1 Introduction

The advancement of digitalisation has posed new and increasing challenges for the teaching profession. Students must be trained in the so-called ‘four C's’ (critical thinking, communication, cooperation, and creativity) as overarching competencies in addition to mastering digital literacy skills in preparation for life in an increasingly digital society (SWK, 2022), as summarised, for example, in the ‘Key Competences for Lifelong Learning’ (European Commission, 2019). Further frameworks emphasize, that teachers must acquire these competencies themselves to be able to teach them to their students, e.g., Redecker and Punie (2017), UNICEF’s ‘Educators’ Digital Competence Framework’ (2022), UNESCO ICT Competency Framework for Teachers (2011). However, despite teachers’ decisive role in students’ success with respect to digital competencies (European Commission, 2019), this aspect has to date received insufficient attention in teacher education. This is corroborated by the findings of a survey conducted by the Organisation for Economic Co-operation and Development (OECD), according to which 39% of teachers in the European Union (EU) admitted that they felt inadequately prepared for the everyday use of digital technologies in the classroom (European Commission, 2020a), compared to an OECD average of 43% (OECD, 2019). Significantly, teachers’ professional knowledge regarding information and communication technology (ICT) integration in the classroom is not only valuable in extreme situations, such as during the recent COVID-19 pandemic, e.g., to enable flexibility in action (König et al., 2020); rather, a professional knowledge base of this nature will also help teachers successfully implement innovations in the classroom (Mwendwa, 2017; König et al., 2022a).

Various initiatives at the national and international levels have established the goal of better preparing teachers for such challenges and providing them with more effective support. In Germany, the Federal Ministry of Education and Research (BMBF) has addressed the problem with various projects affiliated with the ‘Qualitätsoffensive Lehrerbildung’ ([Quality Initiative for Initial Teacher Education] BMBF, 2018). With the ‘Digital Education Action Plan’ (2021–2027), the EU is focusing explicitly on improving teachers’ digital competencies and skills, which is referred to as ‘Priority 2’ in the corresponding paper (European Commission, 2020b). In addition, in 2021, the UNESCO Institute for Information Technologies in Education (UNESCO IITE), Southeast Asian Ministers of Education Organization (SEAMEO), and the Teacher Task Force (TTF) led the first virtual ‘High-Level Forum on Teacher Competencies in the Digital Revolution: Reaching the Unreached’. One output of this forum was an ‘Action Agenda on Improving Teacher Competencies in the Digital Revolution’ (UNESCO IITE et al., 2021).

To address these requirements efficiently and effectively, evidence-based insights into preservice teachers’ professional knowledge as a learning outcome of initial teacher education and the opportunities to learn (OTL) to which they are exposed during their university education are required (Flores, 2020). A frequently cited approach in this regard is Mishra and Koehler’s (2006) technological pedagogical content knowledge (TPACK) model, according to which teachers’ professional knowledge comprises general pedagogical knowledge (GPK), technological knowledge (TK), content knowledge (CK) and the resulting intersections. A cross-subject perspective on teachers’ professional digital knowledge highlights the following knowledge components, which form the TPACK model – often illustrated by the much-reproduced Venn diagram (Mishra & Koehler, 2006, p. 1025): pedagogical knowledge (PK or GPK) on one side, TK on the other, and technological pedagogical knowledge (TPK) at their intersection (Fig. 1).

Fig. 1
figure 1

The TPACK model

Full size image

Recent years have witnessed much educational research dedicated to teachers' professional knowledge. For example, several studies have investigated GPK in German contexts (cf. König et al., 2018; Voss et al., 2015) as well as in other countries worldwide (König, 2014; Leijen et al., 2022). Several such studies also described and analysed the OTL available to student teachers (cf., König et al., 2017; Watson et al., 2018; Terhart, 2019; Depping et al., 2021). Perhaps because the amount of technological OTL offered in teacher education is still very scarce (e.g., Gudmundsdottir and Hatlevik, 2018), TPK and TK have received considerably less attention. For example, Lachner et al. (2019) developed a knowledge test to assess TPK, which has already been implemented in several studies (e.g., König et al., 2020). However, the findings as to whether or to what extent the individual knowledge components can be recorded separately or how OTL may affect teachers' professional knowledge have primarily been obtained from self-reports (cf. Scherer et al., 2017). Gerhard et al. (2023) is an exception in this regard: they applied knowledge tests for GPK and TPK and investigated the influences of the associated learning opportunities. As such, their study constitutes the first step towards countering Lachner et al.’s (2019) assertion that ‘the interplay of these t-dimensions (e.g., TPK) and their basic knowledge components (e.g., PK, TK) are less clear’ (p. 9). However, knowledge in these areas is fundamental to digital literacy for educators (UNICEF, 2022). As recent developments in the field of artificial intelligence, exemplified by ChatGPT, have demonstrated, the education sector faces ever-evolving challenges in light of technological innovations (Zhu et al., 2023). Particularly, TK plays a crucial role in lifelong learning. The persistent evolution complicates the development of instruments that capture the requisite knowledge in its current necessary form (Mishra & Koehler, 2006). To the best of our knowledge, no knowledge test designed to measure TK and its associated OTL has been developed to date. Consequently, scientific knowledge regarding the interaction of GPK, TPK, and TK as well as how OTL may influence teachers’ professional digital knowledge remains limited.

