Examining pre ‐ service teachers' Technological Pedagogical Content Knowledge as evolving knowledge domains: A longitudinal approach

The aim of this study is to outline the development and changes in pre ‐ service teachers' technological pedagogical content knowledge (TPACK) assessments during the first 3 years in teacher education. Specifically, research was conducted at three measurement points over a 3 ‐ year teacher education period. The target group consisted of pre ‐ service teachers ( N = 148) from three Finnish universities. Results indicate a growth in confidence related to all TPACK areas during the research period. The strongest gains were in pedagogical content knowledge. In addition, the gains were larger in other areas related to pedagogical knowledge than areas related to technology or content knowledge. In areas without pedagogical knowledge, the changes were more moderate. In the discussion section, recommendations are provided on the potential of longitudinal use of the TPACK model to study and improve the development of pre ‐ service teachers' TPACK.


| INTRODUCTION
The development of information and communication technology (ICT) has placed new expectations for current and future working life (Spector, 2010). These expectations relate to the working skills needed in the future, such as skills for collaboration, communication, creativity, problem solving, and critical thinking along with ICT skills and ICT literacy (cf. Voogt & Roblin, 2012), are currently considered as 21st century skills. This development creates new possibilities and expectations for today's educational systems. Within the context of the Finnish educational system, especially the emphasized role of ICT, can be seen in the National Curriculum (Finnish National Board of Education, 2015) where ICT is seen both as a means for and a target of learning.
ICT is part of the everyday world of today's youth and pre-service teachers. Yet, acquiring the knowledge and skills to take advantage of ICT in pedagogically meaningful ways, to understand the benefits and potential of ICT for educational purposes is challenging (Lei, 2009;, today's pre-service teachers have rather positive attitudes in general concerning the possibilities of ICT in education, still when it comes to concrete work, that is, actually using ICT in the classroom, attitudes are more reserved. In addition, based on a review by Brown and Englehardt (2017), it seems that pre-service teachers are often rather uncomfortable with integrating technology into teaching. This poses challenges for teacher education and according to Kirschner and Selinger (2003), teachers' inadequate skills and expertise are the bottleneck of taking advantage of ICT in education.
The technological pedagogical content knowledge (TPACK) model serves as a framework for studying pre-service teachers' knowledge related to the use of ICT in education based on three foundational knowledge areas: technological knowledge (TK), pedagogical knowledge (PK), and content knowledge (CK) as well as three combined areas: technological pedagogical knowledge (TPK), technological content knowledge (TCK), and pedagogical content knowledge (PCK) (Mishra & Koehler, 2006). The TPACK framework is an actively used model, and according to Harris, Phillips, Koehler, and Rosenberg (2017), there are over 1,200 publications related to the TPACK framework. Despite the active use of the TPACK framework, there is a need for more longitudinal studies focusing on the development of TPACK (Hofer & Grandgenett, 2012;Valtonen et al., 2017;Voogt & Roblin, 2012) in order to support the understanding of pre-service teachers' TPACK during the teacher education process. The current study provides a longitudinal perspective on pre-service teachers' TPACK areas based on three measurements conducted during the first 3 years of teacher education (i.e., Bachelor of Arts [Education,180 ECTS]). The aim is to examine pre-service teachers' yearly TPACK assessments and changes in assessments in the context of Finnish teacher education.

| BACKGROUND OF THE STUDY
Integration of ICT by (pre-service) teachers has been studied using different theoretical frameworks. One of these frameworks is the theory of planned behaviour (TPB) by Ajzen (1991). Based on the theory of planned behaviour framework, certain behaviour, in this case the use of ICT in education, is determined by attitudes, subjective norms, and perceived behavioural control related to behaviour. Another model used is the technology acceptance model by Davis (1989). According to the technology acceptance model, perceived usefulness of technology and ease of technology use are the main factors affecting the use of technology. Again, the unified theory of acceptance and use of technology by Venkatesh, Morris, Davis, and Davis (2003) suggests that acceptance of technology is affected by performance expectancy, effort expectancy, social influence, and facilitating conditions. There are several studies focusing on (pre-service) teachers' use of technology using the abovedescribed frameworks (see. Birch & Irvine, 2009;Teo, 2011;Teo & Tan, 2012;. Still, it can be argued that these models are problematic in an educational context because the lack of areas focusing on pedagogy or content areas. In order to better highlight these, the association between technology, pedagogy, and content, the TPACK framework was used to frame this research.

