the processing of other people’s errors (advocatory error; Oser et al., 1999) also appear useful to deconstruct misconceptions (Klein et al., 2017; L. Stark et al., 2023). However, each misconception or error must be addressed individually; a course of study cannot achieve this for lifelong occurring mistakes and misconceptions. Reflecting on anecdotal evidence (e.g., pedagogical situations or lesson plans), in contrast, allows pre-service teachers to independently check their knowledge and deconstruct misconceptions. However, at the content level, there has been little research on how pre-service teachers deconstruct misconceptions in reflections (Patry, 2014).

1.3. Process Level of Reflection (Epistemic Processes)

In addition to the content of a reflection, information processing also plays a role in how reflection contributes to a pre-service teacher’s professionalization. In analyzing the process level of reflection, we build on the concept of epistemic cognition. Epistemic cognition encompasses all processes involving knowledge and the acquisition, judgment, processing, or application of knowledge (Buehl & Fives, 2016).

In our research, we refer to the AIR model of epistemic cognition that distinguishes and describes three components of epistemic cognition (Chinn et al., 2014): aims and values, ideals, and reliable epistemic processes. Reliable processes are causal schemes that specify how knowledge and other epistemic products are produced in a reliable way. Part of a reliable process is knowing the circumstances under which it works or does not work (Buehl & Fives, 2016; Chinn et al., 2014). Reliable processes may encompass implicit and explicit “strategies, skills, or other cognitive actions” (Barnes et al., 2020, p. 2) for acquiring and processing information. In the reflection process, the three reflection skills noticing, reasoning, and generating alternative actions (Weber et al., 2023) may each require the use of different (reliable) epistemic processes.

Aims and values describe goals that motivate people to engage with information (e.g., to understand a pedagogical situation) and the importance ascribed to these goals (Barnes et al., 2020; Chinn et al., 2014). Buehl and Fives (2016) elaborate on epistemic cognition for teachers; they emphasize the dual task of teachers themselves to learn but also to teach, which manifests in two types of epistemic aims. This can be illustrated by an example of a reflection on lesson planning: when reading another practitioner’s lesson plan, a pre-service teacher may set two different epistemic aims, namely one focused on personal learnings (I want to understand the situation; epistemic stance), and one aimed at the learning of learners (I want to develop alternative courses of action; epistemically informed practice, Buehl & Fives, 2016). These aims may be represented in different epistemic processes.

Epistemic ideals describe criteria individuals use to evaluate progress towards an epistemic aim or an achievement of an epistemic aim (e.g., true explanations are not threatened by counterevidence; Chinn et al., 2014). These ideals are shaped by underlying beliefs about the nature of knowledge and knowing (Barnes et al., 2020). We assume that specific epistemic processes are used for the application of epistemic ideals in reflection, which may include evaluative processes (Barzilai & Chinn, 2018).

The apt-AIR model (Barzilai & Chinn, 2018), an extension of the original AIR model to include the level of epistemic performance and its promotion (Barzilai & Chinn, 2024), describes “five interweaving aspects of competent engagement with epistemic aims, ideals, and processes” (Mor-Hagani & Barzilai, 2022, p. 2), two of which are “cognitive engagement in epistemic performance” (Mor-Hagani & Barzilai, 2022, p. 2) and “adapting epistemic performance” (Mor-Hagani & Barzilai, 2022, p. 2). Both include the ability to consciously choose and use (cognitive) epistemic processes to achieve epistemic aims. In order to foster pre-service teachers’ abilities to do so, which means to foster their epistemic growth (Mor-Hagani & Barzilai, 2022), it is important to know which processes lead to the successful achievement of an epistemic aim and subsequently an epistemic stance (Buehl & Fives, 2016).

Given that epistemic emotions, such as curiosity or confusion, influence the processes of self-regulated learning (Muis et al., 2018), we assume that they also influence the use of different epistemic processes in reflection. For this reason, we include them in our study, although epistemic emotions are not explicitly mentioned in the AIR model.

Previous research has examined the epistemic processes of (pre-service) teachers in various ways: Barnes et al. (2020) examine which components of epistemic cognition teachers use for classroom assessment using think-aloud data. The authors show that epistemic processes are complex, and that engaging in epistemic processes often takes a long time and is interrupted. Csanadi et al. (2021) examined the extent to which pre-service teachers engaged in epistemic processes of scientific reasoning when solving a pedagogical problem. The use of processes depends on working in dyads or alone. Based on this research, Schellenbach-Zell et al. (2023) examined the processes of pre-service teachers who reflect on a pedagogical situation. They distinguished analysis-related processes with the subcodes “description of theory”, “use of scientific evidence”, and “interpretation”, and action-related processes with the subcodes “formulation of action alternatives” and “derivations for one’s own professionalization”. Giving prompts and feedback can improve the use of epistemic processes, in the sense of a deeper engagement (more description of theories, more interpretation of situations, more implications for one’s own professional development) (Schellenbach-Zell et al., 2023).

None of the research studies establishes links between epistemic processes (process level of reflection) and contents (content level of reflection). Considering the AIR model, however, it seems worthwhile to take a closer look at this, as a reliable application of epistemic processes can lead to the achievement of aims and values (e.g., recognizing misconceptions in lesson planning).

1.4. Research Questions

In light of this theoretical background, we pose the following exploratory research questions. The first research question concerns the content level of reflection:

(RQ1) Do pre-service teachers recognize (noticing) and process (reasoning, concluding, alternative actions) the practitioner’s misconceptions in their reflection?

The second research question concerns the process level of reflection:

(RQ2) Which epistemic processes can be found in pre-service teachers’ reflections?

In the combination of both levels, we pose the third research question:

(RQ3) Are there indications of connections between the use of epistemic processes and the recognition/processing of the misconceptions?

2. Materials and Methods

2.1. Participants and Study Design

Since our research aims to capture a broad spectrum of perspectives and the use of epistemic processes in recognizing misconceptions has been scarcely investigated, we pursue a qualitative, exploratory approach.

Our sample consisted of eleven pre-service teachers in their master’s program in Germany, who aim at becoming primary school teachers and teaching the subject of primary science education. The participants were, on average, 26.45 years old (SD = 4.80) and in their second or third master’s semester (M = 2.82; SD = 1.57). Nine participants identified as female, and two as male.