The present article addresses the corresponding research lacuna. We used data collected in 2022 as part of a central teacher education quality assurance initiative (‘Bildungsmonitoring’) implemented by the University of Cologne, which surveyed bachelor’s and master’s student teachers. The survey is part of a larger project entitled ‘Zukunftsstrategie Lehrer*innenbildung Köln (ZuS): Heterogenität und Inklusion gestalten‘. ZuS has been implemented in the ‘Qualitätsoffensive Lehrerbildung’ (Quality Initiative for Initial Teacher Education), a joint initiative of the Federal Government and the Länder that aims to improve the quality of teacher training. The programme is funded by the Federal Ministry of Education and Research [grant numbers 01JA1515, 01JA1815, and 01JA2003 (DiSK)]. In our study, we used three different test instruments to measure GPK, TPK, and TK. Preliminary analyses that focused exclusively on a bachelor’s student sample and the simpler analysis of the bilateral relationship between GPK and TPK have already been published (Gerhard et al., 2023). In the present article, we offer new insights into the GPK–TPK–TK triad using a larger database of both bachelor’s and master’s student teachers. We thus build on the above-mentioned preliminary analyses (Gerhard et al., 2023) while extending the previously obtained insights by accounting for TK as another important component of teachers’ knowledge for ICT integration in the classroom.

We choose these three knowledge components to provide an overview irrespective of specific teaching subjects, since indications of lacking technological OTL have been observed even at the generic level (Gerhard et al., 2023). Since digitalisation advances across all areas of society and is not limited to certain subject or content areas, schools and universities are required to take measures that will be applicable to teachers across all subjects and school types. Hence, associated knowledge is therefore of a fundamental nature and frequently independent of subject-specific requirements at first. Research indicates that the mere use of digital resources has not yet improved the teaching–learning process. Rather, the way in which they are used is crucial (Baker et al., 2018; Lei & Zhao, 2007) in terms of developing (preservice) teachers’ competencies beyond their subject boundaries (Blikstad-Balas, 2023). As such, this paper focuses on the generic components of (preservice) teachers' professional digital knowledge, i.e., GPK, TPK, and TK.

Feasibility of conducting our research was another reason: To capture technological content knowledge (TCK), various subject-specific knowledge tests would need to be developed for the different teaching subjects and their academic disciplines. This was not feasible in our research project, and analysing the professional knowledge with TCK of various teaching subjects would inhibit research questions going far beyond the scope of the present article.

Moreover, comparing bachelor’s and master’s students regarding their knowledge allows for a more detailed examination of the knowledge acquisition within the three knowledge components. This analysis can yield insights into formal and informal OTL in these domains. With regard to the TPACK model, we anticipate achieving systemic comparability, as these students originate from the same institution before the divergence occurs during teacher training, and practical experience increase for in-service teachers.

2 State of research

2.1 Teachers’ professional digital knowledge

The structure of preservice and in-service teachers’ professional knowledge has changed in recent decades for reasons that include the new challenges that have arisen in the context of digitalisation (Koehler & Mishra, 2009; Mußmann et al., 2021). As early as 2006, Mishra and Koehler acknowledged these new challenges and added a technology component to the knowledge model developed by Shulman (1987). The resulting TPACK model includes a total of seven knowledge components (Mishra & Koehler, 2006): PK/GPK, CK, TPK, TK, technological content knowledge (TCK), pedagogical content knowledge (PCK), and TPACK.

We focus on GPK, TPK and TK because these are important competencies across subjects (see Sect. 3). The current status of research on this topic is outlined below.

2.1.1 General pedagogical knowledge (GPK)

Shulman (1987) identified GPK as one category of the knowledge base for teaching ‘with special reference to those general principles and strategies of classroom management and organisation which seem to transcend the subject area’ (p. 8). Various authors have demonstrated that GPK can be conceptually distinguished from other knowledge components, particularly those that pertain to specific subjects (Shulman, 1987; Baumert et al., 2010; König et al., 2022b). In a systematic literature review, König (2014) delineated three content areas of GPK: instructional process, student learning, and assessment. The area of student learning explicitly illustrates GPK’s relevance for learners. Researchers broadly agree that GPK is a key determinant of teaching quality that affects students’ learning growth and motivational development (König et al., 2021; Leijen et al., 2022).

2.1.2 Technological pedagogical knowledge (TPK)

The Venn diagram that visualises TPACK (Mishra & Koehler, 2006, p. 1025) illustrates TPK as resulting from the superposition of TK and GPK (Fig. 1). Shulman proposed the amalgam hypothesis, which suggested that PCK represents the correlation of the knowledge components PK and CK (Shulman, 1987). This hypothetical amalgam of PK and CK was later proven by Tröbst et al. (2018), among others. To the best of our knowledge, the extension of the amalgam hypothesis to the technological domain, i.e., PK, TK, and TPACK, has hitherto only been possible on the basis of self-reports. Luo et al. (2022), for example, have investigated the relationships between PK, TK, and TPK and their influence on TPACK. To demonstrate technology integration, they resort to items on self-efficacy, among others, and their results reveal significant correlations between the three knowledge components, i.e., PK, TK, and TPK (Luo et al., 2022). Existing definitions of TPK – such as that devised by Zhang et al. (2019), according to which TPK comprises ‘knowledge about the use of new technologies to support general pedagogical activities’ (p. 3443) – point in the same direction. This definition also clarifies TPK’s distinction from TCK and TPACK, as TPK encompasses knowledge that is relevant to all subjects and is thus subject-independent (Guggemos & Seufert, 2021). However, this is not the sole reason that TPK is regarded as crucial for all teachers (Gerhard et al., 2023; Guggemos & Seufert, 2021): TPK is also considered an important requirement for teachers’ ICT integration in the classroom (Gerhard et al., 2023; Lachner et al., 2019; Koehler & Mishra, 2009). Koh et al. (2013), for example, demonstrated that TPK – like TCK – not only exerts a positive influence on TPACK but that this influence is greater than that of TK and GPK. To capture the TPACK model’s seven components, they used a survey developed by Chai et al. (2011). Koh et al.’s adapted survey comprised 30 items rated on a seven-point Likert scale measuring the degree of agreement or disagreement with statements about the seven different TPACK components.

2.1.3 Technological knowledge (TK)

Alongside GPK and CK, TK is one of the three basic components of the TPACK model (Mishra & Koehler, 2006) and ‘refers to [(preservice)] teachers’ knowledge and proficiency with technology tools’ (Shinas et al., 2013, p. 341). In this context, Mishra and Koehler (2006) expressed criticism regarding the consideration of TK in isolation and instead advocate the integrated promotion of all technology components in professional and practical contexts. Guggemos and Seufert (2021) support this position, but Kaplon-Schilis and Lyublinskaya (2020) reported different findings. Using knowledge tests rather than self-reports in their study, they examined the basic components of the TPACK model with a subject focus on mathematics and science and tested their influence on TPACK without considering the intersections, i.e., TPK, TCK, and PCK. Although they used knowledge tests to capture TK, PK, and CK, TPACK was assessed using the TPACK Levels Rubric: ‘The rubric was used to assess the teachers’ written artifacts (lesson plans and authored curriculum materials) and observed behaviours (PD presentations and classroom teaching through observations)’ (Lyublinskaya & Tournaki, 2012). Their findings reveal that their TPACK measure is independent from TK, PK, CKM, and CKS. Further analysis using multiple linear regression demonstrated that TK, PK, and CK are not significant predictors for their measure of TPACK. On this basis, they also refuted Koh et al.’s (2013) findings regarding TK’s direct positive influence on TPACK (see the previous section for a description of the instrument used). Future studies should address the question of whether these studies’ mixed findings may be attributable to the different approaches used to measure TPACK.

Despite the seemingly contradictory statements regarding the isolated consideration of TK, however, scholars agree that TK is important and necessary for preservice teachers (Guggemos & Seufert, 2021). The action-theoretical model designing digital resources highlights the importance of digital resources as an essential object of TK with respect to ICT integration in the classroom (Heine et al., 2022). In this context, it is precisely teachers’ knowledge about such resources that aligns with TK (ibid.). Such knowledge encompasses, among other things, competent handling of CC licences or knowledge of copyright (cf., Redecker & Punie, 2017). TK is thus an integral part of teachers’ knowledge base that facilitates their integration of ICT in the classroom. As such, it is important for all teachers, irrespective of the specific subjects that they teach.

2.2 Opportunities to learn (OTL)

In developing their professional (digital) knowledge during their initial teacher education, it is imperative that preservice teachers be provided with OTL (König et al., 2017; SWK, 2022). Moreover, student teachers’ acquisition of professional knowledge can also be evaluated by how they respond to OTL (Blömeke et al., 2014; Floden, 2015). OTL include, among other things, the content with which student teachers have engaged up to a certain time point (Schmidt et al., 2011; König et al., 2017). Regarding the structure of academic OTL, ‘the key components such as content of the subject […], the content of subject-specific didactics and the content of general pedagogy can be identified and empirically separated’ (König et al., 2017). Accordingly, this structure also reflects the German university teacher training system and the associated knowledge components, whereby the educational sciences (GPK), subject sciences (CK), and subject didactics (PCK) are involved in prospective teachers’ academic training.

While general pedagogical OTL has already been investigated and described extensively and its positive influence on preservice teachers' knowledge verified on several occasions (cf. König et al., 2017; Terhart, 2019; Depping et al., 2021), technological and technological pedagogical OTL remain relatively unexplored (Gerhard et al., 2023; Jäger-Biela et al., 2020). Wilson et al.’s (2020) meta-analysis of 38 studies goes some way toward addressing this desideratum, indicating an average positive effect of teacher education courses concerning technology integration on preservice teachers' knowledge (d = 1.057). However, it was not possible to determine conclusively which specific aspects of the courses contributed to the increase in knowledge (Wilson et al., 2020). König et al. (2022b) identified two intervention studies (pre–post design) in a scoping review that examined the effectiveness of OTL in improving preservice teachers' planning skills: both Neumann et al. (2021) and Zimmermann et al. (2021) reported significant increases in practice-relevant skills. The new approaches to teaching and learning that were required during the COVID-19 pandemic also demonstrated the importance of digital knowledge and its associated skills. König et al. (2020) found that students teachers’ exposure to (digital-related) OTL during their university education positively influenced their ability to cope with the demands that the COVID-19 pandemic imposed on teachers. By contrast, Gerhard et al. (2023) observed no direct effect of technological pedagogical OTL on TPK. However, it was also evident that student teachers had not had significant exposure to OTL, mainly because they were surveyed as early as in their sixth semester of their bachelor’s programmes – at least two years before graduating from university, after passing their Master of Education programme, which requires a further four semesters of study. This confirmed that OTL, which are instrumental in developing (preservice) teachers’ professional digital knowledge, are not sufficiently anchored in the curricula for all student teachers (Bertelsmann Stiftung 2021; Jäger-Biela et al., 2020).