During a five-month practical phase, these pre-service teachers created five written reflections on pedagogical situations. They were trained to write a knowledge-based reflection with the help of a video depicting an authentic lesson in a two-day preparatory workshop before the beginning of the practical phase. The participants were neither given specific literature nor guidelines or restrictions on their choice of literature. This means that the participants were able to research and use the literature independently for their reflection (both on the Internet and in the university library, but also from any other source). This gave the participants the opportunity to fill in any gaps in their knowledge individually. Before the data were collected, the participants had already written two reflections, on which they received criteria-led feedback from educational science scholars.

Based on the work of Schellenbach-Zell et al. (2023), the participants received supportive prompts on how to write the reflective essay. The prompts were intended to stimulate the participants to make use of professional knowledge from all three areas of professional knowledge (“Which theories and empirical findings can you refer to in order to come to a well-founded conclusion? […] You may use educational science as well as didactic or subject-specific sources”, cf. Supplement S1).

For data collection, the participants received a vignette in text form: a fictitious pre-service teacher in a practical phase—a novice practitioner—plans a fourth-grade science lesson, in which the learners are to understand why sea levels rise when the Earth’s ice masses melt due to global warming through an experiment. The lesson plan is presented in written, narrative form (cf. Supplement S2). The description contains three misconceptions that become evident in planning decisions, which contradict an evidence-based teaching. All aspects are grouped around one phase of the lesson, that is, carrying out an experiment in group work.

First, in the area of PPK, the lesson plan is based on the learning styles according to Vester (2021). The teacher divides the class into groups to ensure that each learning style is represented by a learner in each group, with the consequence that, for example, visual learners only watch the experiment and so on. However, there is no scientific evidence supporting the existence of learning styles that are not only preferred but also lead to measurably better learning outcomes (Kirschner, 2017). Bauer and Asberger (2022) describe in detail why learning styles are a myth and give reasons why it may even be detrimental to successful learning if learners are categorized in this way, e.g., because learners are pigeonholed and no longer receive individual support.

Second, in the area of CK, the lesson plan describes an experiment that is technically incorrect, as explained by Noerdlinger and Brower (2007) and Schlichting (2018): An ice cube melts in a glass of water and learners are supposed to see the water level rising. However, an ice cube has more volume than the same amount (in weight) of water due to its lower density. When it melts, its volume decreases and its density increases while its weight remains the same. In consequence, the same amount of water is displaced. In the planned experiment, the water level would therefore not rise. The hypothesis that melting icebergs contribute to the rise in sea levels is still valid. Freshwater, and thus freshwater icebergs, have a lower density than salt water (ocean). Therefore, the water level would still rise when a cube of freshwater ice melts in a glass of salt water. However, this effect is so minimal that the experiment could be classified as unsuitable for primary school lessons from a PCK point of view. This illustrates the importance of taking an integrated rather than isolated view of the different areas of knowledge into consideration. CK correctness has an impact on PCK implementation: A better option would be to place an ice cube on a stone. This setup would effectively and comprehensibly demonstrate how ice melts on land masses, which significantly increases ocean volume, and is far more appropriate for primary school children.

Third, from a PCK point of view, the experiment is not implemented adequately because important elements, such as working out a research question with the learners or providing scientific control in the experiment, are missing (Frischknecht-Tobler & Labudde, 2013; Hartinger, 2020).

The participants could address any aspects of the situation as they were only instructed to give feedback to the fictitious pre-service teacher. They were not told what to address or to address misconceptions at all.

2.2. Think-Aloud Protocols

Based on the research of Barnes et al. (2020), we used think-aloud protocols to analyze the reflection process. The participants recorded the think-aloud data without the supervision of the researchers. They were provided with detailed instructions (which were explained to the participants; questions could be asked), and then recorded the reflection process with their own devices. This allowed the participant teachers to take breaks and write the reflection at their preferred times and reduced the feeling of being “monitored”.

The complete think-aloud protocols (in the following abbreviated as TAs; for full instructions, cf. Supplements S3) covered an average of 1 h and 29 min of speaking time (SD = 01 h 12 min; Min = 25 min; Max = 04 h 05 min). They were transcribed in a content-related semantic way according to Kuckartz and Rädiker (2024). The protocols were then rated at the content level and coded at the process level.

3. Coding and Results

3.1. Rating and Results Content Level (Misconceptions)

First, we assessed the TAs in terms of their recognition and further handling of the misconceptions as a whole. For this purpose, the levels noticing, reasoning, conclusion, and alternative actions were rated separately for each content dimension (CK, PCK, and PPK). This resulted in a total of twelve individual ordinal-scaled ratings for each TA, as seen in

Table 1. Coding manual content elaboration.

Process	0	1	2	3
Noticing	No irritation is triggered/no focus on the concept	Irritation is triggered/focus on the concept is expressed
Reasoning	No (further) engagement with the concept	Engagement based on previous or everyday knowledge (subjective theories, experiences)	Engagement based on non-conflicting evidence	Engagement based on conflicting or multi-perspective evidence
Conclusion	Concept is accepted	Concept is accepted to a limited extent/partially confirmed	Concept is not accepted
Alternative actions	No alternative planning options	Alternative planning options without scientific foundation	Elaborate alternative planning options

. In the noticing dimension, we distinguished whether the misconception was addressed in any way (1) or not (0). As soon as the participants paid attention to the corresponding misconception or expressed irritation, it was rated (1). The reasoning category was scored on four levels: either the concept was not addressed further (0), or the concept was addressed based on everyday knowledge (1), or the concept was addressed using non-conflicting evidence (2), or the concept was addressed using conflicting or multi-perspective evidence (3). Non-conflicting evidence involves relying exclusively on sources that support the misconception or do not address it at all. Conflicting literature refutes or criticizes the concept. The conclusion dimension was scored at the following three levels: the concept is accepted (0), the concept is accepted to a limited extent (1), and the concept is not accepted. An example statement in which the learning styles concept has been partially adopted is as follows: “I’ve done a bit of research and actually found out that these learning styles have been expanded in more recent studies.” (TA7, seg. 170) or “Um I think that um different types of learners and cooperative learning arrangements go very well together.” (TA6, seg. 149).1 Finally, the alternative actions dimension was rated with no alternative planning options identified (0), alternative planning options identified without scientific basis (1), and elaborated alternative planning options identified (2).