3 Research questions and hypotheses

This article is guided by two major research questions. First, we examine the structure of preservice teachers’ professional knowledge required for ICT integration in the classroom across subjects. Second, we analyse the relationship between these knowledge facets and OTLs during teacher education. Considering the current state of the research (outlined above), we address the two research questions (RQ) with the following hypotheses (H):

  • RQ1:What structure exists between the three knowledge components of GPK, TPK, and TK among student teachers?

We hypothesise that the above three components of preservice teachers’ professional knowledge can be empirically separated (H1). The reasons for this assumption are, on the one hand, the existing fragmentation in teacher education. Hitherto, preservice teachers have acquired their professional knowledge through lectures and coursework pertaining to their subjects, subject didactics, and general pedagogy (see Sect. 4.1.4). Although the curricular structure relates to CK, PCK, and GPK, these knowledge components remain distinct, as few OTL explicitly target their linkage (König et al., 2017). We also expect this separation for the ‘new’ technological components, as they have not yet been integrated across all areas. On the other hand, we are also following Gerhard et al.s’ (2023) findings that few OTL specific to these knowledge areas have been offered to date. We assume that this circumstance has led to technological OTL being added to rather than merged with the existing curriculum (Scheiter, 2021). On this basis, we hypothesise that the different knowledge components can be recorded separately. In line with Mishra and Koehler’s (2006) model, we schematically favour the right-hand representation – which depicts professional knowledge in a three-dimensional model – over the left-hand model (Fig. 2), which considers professional knowledge as an overall factor without individual, separable components.

  • RQ2: Do preservice teachers’ knowledge levels differ in relation to their exposure to OTL?

Fig. 2
figure 2

Modelling teachers’ professional knowledge as one-dimensional (left) and three-dimensional (right)

Full size image

Regarding the preservice teachers’ knowledge, we assumed that the master’s students at the end of their first academic year would outperform the bachelor’s students at the end of their first academic year in all three knowledge tests (H2a). Those enrolled in the master’s programme had studied for three years more than those enrolled on the bachelor’s programme, and thus the master’s students had been exposed to significantly more OTL, allowing them to acquire more sophisticated professional knowledge with respect to facts, principles, and concepts and to enhance their existing knowledge. The modular structure of the teacher training programme at the University of Cologne ensures that some OTL are only offered in more advanced semesters. Thus, for example, all areas of competence are only fully covered upon completion of the degree. Simultaneously, this structure facilitates cumulative learning, as the offered content (partly) builds on that studied at earlier stages of the programme. Therefore, we assume that increasingly in-depth OTL will be offered to master’s students in relation to all components of their professional knowledge (GPK, TPK, and TK). This will also have an impact on the level of knowledge in the three tests (H2b).

4 Method

4.1 Sampling design

The data for this study were collected as part of a central teacher education quality assurance initiative at the University of Cologne (cf. König et al., 2018). The University of Cologne is among the largest teacher education universities in Germany and the EU. The data collection period took place during the summer term in 2022. Both bachelor’s and master’s students at the end of their second semester were surveyed, providing two distinct groups for possible comparisons with respect to competence levels. Whereas the bachelor’s students were in their first academic year at university, the master’s students had completed their three-year bachelor’s programme and had almost finished the first half of their master’s programme (the Master of Education programme comprises two years in total). Interpretations of possible differences in knowledge (RQ 1) and OTL (RQ2) between the bachelor’s and master’s students thus relate to the three-year difference in study between the two groups (Table 1). All students in the target group were contacted via their internal university email address and invited to participate in the survey online (N = 1249 bachelor’s and N = 737 master’s students). Each student was offered a small financial incentive (10 euros) to complete the survey.

Table 1 Sample characteristics
Full size table

A total of n = 702 students participated: n = 386 bachelor’s and n = 316 master’s students (response rate: 30.9% bachelor’s; 42.9% master’s; see Table 1).

For improved comparability of the test results, we considered only those students who had completed all three knowledge tests in our later analyses: a total of 619 students (320 bachelor’s, 299 master’s students; dropout: n = 83). Considering the sample ‘included for analyses’ relative to the population, we observed a slight overrepresentation of preservice primary school and special needs education teachers compared to the bachelor’s study population with differences of 4.3% and 4.0% respectively, while preservice grammar school and comprehensive school teachers were slightly underrepresented in the bachelor’s sample. Here, the difference was 6.2%. In terms of gender, our sample for the analyses shows an overrepresentation for the female students in the bachelor’s programme compared to the population with a difference of 6.6% (Table 1). Examination by χ2-tests shows that these deviations are statistically significant (Teaching programme type: χ2 (n = 320; df = 4) = 11.04, p = 0.03; Gender: χ2 (n = 320; df = 1) = 7.38, p < 0.01). This does not apply to the differences between our master’s sample and the population. Information on mean age and GPA of the population was not available.

4.2 Test instruments

4.2.1 Professional knowledge

To assess student teachers’ GPK in this study, we used the test instrument developed for the Teacher Education and Development Study – Mathematics (TEDS-M; Blömeke et al., 2010; König & Blömeke, 2010). The three content areas already described in Sect. 4 are represented in the test by four subcategories: structure, adaptivity, assessment, and classroom management/motivation. The test consists of 42 items in total (see Table 2 for examples). Various earlier studies that used the survey instrument had already demonstrated its construct and curriculum validity (König, 2014; König et al., 2018; König et al., 2022b). Prognostic validity has also been demonstrated for instructional quality and student learning in secondary mathematics (König et al., 2021).

Table 2 Item examples from the GPK, TPK, and TK tests
Full size table

The TPK test, developed in 2020 by an interdisciplinary team of educational researchers and media psychologists (Gerhard et al., 2020; 2022; 2023), is based on six different content dimensions: classroom management, structuring, diagnosis, evaluation, motivating learners, and dealing with heterogeneity of learning groups. The TPK test combines this with the interdisciplinary, generic use of (digital) technologies (Gerhard et al., 2023). In total, the instrument includes 34 items (see Table 2 for examples).

Our research group newly developed the TK test in 2021, with digital resources serving as the central reference based on findings from Heine et al.s’ (2022) literature review. One key area of focus was the conceptual placement of digital resources in the TPACK model. The results revealed that knowledge about digital resources in particular can be assigned to the area of TK. Dealing with digital resources, meanwhile, lies in the transitional area between TK and TPK. Given that digital resources could be identified as a central topic of TK, we decided to include this as a focal point in the knowledge test as well. Moreover, the DigCompEdu also assigns them a high level of importance for preservice teachers through a separate competence area (Redecker & Punie, 2017). The test’s design is intended to address several critiques. One of these pertains to the transfer of TK in isolation and addresses the significant fluctuation of this knowledge domain (Mishra & Koehler, 2006). To ensure that the test remained as stable as possible against such fluctuations over longer periods of time, the questions relate, among other things, to overarching aspects, such as the handling of licences, copyright, and data protection. These contents are also anchored, for example, in the Medienkompetenzrahmen NRW ([Media competence framework North Rhine-Westphalia], Medienberatung NRW, 2019). At the same time, it is necessary to embed the test items in authentic classroom contexts (Mishra & Koehler, 2006): although the test deals exclusively with TK content, this is situated in a pedagogical context. The aim is to provide examples of use from the school sector and thereby clearly demonstrate the relevance of this knowledge to preservice teachers (see Table 1 for examples).

The short test used in this study comprises 13 items. Based on the results of Heine et al.s’ (2022) review, we distinguished between ‘knowledge about’ (7 items) and ‘dealing with’ (6 items) digital resources with respect to the cognitive demands. Nevertheless, care was taken to address pure TK despite the proximity to the intersections in the TPACK model – particularly TPK. As such, we expect that TPK and TK can be empirically separated (RQ1).

4.2.2 OTL

We used three different survey instruments to capture OTL. In all three, the students were asked to indicate whether, during their initial teacher education, they had already been exposed to specific content in general pedagogy, technological pedagogy, and technology. The response options were dichotomous (‘yes’ = 1 and ‘no’ = 0). A survey instrument developed by König et al. (2017) closely aligned with the subcategories of the GPK test was used to capture OTL relating to GPK. This 37-item instrument assesses the pedagogical OTL. The items are distributed over four subscales: structuring, adaptivity, assessment and classroom management/motivation (Table 3). The same applies to the instrument for surveying OTL regarding the TPK (Gerhard et al., 2022, 2023). Analogous to the structure of the TPK test, the OTL instrument consists of six different subscales and the 31 items are closely aligned with them. The subscales are classroom management, structuring, diagnosis, evaluation, motivating learners, and dealing with heterogeneity of learning groups (Table 3). We removed the evaluation subscale for analyses (Gerhard et al., 2022). The test to assess OTL in relation to TK consists of 13 items and also aligns with the TK test’s structure. The scale refers to different forms and aspects of digital resources, as conceptualised by Heine et al. (2022).

Table 3 OTL items and reliability of the scales (Cronbach’s Alpha)
Full size table

4.3 Scaling and data analysis

To address our first research question (RQ1), we conducted a scaling in accordance with item response theory (IRT) using the ConQuest software package (Adams et al., 2020). Each test item is assigned an item parameter (‘difficulty parameter’) based on its empirical frequency (‘solution rate’), and all individuals are assigned an ‘ability parameter’ based on their response behaviour using the maximum likelihood procedure.

ConQuest allows the generation of deviance statistics as well as the determination of the Akaike information criterion (AIC) and the Bayesian information criterion (BIC). To increase the analytical power of the scaling analysis, the maximum possible number of cases was included respectively in the modelling (Bond et al., 2020). The deviance statistic (i.e., –2*log likelihood) of the model was generated. The deviation index (deviance; Wu & Adams, 2006) compares the global fit of the models examined and provides information on which of the models best fits the data (degree of goodness of fit). A smaller deviation indicates a better fit. To assess the empirical reliability of the three test instruments, we applied the expected a posteriori estimation (EAP; De Ayala et al., 1995). This is comparable to Cronbach's alpha and permits an unbiased description of the population parameters (Wu et al., 1997). If the three-dimensional model were to fit the data better, the IRT test results from the three-dimensional modelling (EAP estimates derived from the ConQuest scaling analysis) would be used to present the descriptive statistics.

To test the statistical influence of OTL on knowledge in terms of correlations, we first combined the individual items of the OTL into subscales (‘item parceling’; Bandalos & Finney, 2009). Next, we calculated several regression models with the software package Mplus (Muthén and Muthén, 1998-2017). The EAP scores of the three different knowledge tests were each considered as dependent variables.