Two raters coded the entire material. We achieved a Cohen kappa of $\kappa =0.85$ for the noticing dimension, $\kappa =0.96$ for the reasoning dimension, $\kappa =0.96$ for the conclusion dimension, and $\kappa =0.71$ for the noticing dimension. Differing ratings were discussed and agreed upon.

As shown in

Table 2. Noticing and processing of misconceptions.

	PPK				PCK				CK
Code	Noticing	Reasoning	Conclusion	Alternative actions	Noticing	Reasoning	Conclusion	Alternative actions	Noticing	Reasoning	Conclusion	Alternative actions
TA1	1	3	2	2	1	0	0	0	1	0	0	0
TA2	1	3	2	2	1	0	0	0	1	0	0	0
TA3	1	3	2	2	1	0	0	0	0	0	0	0
TA4	1	3	2	2	1	0	0	0	0	0	0	0
TA5	1	3	2	1	1	2	0	0	0	0	0	0
TA6	1	3	1	2	0	0	0	0	1	0	0	0
TA7	1	3	1	1	1	0	0	1	1	0	0	0
TA8	1	3	1	1	0	0	0	1	0	0	0	0
TA9	1	3	1	1	1	0	0	0	1	0	0	0
TA10	1	2	1	0	1	0	0	0	0	0	0	0
TA11	1	1	0	1	1	3	2	2	0	0	0	0

Note. The meaning of the values can be found in the coding scheme in Table 1. The principle is, the darker, the more elaborate.

, we can answer our first research question as follows: All participants noticed that grouping learners according to learning styles requires their attention. Moreover, all of them addressed the concept of learning styles in their reflective reasoning. While nine participants employed multi-perspective evidence to analyze the issue, one relied exclusively on non-conflicting evidence, such as the texts by Vester (2021), and another referred only to personal experiences without incorporating scientific evidence. The findings from the reasoning process are used in different ways to draw conclusions and develop alternative actions. Some participants describe the concept as empirically wrong (5), while others partially adopt it (5). One participant does not mention the concept at all in the conclusion. Some of the verbalized alternative actions are elaborated and result from reasoning (5), while others are suggested without reference to the evidence (5). At a descriptive level, an evidence-based conclusion also seems to be related to evidence-based alternative courses of action.

Even though almost all participants perceive the experiment as relevant in a subject-didactic manner (9 out of 11), there are only two participants who continue to deal with it in the further course of their reflection. One participant draws on non-conflicting evidence and does not formulate a conclusion or alternative actions. Another participant deals with the topic using multi-perspective evidence and formulates elaborate conclusions and alternative courses of action.

At the content knowledge level, five people become aware of the experiment, but none of them engage further with the topic.

3.2. Coding and Results Process Level (Epistemic Processes)

Second, we coded the TAs regarding the epistemic processes used by the participants with the software MAXQDA24, Version 24.7.0 (VERBI Software, 2024). At first, each TA was segmented by one researcher into coding units, specifically segments (in the following, seg.) (O’Connor & Joffe, 2020), and then all coding units were coded. To calculate the interrater agreement, we used kappa ${\kappa }_n$ according to Brennan and Prediger (1981) as the four-field table associated with our coding has an unequal marginal distribution, which could lead to paradoxical values with Cohen’s kappa (Kuckartz & Rädiker, 2019). In total, three protocols were coded by three researchers (fully consensual). The other eight think-aloud protocols were split: Team A coded two protocols $({\kappa }_n=0.63)$, and Team B coded six protocols (${\kappa }_n$ = 0.83). One researcher was in both teams and was thus able to inform the other two researchers about adjustments to the coding manual. The teams initially coded separately and then discussed contradicting results. Inconsistencies were used to adapt the coding manual. The results were “good/substantial” for Team A and “almost perfect” for Team B according to Landis and Koch (1977).

Based on the theoretical considerations, we established the process categories noticing processes, reasoning processes, and conclusion processes (Buehl & Fives, 2016; Weber et al., 2023). We further differentiated these categories partly deductively (based on Schellenbach-Zell et al., 2023; 2.4, 2.5, 3.1, 3.2, 3.3) and partly inductively (1.1, 1.2, 1.3, 1.4, 2.1, 2.2, 2.3), with codes for the different epistemic processes. We added the process category application of epistemic ideals inductively, as there were other processes occurring that did not fit in one of the first three categories (4.1, 4.2, 4.3, 4.4, 4.5). We found that the participants used a number of epistemic processes that we could not assign to the existing categories, but they all seemed to have one thing in common: they could serve to monitor whether or to what extent an epistemic aim may have already been achieved. They therefore seemed to represent the implementation of the epistemic ideals.

Moreover, we added a category with other processes to which we assigned clearly non-epistemic processes (5.1, 5.2). The final coding manual can be seen in

Table 3. Coding manual epistemic processes.