Different regression models were calculated for each of the three knowledge tests. Model 1 may be considered the ‘base model’. The study section serves as the independent variable. In Model 2, the effects of study section are determined using gender and grade point average (GPA) as control variables. Subsequent models also consider the effects of OTL. Given that the GPK-OTL are correlated with one another, not only the OTL but also the individual subscales are included in the calculation to avoid multicollinearity. The same applies to the TPK-OTL but not to the TK-OTL, since there is only one scale in total here.

5 Results

5.1 Findings on the structure of professional knowledge (RQ1)

To address our first research question (RQ1) concerning the structure of preservice teachers' professional knowledge, we calculated different structural models (see the schematic representation for the one- and three-dimensional models in Fig. 2). Table 4 illustrates the deviance for different dimensional models, including the calculated results for AIC and BIC. They are based on the deviance statistics and consider both the number of parameters and the sample size.

Table 4 Deviance statistics of the different dimensions
Full size table

AIC and BIC are lower for the three-dimensional than the one-dimensional model. A significant improvement in model fit was indicated by the deviance statistics. To perform a more in-depth analysis, in addition to examining our hypothesised structure of having three-dimensions (H1), we additionally computed several two-dimensional models with GPK, TPK, or TK each being compared with the other two knowledge components subsumed to a second dimension, respectively. As the findings presented in Table 4 illustrate, none of these two-dimensional models fit the data better than the hypothesised three-dimensional model.

The results support the first hypothesis (H1) and favour the three-dimensional model over the one-dimensional model for preservice teachers’ professional digital knowledge.

Furthermore, all weighted mean squares of the 89 items are within the recommended range for both the one-dimensional (0.89 < MNSQ < 1.15) and three-dimensional (0.84 < MNSQ < 1.17) models. The average item discrimination ranges from 0.26 to 0.28. Table 5 summarises the central statistical parameters from the IRT scaling analysis of the three-dimensional model.

Table 5 Statistical parameters from the IRT scaling analysis of the three-dimensional model
Full size table

Table 6 details the latent intercorrelations for the three surveyed knowledge components. While high intercorrelations are shown for TK and TPK (> 0.8), those between GPK and TK or between GPK and TPK may be classified as moderately high (> 0.5) following Cohen’s (1992) recommendations.

Table 6 Latent intercorrelations of student teachers’ GPK, TPK, and TK
Full size table

ConQuest – the software used for IRT scaling – provides an item-person map. Figure 3 illustrates this for the three-dimensional model, with the student teachers’ ability shown on the left side. A single X represents 3.6 student teachers. The difficulty level of the test items is indicated on the right side. Each item has a number corresponding to the numbering in the scaling analysis (1–89). In the three-dimensional model, each item is assigned to a single dimension. Students whose ability parameter on that dimension is equal to the difficulty parameter for that item have a 50% chance of success on that item (Adams et al., 2022). Thus, the higher a person’s ability parameter is on this scale, the more likely it is that they will successfully complete the task. The three test instruments cover the students' abilities satisfactorily. This may be deduced from the fact that the difficulty levels of the items (right) reflect the individual abilities (left) well. Furthermore, the three-dimensional model and its results demonstrate that it was possible to create a test score for each dimension of professional digital knowledge, i.e., GPK, TPK, and TK. The three instruments’ EAP reliabilities are each within an acceptable range (Table 5). Given that the three-dimensional model fits better, we exported the EAP estimates from it. For further analyses and improved readability, the EAP estimates were linearly transformed to a mean of 50 and a standard deviation of 10 for each of the GPK, TPK, and TK scales (Table 7).

Fig. 3
figure 3

Item–person map of three-dimensional Rasch scaling

Full size image
Table 7 Testing for test score differences among bachelor and master student teachers
Full size table

5.2 Findings on the relationship between knowledge and OTL (RQ 2)

The statistics outlined in Table 7 demonstrate that, according to our second hypothesis (H2), master’s students significantly outperform bachelor’s students in all three knowledge tests. Based on the transformed EAP estimates, we calculated a t-test for independent samples. Given that hypothesis testing always depends on sample size, we also computed effect size d (Cohen, 1992). While the effect for GPK can clearly be classified as strong, it is in the upper-middle range for TPK and TK (Cohen, 1992).

Table 8 details the descriptive statistics of the OTL subscales. Comparison of the mean values of the individual scales between bachelor’s and master’s student teachers reveals a difference for all subscales, indicating that master’s students had been exposed to more OTL than bachelor’s students. The effect of the differences evaluated by η2 in the last column varies considerably. By far the largest effect can be seen in the difference of the subscale 'P-Structuring' in the area of pedagogical OTL. While the difference in means is absolutely 0.48, the effect size may be classified as large, with η2 = 0.53. The other pedagogical OTL subscales also reveal a significant difference between the two groups of students with a mean of approximately 0.30. By contrast, the difference for OTL with a technological component is small. Although medium effects are still shown for the subscales of technological pedagogical OTL (Cohen, 1992), the average difference in absolute terms is small (0.15). For the pure technological OTL, the difference is of no practical significance.

Table 8 Descriptive statistics for the OTL scales
Full size table

For the technological knowledge components, OTL scarcely appears to play any part in teacher education: even in the master's programme, only 30% of the students on average indicated that they had been exposed to the content in their studies hitherto. For the pure technological content, the value for master’s students is even lower at 15%.

To assess the possible influences of the different OTLs on students’ performance in the knowledge tests, the intercorrelations of study variables were calculated using Mplus (latent) and SPSS (manifest) (Table 12 in the Appendix).

In the subsequent regression analyses, the relationships between OTL and knowledge persist for the various dependent variables, even after controlling for gender and GPA. The negative GPA correlations are due to the German scoring scheme, wherein a 1 represents the best score and a 4 represents the worst score.

Table 9 summarises the results of the regression analyses with the GPK test score as the dependent variable. In addition to the influences of the different subscales, we also determined the influence of the overall score for the OTL in the GPK domain (GPK-OTL). With the exception of the ‘P-OTL Assessment’ subscale, the other OTL for GPK show highly significant statistical influences on students’ GPK knowledge test scores (Table 10).

Table 9 Findings from regression analyses with GPK test score as dependent variable
Full size table
Table 10 Findings from regression analyses with TPK test score as dependent variable
Full size table

The regression analyses with TPK knowledge test score as the dependent variable indicate that OTL in the TPK domain has a small, insignificant negative impact on test scores, as demonstrated by the overall score TP-OTL score (− 0.050). In the case of the subscale ‘TP-OTL Dealing with Heterogeneity of learning groups’, this is of minor significance. Only the subscale ‘TP-OTL Structuring’ shows slightly positive influence.

The regression analyses with the EAP score of the TK knowledge test reveal results similar to those for the TPK (Table 11). However, the influence of TK-OTL may be evaluated as stable with − 0.001.

Table 11 Findings from regression analyses with TK test score as dependent variable
Full size table

6 Discussion

Herein, we have aimed to clarify the structure of teachers’ professional knowledge (RQ1). Building on conceptual work by Shulman (1987) and Mishra and Koehler (2006) regarding teacher knowledge components, we argued that the professional knowledge required for teachers’ integration of ICT in the classroom across subjects is multidimensional in nature. According to our hypothesis (H1), generalised knowledge is insufficient to cope with the diverse and increasingly complex demands of digitalisation in education (Mußmann et al., 2021). Rather, teachers require individual knowledge components, such as GPK, TPK, and TK. Our second hypothesis (H2a), which assumed a higher level of knowledge across all three knowledge components among master’s students compared to bachelor’s students, was supported. However, our findings confirmed the expected link only for the relationship between general pedagogical OTL and GPK. The detailed insight that initial teacher education programmes offer into curricular content and the level of knowledge that student teachers exhibit not could be confirmed for either TPK or TK. Therefore, we could not fully confirm H2b.

6.1 Structure of professional knowledge

Three standardised knowledge tests were used to address the question regarding the structure of preservice teachers’ professional knowledge required to meet the challenges associated with integrating ICT into classroom teaching across subjects (RQ1). The findings from the IRT scaling analysis reveal that the three-dimensional model is a significantly better fit for the data than the one-dimensional model. While the three-dimensional model distinguishes between GPK, TPK, and TK, the one-dimensional model operationalises knowledge across all test items.

The correlation between the knowledge tests with the technical components (TPK and TK) is higher than that shown by both with GPK. However, the correlation between TPK and GPK is slightly higher than that between TK and GPK. Nonetheless, the fact that they are discrete constructs was also verified by the three different two-dimensional models of the IRT scaling analyses. The three-dimensional model is also superior to the three models (Table 4). However, the three knowledge components are not inter-correlated in the same way. One possible reason for the tests’ stronger correlation with the technological components may be that the TK items were embedded in educational contexts to raise the preservice teachers’ awareness of the technical aspects’ relevance for school and teaching (Sect. 6.2.1). At the same time, it may be more difficult to delineate TK and TPK from one another. Nonetheless, empirical findings highlight the importance of distinguishing between TK and TPK.

6.2 Different levels of professional knowledge

In addressing RQ2, we investigated whether differences in student teachers’ knowledge in relation to their OTL could be discerned. Our analyses revealed that preservice teachers attain different levels of knowledge depending on their initial teacher education career stage.

However, superior performance can only be attributed to the students’ OTL to a limited extent (H2b). The regression analyses showed that GPK-OTL in particular exert a positive statistically significant influence on student knowledge. Contrary to our expectations, the influence of OTL with a technological component is negligible, if not non-existent. This may be attributed to several possible factors.

Our survey found that student teachers are exposed to very few TK-OTL and TPK-OTL. Thus, the demonstrably low correlation between technological OTL and students’ TK and TPK may be due to a lack of provision at the institutional level (Gerhard et al., 2023). The relatively limited provision of technological OTL may also be due to its curricular status as not yet conventional, as is the case for GPK-OTL in Germany (Terhart, 2019). Teacher education is progressing slowly in terms of integrating offerings that involve a technological component (Gudmundsdottir & Hatlevik, 2018). Based on our survey of students from the University of Cologne, we can corroborate Bertelsmann’s (2021) assertion that the incorporation of mandatory courses on digital media literacy into the curriculum showed only minimal progress between 2017 and 2020. Our findings extend this observation into 2022.

Although we were unable to find a direct link between the content aspects of technological pedagogical or technological OTL and corresponding preservice teacher TPK and TK, it is promising that both knowledge components nevertheless appear to play a significant role in initial teacher education. Given that the master’s students outperformed the bachelor’s students, we assume that specific OTL supports the acquisition not only of conventional GPK but also of such innovative knowledge facets as TPK and TK. This is also evident in the comparison of students based on their chosen school type.