Process	Description	Example
1 Noticing processes
1.1 Reading the situation	Rereading and repetition of the planning description.	A prospective teacher in the practical semester plans a lesson on the topic ‘The sea level is rising—how so?’ in a fourth-grade class. (TA1, seg. 4)
1.2 Paraphrase and categorization of the situation description	Making sense of the description, realizing the meaning, and making first inferences.	Okay, so the hairdryer is supposed to symbolize solar radiation, I guess. (TA1, seg. 30)
1.3 Epistemic emotions	Verbalizing emotions that relate to knowledge and the generation of knowledge (surprise, curiosity, confusion). Verbalizing other emotions (non-epistemic emotions can also have epistemic consequences).	And, um, yes, I’m curious to see whether, um, yes, what I find out and whether the whole thing, as I said, makes sense and really leads to better learning outcomes. (TA2, seg. 43)
1.4 Identification/judging	Identify with or distance oneself from a planning decision, personal reference, no reference to knowledge bases.	But still a nice idea, I think. (TA1, seg. 37)
1.5 Problem identification/questioning	Expressing aspects that remain open in the description or making content-related comments that critically question the description.	Okay, I wonder how she knows how the children can learn best. (TA1, seg. 39)
2 Reasoning processes
2.1 Reference to knowledge of unclear origin	Recourse to (prior) knowledge of unclear origin; also experiences from own school days.	I actually thought that children can always learn best when they are active themselves? (TA1, seg. 48)
2.2 Reference to experiences from pedagogical practice	Recourse to experience from own pedagogical practice (planning, implementation, observation).	I’ve also had a lot of experience with seated circles and used them myself during the practical semester. Um, you can simply talk to the children better, I think, and you’re also at eye level with them. (TA1, seg. 71)
2.3 Reference to the curriculum	Reference to the curriculum, political education goals, “Perspektivrahmen Sachunterricht”.	Um, I don’t remember this topic being explicitly required by the curriculum. (TA7, seg. 85)
2.4 Reference to knowledge from academic studies and theory	Naming of scientific terms/theories, explicit reference to the studies.	Well, of course I’ve heard about learning styles before, but I also think it would have been in some educational science lecture with Mrs X or something like that, that it can’t be separated like that. (TA8, seg. 32)
2.5 Reference to ‘new evidence’	Recourse to “new” knowledge that is learnt or refreshed in the reflection process; knowledge is explicitly taken from the literature sources used for the reflection.	Okay here in the text it says metacognition is a collective term for a series of phenomena that have to do with knowledge and control over one’s own cognitive functions, for example learning memory et cetera. (TA1, seg. 200)
3 Conclusion processes
3.1 Knowledge construction/theory-informed interpretation	New knowledge gain, integration of knowledge from different sources and/or with prior knowledge. Interpreting the planning description with the help of evidence, establishing a connection between knowledge and situation.	And it is the case that several authors now classify not only Vester and Offner, um, according to different learning styles, but also, um, others. (TA1, seg. 175)
3.2 Self-reflection	Reflection on own knowledge, attitudes, and professionalization process.	Hm (thinking) yes, I definitely still have to deal with it, um, to what extent self-directed learning is realizable. Hm, I have to acquire knowledge about the implementation of self-directed learning. (TA1, seg. 281)
3.3 Naming of planning alternatives	Formulation of alternatives for the planning decisions.	In addition, I would have given the pupils a worksheet on which they should make individual notes. (TA4, seg. 251)
4 Application of epistemic ideals—processes
4.1 Naming and explaining work strategies	Describing how one proceeded with literature search and assessment of source quality; also verbalizing the intention to summarize information.	And, um, yes, I searched in various/different, um, databases, including, um, pedocs. Um, yes, of course also in the university library catalogue and in Fachportal-Pädagogik. I looked for the studies in particular. (TA1, seg. 100)
4.2 Justification of work strategies	Justifying one’s procedures.	…because I tried somehow. The lecturer told me that when I listed studies last time, yes, I should go into more detail about the structure of the studies and things like that. (TA8, seg. 79)
4.3 Meta-cognitive emotions/cognitive judgement	Emotions with regard to own work or sources; judgement of own work, conclusions without reference to theoretical arguments.	I have to say that, um, yes I spent a lot of time on it because it was very, very difficult for me to find something suitable, um, for the situation, especially, um, studies. TA1, seg. 87)
4.4 Evaluation	Retrospective evaluation of the work process.	Because I had written it quite well before, in my opinion. (TA6, seg. 147)
4.5 Expression of lack of knowledge/desire for new evidence	Verbalize missing knowledge to fully reflect on the situation, concretize what needs to be reworked, desire for evidence.	Okay, I don’t know Vester now, I’d have to read it again. (TA7, seg. 36)
5 Other processes (non-epistemic)
5.1 Coordination of the workflow	Description of the procedure for writing the reflection at an organizational level. The focus is on working on the task, time management, etc.	Okay, then I’ll deal with, um, the study now. Exactly, the last thing I stopped at was transferring the theory to the situation. (TA1, seg. 249)
5.2 Textualization	Description of the procedure for writing the reflection at an organizational level. The focus is on working on the task, time management, etc. The students write down their ideas. In doing so, they repeat what they have previously said or verbalize thoughts on word choice, grammar, etc.	I’ve already used ’distinction’, what can I use as a synonym? (TA1, seg. 158)

.

With this coding scheme, we can now address RQ2 and name specific epistemic processes that occur in participants’ reflections.

We found various epistemic processes and organized them into the categories noticing, reasoning, conclusion, and application of epistemic ideals, as well as non-epistemic processes (which, however, influence the application of epistemic processes, at least quantitatively).

Table 4. Descriptive statistics of the number of epistemic processes used per think-aloud protocol.

Process	Min	Max	M	SD
1 Noticing Processes	4	78	32.90	22.64
Reading the situation	1	22	9.73	7.32
Paraphrase and categorization of the situation description	0	7	3.36	2.01
Epistemic emotions	0	10	4.27	2.90
Identification/judging	2	18	8.27	4.80
Problem identification/questioning	1	21	7.27	5.61
2 Reasoning Processes	9	100	42.08	30.78
Reference to knowledge of unclear origin	1	22	10.91	7.70
Reference to experiences from pedagogical practice	0	14	4.36	4.59
Reference to the curriculum	0	7	1.36	2.25
Reference to knowledge from academic studies and theory	1	14	6.09	4.46
Reference to ‘new evidence’	7	43	20.36	11.78
3 Conclusion processes	8	49	30.72	14.28
Knowledge construction/theory-informed interpretation	6	27	19.18	7.41
Self-reflection	1	8	5.09	2.43
Naming of Planning Alternatives	1	14	6.45	4.44
4 Application of epistemic ideals—processes	12	76	36.19	20.25
Naming and explaining work strategies	9	40	19.55	9.17
Justification of work strategies	0	5	2.00	1.79
Meta-cognitive emotions/cognitive judgement	1	22	10.36	6.83
Evaluation	0	3	0.64	1.03
Expression of lack of knowledge/desire for new evidence	2	6	3.64	1.43
5 Other processes (non-epistemic)	11	169	51.91	49.80
Coordination of the workflow	11	64	35.27	18.06
Textualization	0	105	16.64	31.74

shows the descriptive statistics in absolute values (exact number of processes used per category), and

Table 5. Descriptive statistics of the distribution of epistemic processes in the think-aloud protocols.

Process	Min	Max	M	SD
1 Noticing processes	3.00%	43.00%	17.09%	11.69%
Reading the situation	1.00%	11.00%	4.64%	3.20%
Paraphrase and categorization of the situation description	0.00%	5.00%	2.00%	1.34%
Epistemic emotions	0.00%	7.00%	2.27%	2.00%
Identification/judging	1.00%	9.00%	4.45%	2.58%
Problem identification/questioning	1.00%	11.00%	3.73%	2.57%
2 Reasoning processes	6.00%	52.00%	22.17%	14.11%
Reference to knowledge of unclear origin	1.00%	13.00%	5.45%	3.64%
Reference to experiences from pedagogical practice	0.00%	8.00%	2.36%	2.62%
Reference to the curriculum	0.00%	4.00%	0.73%	1.27%
Reference to knowledge from academic studies and theory	1.00%	8.00%	3.36%	2.50%
Reference to ‘new evidence’	4.00%	19.00%	10.27%	4.08%
3 Conclusion processes	6.00%	37.00%	18.00%	10.60%
Knowledge construction/theory-informed interpretation	5.00%	23.00%	11.18%	6.31%
Self-reflection	0.00%	4.00%	1.18%	1.25%
Naming of planning alternatives	1.00%	10.00%	5.64%	3.04%
4 Application of epistemic ideals—processes	7.00%	37.00%	19.54%	10.16%
Naming and explaining work strategies	0.00%	2.00%	0.36%	0.67%
Justification of work strategies	1.00%	4.00%	2.18%	0.87%
Meta-cognitive emotions/cognitive judgement	4.00%	18.00%	10.73%	4.88%
Evaluation	1.00%	5.00%	2.82%	1.40%
Expression of lack of knowledge/desire for new evidence	1.00%	8.00%	3.45%	2.34%
5 Other processes (non-epistemic)	6.00%	57.00%	24.00%	15.91%
Coordination of the workflow	6.00%	27.00%	18.18%	6.71%
Textualization	0.00%	30.00%	5.82%	9.20%

shows the descriptive statistics in relative values (proportion of processes used per think-aloud protocol, calculated by dividing the number of processes used per category by the overall number of processes coded per TA). The utilization of the processes appears to be relatively even across the categories.

3.3. Qualitative Comparison of Three Cases

In the following chapter, we contrast three TAs in order to answer RQ3 and exemplify our results. Our aim is to work out connections between the content and the process level of reflection. For this purpose, we selected three cases that show strong contrasts in terms of the content level: The first case, TA1, elaborates on the learning styles concept (PPK) at all levels (noticing, reasoning, conclusion, alternative actions). The experiment is noticed, but neither taken up further from a PCK point of view nor from a CK point of view. The second case, TA10, elaborates less on the learning styles concept. After noticing, the participant uses only one-sided evidence for reasoning and partially adopts the concept of learning styles in the conclusion. The participant does not verbalize alternative actions concerning learning styles. The experiment is noticed from a PCK point of view, but not from a CK point of view. There is no further elaboration. The third case, TA11, differs, because it barely reflects on the learning styles after noticing, but elaborates on the experiment as a teaching method (PCK). The participant elaborates at all four levels, noticing, reasoning, conclusion, and alternative actions, at an advanced level and recognizes and processes misconceptions in the planned implementation of the experiment.

Regarding the quantitative use of epistemic processes, the TAs do not differ considerably (see

Table 6. Absolute and relative process utilization in the contrasted TAs.

Total Number of Processes	TA1	TA10	TA11
	281	208	153
1 Noticing processes	51 (18.41%)	33 (16.10%)	22 (14.86%)
Reading the situation	21 (7.58%)	8 (3.90%)	2 (1.35%)
Paraphrase and categorization of the situation description	7 (2.53%)	4 (1.95%)	4 (2.70%)
Epistemic emotions	4 (1.44%)	3 (1.46%)	5 (3.38%)
Identification/judging	10 (3.61%)	9 (4.39%)	5 (3.38%)
Problem identification/questioning	9 (3.25%)	9 (4.39%)	6 (4.05%)
2 Reasoning processes	65 (23.47%)	44 (21.46%)	46 (31.08%)
Reference to knowledge of unclear origin	22 (7.94%)	12 (5.85%)	6 (4.05%)
Reference to experiences from pedagogical practice	4 (1.44%)	5 (2.44%)	12 (8.11%)
Reference to the curriculum	0 (0.00%)	2 (0.98%)	1 (0.68%)
Reference to knowledge from academic studies and theory	2 (0.72%)	6 (2.93%)	12 (8.11%)
Reference to ‘new evidence’	37 (13.36%)	19 (9.27%)	15 (10.14%)
3 Conclusion processes	43 (15.52%)	33 (16.10%)	33 (22.30%)
Knowledge construction/theory-informed interpretation	24 (8.66%)	26 (12.68%)	23 (15.54%)
Self-reflection	8 (2.89%)	5 (2.44%)	7 (4.73%)
Naming of planning alternatives	11 (3.97%)	2 (0.98%)	3 (2.03%)
4 Application of epistemic ideals—processes	33 (11.91%)	67 (32.68%)	27 (18.24%)
Naming and explaining work strategies	15 (5.42%)	40 (19.51%)	10 (6.76%)
Justification of work strategies	0 (0.00%)	3 (1.46%)	1 (0.68%)
Meta-cognitive emotions/cognitive judgement	13 (4.69%)	21 (10.24%)	13 (8.78%)
Evaluation	0 (0.00%)	0 (0.00%)	0 (0.00%)
Expression of lack of knowledge/desire for new evidence	5 (1.81%)	3 (1.46%)	3 (2.03%)
5 Other processes (non-epistemic)	89 (32.13%)	31 (15.12%)	25 (16.89%)
Coordination of the workflow	52 (18.77%)	30 (14.63%)	25 (16.89%)
Textualization	37 (13.36%)	1 (0.49%)	0 (0.00%)

). However, we would like to point out some aspects as an example that stand out when considering the relative proportions of processes utilized: TA11 uses considerably more reasoning processes than TA1 and TA10. TA11 refers more to experiences from pedagogical practice and knowledge from studies than the others. It is noticeable that TA10 names and explains working strategies more than the others. Moreover, TA1 uses a strikingly large proportion of non-epistemic processes to textualize thoughts.

However, a qualitative comparison of exemplary epistemic processes reveals various differences.