As the present study is one of the first to apply standardised tests rather than self-reports, the mean differences in TPK and TK between bachelor’s and master’s students, with middle effect size, emerge as an important component amid the various research desiderata relating to digitalisation in educational contexts. However, future research should continue to investigate which OTL are most salient with respect to fostering such components of the teacher professional knowledge base.

6.3 Pedagogical implications

The results of our analyses generally illustrate that the classification of knowledge components as proposed by Mishra and Koehler (2006) is useful. Teachers’ professional knowledge must be viewed as a multidimensional construct. While we might consider this to be an argument in favour of no longer taking a critical view of the isolated promotion of TK (Guggemos & Seufert, 2021; Mishra & Koehler, 2006), we still should not definitively reject interdisciplinary promotion. The COVID-19 pandemic, among other phenomena, has highlighted that teachers’ ability to be flexible in their thoughts and actions is a competence that transcends disciplines and knowledge (König et al., 2020). TK, in particular, can enhance this flexibility for the current era of digitalisation (Heine et al., 2022).

To support preservice teachers in developing their digital competencies, the promotion of all three knowledge components, i.e., GPK, TPK, and TK, should be accounted for during teacher education programmes and early career teacher induction. The recent rapid spread of ChatGPT has highlighted the significance of teacher education and teacher professional development institutions responding promptly and effectively to technological innovations. Instead of prohibiting these innovations at the university level (The Guardian, 2023), OTL need to be established that fit the rapidly emerging challenges. These avenues should facilitate the development and enhancement of preservice teachers’ competencies, in particular related to TK, in utilizing such innovations effectively. Additionally, for preservice teachers, specific OTL approaches must be implemented, enabling them to impart these competencies to their future students, which demands an appropriate acquisition of TPK. Thus, the demands that international educational institutions place on preservice teachers’ digital competencies may be met (UNESCO, 2011; UNICEF, 2022; Redecker & Punie, 2017).

6.4 Limitations and future research

Although the study was conducted at only a single university in Germany, we are confident that the findings from our analyses for RQ1 are relevant both nationally and internationally. As noted in the introduction, various national and international institutions have made it their goal to prepare teachers for the demands of digitalisation. This undertaking is particularly relevant to teacher education institutions, e.g., Redecker and Punie (2017). The widespread use of the TPACK model at the international level also indicates that this is a topic of global interest.

This study explicitly refers to the entire professional knowledge of (preservice) teachers. In the TPACK model, different knowledge areas are addressed, of which we have examined the three cross-disciplinary ones (GPK, TPK and TK). Nevertheless, the exclusion of subject-specific knowledge components represents a limitation. As such, future investigations focusing on specific subjects and involving the various subject didactics and disciplines are warranted. For instance, the DiKoLeP project (Digital Competencies of Pre-Service Physics Teachers; Große-Heilmann et al., 2022) has devised a university teaching approach to enhance the digital-media PCK among preservice physics teachers. Future research could aim to combine cross-disciplinary and subject-specific domains to comprehensively address the entire TPACK model.

The cohort comparison, i.e., the comparison between bachelor's and master's study sections, is based on a simple cross-sectional analysis. Generalised conclusions derived from study section on the level of knowledge are therefore limited in their validity. However, we compared the two samples on the basis of several socio-demographic characteristics and consequently, the samples are comparable samples (see the Methods section). Nevertheless, it would be interesting for future studies to examine the individual developmental progression of the students on a longitudinal study plan. It is possible that students benefit from early learning opportunities and can then make better use of further learning opportunities later in their studies.

The test instrument used to measure TK comprises 13 items only. Therefore, the findings potentially limited generalizability cannot be overlooked. An extended test version, consisting of 25 items, that was not available for the present study, will be applied in future investigations.

As the findings reveal, few OTL are provided in the area of TK. The concrete influence that these exert on students’ performance and knowledge levels can thus be assessed only to a limited extent. For this reason, a learning module was developed within the context of a seminar. In accordance with intervention study guidelines, the OTL and knowledge will be surveyed and analysed again.

Moreover, the query regarding OTL was conducted through student self-report, and comparison with the university curriculum with respect to the technological components of OTL was only made for TPK (Gerhard et al., 2023; Jäger-Biela et al., 2020). However, the students’ self-reports suggest that digitalisation has yet to fully assert itself in the teacher education curriculum, at least at the University of Cologne. To fully appreciate the reasons for the differences between the bachelor’s and master’s students, further investigation will be necessary. Informal OTL may play a role, but these were not surveyed in the present study.

7 Conclusion

In this study, 619 teacher education students from the University of Cologne participated. Among them, 320 were in their second bachelor’s semester, and 299 were in their second master’s semester. Through the use of three distinct knowledge tests assessing GPK, TPK, and TK, it was empirically demonstrated that these knowledge components can be distinguished from each other.

The evidence supporting this conclusion was derived from an IRT-Rasch scaling analysis, which indicated that the three-dimensional model outperformed the one-dimensional model. Additionally, master’s students outperformed bachelor’s students in all three knowledge components. However, when analysing OTL, only the regression analysis for GPK reveals that general pedagogical OTL had a statistically significant impact on the difference in knowledge levels between bachelor’s and master’s students. Technological OTL provision for preservice teachers was lacking.

This raises the question of whether the limited availability of technological OTL or the effectiveness of existing curricular provision accounts for this finding. Further investigation in future research is warranted to delve into this matter.