3.3.1. Noticing Processes

With regard to noticing processes, the following characteristics appear:

Paraphrase and categorization of the situation description: TA1 reconstructs the experiment at all three levels (CK, PCK, PPK). No misconceptions are recognized yet, but an open attempt is made to understand the lesson plan. TA10, on the other hand, assigns the description of the situation directly to the concept of exploratory learning. This could indicate a more open perception in TA1 compared to a narrower, focused perception in TA10.

Problem identification/questioning: It is noticeable that TA1 and TA11 tend to formulate questions, such as “Okay now I wonder how the child is supposed to learn by just listening?” (TA1, seg. 55), and often ask questions with question words (why, wherefore, etc.). TA10, on the other hand, tends to make statements: “The, um, division according to learning styles seems to be problematic in the, um, description of the situation.” (TA10, seg. 188). This difference may indicate a different epistemic attitude (aims and/or ideals and values) that influences the entire reflection.

3.3.2. Reasoning Processes

The use of epistemic processes for reasoning appears to be different, especially with regard to the selection of content and sources:

Reference to knowledge of unclear origin: TA1 expresses thoughts with direct reference to learning styles (“I actually thought that children can always learn best when they become active themselves?”; TA1, seg. 48). TA10 expresses various ideas about references that can be made. However, it is noticeable that these do not deal with the learning styles, but rather focus on other theories, such as exploratory or discovery learning, interest, group work in general, and discussions with the mentor (“So it has to be a well-chosen video and of course afterwards, um, I think the children’s interest should remain constant”; TA10, seg. 29). The third participant, TA11, sets a different focus: five out of six coded segments refer to experimentation (“there is a lot of potential in independent experimentation”; TA11, seg. 80). Different priorities in terms of content may therefore be set here. The content foci are also reflected in the use of the other epistemic processes related to knowledge.

Reference to experiences from pedagogical practice: When the participants make references to their own pedagogical experiences, it is noticeable that TA1 directly compares the situation with their own experiences (“No, I haven’t experienced it yet. I haven’t experienced pupils being divided into, what were they called, auditory, visual, haptic and, what was the other one? Intellectual learning styles were categorized. Hm. Just support groups or something. Um, but I don’t think that fits in there now. Um, yes, exactly.”; TA1, seg. 134). TA11 also compares the lesson plan directly with his own experiences and thus uncovers deviations. All considerations are aimed at the implementation of the method of experimentation (“Yes. Protocol [and] assumption is always there with us. And we always experiment together, there are no observers only—except for the demonstration experiment.”; TA11, seg. 72). TA10, however, makes comparisons at a more superficial level: “Well, um, I’ve had a similar situation before, because I also had to prepare a lesson and after the lesson I talked to my mentor about the procedure and everything.” (TA10, seg. 108). When the participant relates the experiences to the topic of group work, considerations remain on the surface of the “group work” method and do not aim at deeper structures. This different application of epistemic processes may indicate that TA1 and TA11 are more successful in critically relating and analyzing existing pedagogical experiences.

Reference to knowledge from academic studies and theory: The references made to university knowledge initially appear similar for all three think-aloud protocols. TA1 refers to intrinsic motivation and cooperative learning. TA10 refers to exploratory learning, research-based learning, and interest. TA11 refers to experimentation and exploratory learning. There are some overlaps in the selected topics, but individual priorities are set. The participants therefore appear to consider different knowledge from their university studies to be relevant for the situation; the ability to select the appropriate bodies of knowledge from existing prior knowledge could play an important role in the ability to reflect.

Reference to ‘new’ evidence: TA1 draws on sources on cooperative learning, self-directed learning, meta-cognition, and learning styles. The learning styles are described on the basis of two sources that support the learning styles hypothesis. These statements are then dissolved again (“These propagated learning styles are nowadays (…) strongly criticized as/or for the time being, because they do not exist. So, it is assumed that there are not different types of learners, but that it is, um, a consequence of habitual preferences of modes of information processing.”; TA1, seg. 185–186). The concept is distinguished from learning strategies: “However, it has been proven that learners have different learning strategies or metacognitive strategies, which then, um, also influence learning. Exactly, which have an influence on learning behavior.” (TA1, seg. 198). Learning strategies are assigned to the concept of self-regulated learning and referred to cooperative learning. It is striking that the different concepts are clearly related to each other in TA1. In contrast, there seems to be only a mere reproduction of various theories (interest theory and exploratory learning) in TA10; no connections are made. This might indicate a fragmented acquisition of knowledge in TA10 (Kleickmann & Hardy, 2019).

3.3.3. Application of Epistemic Ideals—Processes

Meta-cognitive emotions/cognitive judgement: Regarding the epistemic processes that represent an application of epistemic ideals, both TA1 and TA10 verbalize that they had difficulties finding suitable literature. It is striking, however, that TA1 also verbalizes that they spent a correspondingly long time on this (“I have to say that I spent a long time on this because it was very, very difficult for me to find something suitable, um, for the situation, especially, um, studies.”; TA1, seg. 87). The time spent and appropriate attitude might be a reason for the way the epistemic process is applied.

In TA11, the participant focuses on the reflection prompt “At the level of your professional role and/or your professional self: What do you take away for yourself?” (reflection prompt) and says, “Oh, I don’t like this task, it’s the same thing I just wrote” (TA11, seg. 151). This could be relevant with regard to the quality of the content of this part of the task.

3.3.4. Conclusion Processes

Knowledge construction/theory-informed interpretation: It is noticeable that TA1 regularly summarizes the knowledge components when using knowledge-constructing processes and relates them to each other, as can be seen here: “And it is even the case that metacognitive skills, um, are also part of self-directed learning. Perhaps I can use that as a connecting factor.” (TA1, seg. 239). The knowledge is also often used to interpret the situation: “So this form of learning is definitely not really cooperative.” (TA1, seg. 62). TA11 also relates the theory to the given situation, as can be seen in the following two quotations: “But is it really action-oriented? They act, but they only do it according to a plan” (TA11, seg. 113) and “The more I read the theory and write down the explanations, the more pointless this experiment somehow becomes” (TA11, seg. 124). TA10 remains less specific in the construction of knowledge: “Yes, so overall you can say that theoretically you can’t apply these four learning styles to everyone” (TA10, seg. 137). The statements are formulated in more general terms and not directly related to the situation. Once again, this could indicate a rather fragmented acquisition of knowledge.

Self-reflection: All three TAs differ in the processes of self-reflection. TA1 rejects the learning styles concept and proposes self-directed learning as an alternative. It is striking that the participant reflects on the conditions for the success of self-directed learning and also considers, for example, teaching disruptions or the role of the teacher as important factors. It could be concluded that TA1 does not want to transfer the theories to everyday school life in a recipe-like manner, but wants to deal with them in greater depth. The self-reflective processes of TA10 remain less concrete: “and for that reason, um, I might just have informed myself a bit more.” (TA10, seg. 217). TA11, on the other hand, formulates his personal connection to the topic: “I love experimenting.” (TA11, seg. 58); he also draws personal conclusions on the topic of experimenting. One’s own emotions also seem to play a role in the application of epistemic processes.

Naming of planning alternatives: TA1 formulates alternative actions with concrete references to cooperative and self-directed learning. TA10 also makes references to theories, such as discovery learning. The difference is that TA1 uses these references to formulate action alternatives with regard to group division and task distribution in the groups (instead of using learning styles). TA10 does not refer to learning styles when verbalizing possible alternative actions. TA10 does not seem to transfer the findings on the misconception of learning styles into practice. TA11 mainly formulates alternative courses of action on the topic of experimentation.

3.3.5. Overall Use of Processes

In summary, each of the three TAs looks at one of the three misconceptions; it seems too difficult or too complex to look at and reflect on two or more misconceptions at the same time. Because the participants only ever focus on one misconception, they do not consider the overall picture of lesson planning with all three misconceptions. This means that they themselves could partially adopt misconceptions through reflection.

TA1 and TA11 are similar in the quality of their use of the processes: they formulate questions about the situation, and both apply prior knowledge to the situation in a concrete way. TA10, on the other hand, formulates narrower statements, and the references that are made to prior knowledge are not completely relevant to the given situation but go beyond. TA1 and TA11 make direct references to previous pedagogical experience, while TA10 makes references at a meta-level, such as giving feedback to the situation on lesson planning. No content-related links to previous experience are established. The references to university knowledge are similar for all three TAs. The references to ‘new’ evidence show differences: TA1 presents several theories and relates them to each other. TA10 only describes the theories and does not relate them to each other. In the area of knowledge construction, it is also evident that TA1 and TA11 relate the knowledge to each other and to the situation, while TA10 remains less specific. This is reflected in the alternative courses of action: TA1 and TA11 use the knowledge bases when formulating alternative courses of action. The alternatives of TA10 refer to knowledge, but not to the problem of learning styles. The link is not evident here. With regard to self-references, the following is striking: TA1 and TA10 differ in their perception of the task as difficult to deal with, where TA1 reports having dealt with it for a longer period of time, while TA10 does not. TA11’s enthusiasm for the topic of experimentation and the large amount of prior knowledge and experience are striking.

Thus, there seems to be a link between the quality of use or the fit of the use of epistemic processes and the recognition and resolution of misconceptions. The quantity of the processes used does not seem to play a decisive role. On the contrary, it seems that the use of the processes is influenced by the aims and beliefs that are indirectly permeating.

4. Discussion

Reflection means systematic thinking about a pedagogical (vicarious) experience with the inclusion of scientific evidence—in order to better understand the experience and to draw conclusions for one’s own professional development. It can be especially useful for supporting pre-service teachers in practical phases in processing anecdotal evidence adequately and, for example, in not adopting observed misconceptions unquestioningly. Research has paid little attention so far as to how exactly reflection takes place when misconceptions are recognized and processed. Based on the AIR model of epistemic cognition (Chinn et al., 2014), this article exploratively addresses the question of whether pre-service teachers recognize and reflect on misconceptions in anecdotal evidence and which epistemic processes occur. Finally, we investigate, at a qualitative level, whether connections between epistemic processes and the recognition of misconceptions become visible. To this end, we analyzed eleven think-aloud protocols of pre-service teachers in a practical phase, who wrote a knowledge-based reflection on a fictitious lesson plan. At this point, we would like to emphasize once again that our research is exploratory in nature, which results in some serious limitations. The given study works with a small, specific sample and one specific vignette, which in turn causes a lack of generalizability and transferability. The research results should be replicated with larger and differentiated samples and also with various vignettes.

With regard to RQ1, the pre-service teachers’ handling of the misconceptions at the content level differs for each misconception (PPK, PCK, CK) as well as for the reflection components noticing, reasoning, conclusion, and alternative actions: All pre-service teachers noticed the educational science misconception of learning styles according to Vester (2021). Two participants did not notice a misconception at the PCK level, and six participants did not notice a misconception at the CK level. While all participants reflected on learning styles, only a few continued to deal with the experiment as a teaching method (PCK) and none with the scientific correctness of the experiment (CK). There could be various reasons for this.

First, the pre-service teachers may be exposed to some phenomena more often than others in the course of their university studies and, in consequence, may have more prior knowledge in the respective areas. It remains to be clarified in further research whether this is related to the prior knowledge of the participants or to the reflection task itself. This could be investigated more closely by surveying subject-specific prior knowledge as a control variable. The assumption is likely, for we know that a profound professional knowledge base of teachers also has a positive influence on the quality of teaching (e.g., for educational knowledge analyzed; Voss et al., 2015).

Second, in the given study, the educational–scientific misconception could be easier to recognize than the other misconceptions, for it is not only increasingly discussed in research and teaching, but it also seems to be easily understood as to why the concept might be flawed. The experiment, on the other hand, is often used as a method in everyday life and even laypersons have an idea of what an experiment is. A supposed technical correctness of the experiment seems so plausible that it is not even to be questioned. How close or distant it is to everyday life could play a role in whether a misconception is perceived or not.

Third, the pre-service teachers may have been trained more in PPK and PCK than CK aspects through their previous university studies. Another problem could be that the CK misconception (scientific correctness of the experiment) also influences the implementation at the PCK level (a scientifically correct implementation requires didactically different approaches, such as placing the ice cubes on a stone). This interweaving of misconceptions could have led to biases in our study and should be taken into account in future studies. At the same time, it is very close to actual teaching practice and prospective teachers need the ability to use the knowledge components in an integrated way (Lehmann, 2024). As a consequence, our teacher training programs would have to do a better job at providing coherence in order for pre-service teachers to be able to relate knowledge and working strategies from one area of knowledge to another.

Fourth, the entire possible cognitive capacity of most of the participants might be used to analyze the learning styles concept. It might be too complex for pre-service teachers at the level of their master’s studies to recognize misconceptions at several levels of lesson planning. The ability to become aware of inconsistencies and potential misconceptions should therefore be fostered through university teaching while gradually increasing the level of complexity.

Fifth, the fact that the feedback for the two previous reflections was provided by a lecturer in educational sciences (PPK) and the data collection took place within an educational science course could also have played a role. This may have influenced the reflection and the selection of knowledge. Nevertheless, to mitigate this effect, the participants were prompted to consider all three knowledge areas (CK, PCK, and PPK) in the reflection task.

Apart from this, it is striking that a heterogeneous picture emerges at the level of conclusion (epistemic stance; Buehl & Fives, 2016) and alternative actions (anticipated epistemically informed practice; Buehl & Fives, 2016) in the area of PPK. Five pre-service teachers correctly categorized the learning styles as myths. However, the conclusions drawn by four other pre-service teachers, who also reasoned with multi-perspective evidence, partially accept the concept of learning styles and consider incorporating it into their future action repertoire. There seems to be a gap between the analysis part and the self-referential part of the reflection (as differentiated by Lohse-Bossenz et al., 2023). The self-references might not result directly from the reasoning, which would be consistent with the findings of Lohse-Bossenz et al. (2023), who identified differences between a pure lesson analysis and a reflection with self-references (in regard to self-efficacy). The qualitative comparison of the use of epistemic processes of the three TAs could provide more information.

Regarding RQ2, we were able to describe and inductively differentiate epistemic processes in the reflection categories noticing, reasoning, and conclusion. This differentiation of epistemic processes may help to understand how the reflection of pre-service teachers takes place when dealing with misconceptions. We can now better understand the individual parts of reflection, e.g., we recognized that noticing can be subdivided into different epistemic processes. For example, the distinction between the processes “identification/judging” and “problem identification/questioning” may be insightful, as it depicts whether or how much the pre-service teachers rely on their gut feeling when noticing (identification/judging; possibly implicit knowledge) or refer to knowledge or name discrepancies in a more neutral and concrete way (problem identification/questioning). Emotional attachment/motivation and previous experiences with a topic also seem to play a role and may be further investigated in the context of epistemic cognition. Our research supports the findings of Weber et al. (2023) that enjoyment has a positive effect on reflection.

We also identified processes that reflect the direct application of epistemic ideals. This represents a new operationalization of the AIR model. In previous studies, epistemic aims and values, epistemic processes, and epistemic ideals were mostly investigated separately. However, our qualitative data show an interweaving of epistemic processes and ideals, or a direct manifestation of ideals (and presumably also aims) in the processes.

Which processes are visible could therefore provide information about which aims and ideals the pre-service teachers hold (and the results of RQ3 could provide initial insights into this). At the same time, this could mean that university teaching not only has to convey to pre-service teachers which processes exist and how they can be used reliably—as requested in the apt-AIR model (Mor-Hagani & Barzilai, 2022)—but that teaching has to start off at the level of aims and ideals. That is why we consider these processes to be important variables that should be investigated further in future research. Against this background in particular, the attitude of the pre-service teachers towards the use of scientific knowledge (for us, epistemic ideals) could also play a role (e.g., Merk et al., 2017). According to Zeeb and Voss (2024), this may influence epistemic aims and thus epistemic processes. In the future, it could be worthwhile to examine how utility value interventions can meaningfully support the use of epistemic processes and, above all, their quality.

However, within the design of this study, pre-service teachers were able to record the TA on their own. This way, we were able to record as realistic a process of reflection as possible, in which participants were free to choose the setting. However, the participants could not be accompanied and thus controlled in the recording process. As a result, the material is heterogeneous and not all thoughts might be continuously verbalized. It is therefore always possible that some thoughts were not recorded by the participants, for example, when they are implicit (e.g., routine decisions) or when the participants lack the ability to verbalize (Konrad, 2019). Consequently, this means that we only recorded epistemic processes that the participants consciously used and were able to verbalize.

The answer to RQ3 shows that it is not only decisive which processes are used, but also how they are used—in other words, their quality is important. We selected three TAs that differed strongly in their handling of the misunderstandings (as analyzed in RQ1) and compared them in terms of their process use (as analyzed in RQ2). It can be seen that the two TAs which each resolved a misconception in an elaborate way (TA1 and TA11) used the epistemic processes differently than TA10, which shows an awareness of the learning styles myth but no elaboration on the latter. The processes are used to closely and concretely link and compare the content of the given lesson plan with different bodies of knowledge.

When it comes to capturing the quality of epistemic processes, some aspects remain open. We have only assessed this indirectly via the content-related benefits (recognizing and processing misconceptions). Schellenbach-Zell et al. (2023) and Hartmann et al. (2021) use the reflexivity of application for this purpose and also refer to the concept of meta-reflexivity (Cramer et al., 2023). According to the concept, the ability to explicitly relate bodies of knowledge to one another is an important component of professionalization. Similar demands can also be found in the debate on the coherence of teacher training (e.g., Canrinus et al., 2019; Hellmann et al., 2021). According to our study, it seems to be helpful to explicitly relate different knowledge bases to each other for content-related processing of misconceptions, which corresponds to the concept of meta-reflexivity. Nevertheless, questions remain about how knowledge is interconnected: how do pre-service teachers relate different elements of knowledge to one another, and does it pose problems if they combine these elements in ways that blur distinctions or apply theories inconsistently?

In addition to our analyses, it is striking that some pre-service teachers reflect on the situation with the help of multi-perspective scientific evidence. They relate these well to each other and to the situation, draw conclusions for their own professional development, and develop elaborate alternative courses of action for the situation. In essence, they do a good job of knowledge-based reflection, but still do not recognize misconceptions (that are detrimental to the quality of teaching) in anecdotal evidence. It therefore seems reasonable to train noticing skills in university teacher education—especially with a focus on critically reflecting on anecdotal evidence and independently drawing on scientific evidence to identify potential misconceptions.