| Defining TPACK
TPACK is a theoretical framework for studying and describing (preservice) teachers' knowledge related to pedagogically meaningful use of ICT in education ). TPACK focuses on teachers' knowledge from three foundational perspectives: TK, knowledge related to various available technologies and their characteristics. TK also refers to interest in following the development of new technologies. CK refers to the central theories and concepts of the discipline. In addition, CK refers to the nature of the knowledge and the means of inquiry of the discipline (e.g., biology, mathematics, and history). PK refers to knowledge of learning processes and the readiness to support and guide the learning situation and learning process. PK is a generic form of knowledge related to the theories of learning.
These foundational areas combine as intermediate TPACK areas.
TPK refers to an understanding of the nature of teaching and learning with technology: the benefits and disadvantages of different available technologies for certain pedagogical practices. PCK refers to knowledge of combining CK with PK in a way that makes the CK easy to understand and learn for others. TCK refers to how technology is used to further develop certain content areas (e.g., biology and mathematics) and how technology is used within a certain discipline. Koehler, Mishra, and Cain (2013) define TPACK as "an understanding that emerges from interactions amongst content, pedagogy, and technology knowledge […] knowledge underlying truly meaningful and deeply skilled teaching with technology" (p. 66). The following table (Table 1) contain definitions with examples. The following table is modified based on TPACK review by Chai, Koh, and Tsai (2013).
Usually, TPACK is seen as a balanced entity of three overlapping knowledge areas. However, Doering, Veletsianos, Scharber, and Miller (2009) argue that TPACK should be seen as an evolving and multifaceted entity instead of a static representation comprising equally sized areas. Doering, Veletsianos, Scharber, and Miller (2009) (Koh & Chai, 2014;Valtonen, Kukkonen, Kontkanen, Mäkitalo-Siegl, & Sointu, 2018), indicating that instead of a static entity, TPACK needs to be seen as dynamic and developing. This poses challenges for studies describing the nature and development of pre-service teachers' TPACK areas.
In the PCK literature, Gess-Newsome (1999)  and other areas are seen as latent sources, useful only when transformed into PCK. This discussion can also be seen in TPACK research (Voogt et al., 2013). According to Angeli and Valanides (2009), the question is whether TPACK is a unique body of knowledge itself, constructed from other latent forms of teacher knowledge as a "transformative view" or is TPACK a combination of other forms of teacher knowledge and enactment during teaching in an "integrative view?" In addition to these two extremes, in PCK research, Gess-Newsome (1999) define a position that can be considered also in the TPACK research, that is, in a position between the extremes by recognising the foundational knowledge areas and the actual PCK. According to Gess-Newsome (1999), "New knowledge gained through preparation programmes and teaching experiences increases the organization and depth of both foundational knowledge domains and PCK, though changes in one knowledge base will not necessarily result in changes in others" (p. 13). In the current study, seven TPACK areas were measured over time: TK, PK, CK, TPK, TCK, PCK, and TPACK.

| Cross-sectional and longitudinal studies focusing on TPACK
This section outlines the results from previous studies concerning (pre-service) teachers' TPACK areas. Schmidt et al. (2009) conducted a pilot study in the United States in order to develop a TPACK measurement instrument. They also reported pre-service teachers' assessment of their TPACK. The target group consisted of 124 preservice teachers. Their results indicate that the pre-service teachers perceived their TPK as at the highest level and their level of PK and TPK as rather high as well. The areas assessed as the lowest were CK (science) and PCK. Overall, the variations amongst TPACK areas were rather small. Koh, Chai, and Tsai (2010) studied 1,185 firstsemester pre-service teachers' TPACK in Singapore. Their results indicate that pre-service teachers perceived their PK as the highest and CK as the lowest. Again, the differences between TPACK areas were small and all areas were perceived as above average, that is, no areas were assessed to be particularly low. Similarly, Archambault and Crippen (2009) studied 596 online teachers' TPACK in the United States. Their results indicate that pedagogical and TK and the combined PCK areas were rated highest. Areas of lower ratings were related to TK. Graham et al. (2009) measured in-service science teachers' TPACK (TPACK, TK, TPK, and TCK) as part of the teachers' professional development programme. The results indicate that before the programme, the highest-rated area was TK and TPK and the lowest-rated area was TCK. Nevertheless, the highest gains were in TCK. Although the challenging aspect revealed by these results is that they provide a rather contradicting picture of strong and weak TPACK areas; still, the important role of PK is highlighted in most of the studies.
In addition to cross-sectional studies, there are several studies focusing on the effects of different courses or other interventions for developing TPACK. Typically, these studies contain elements for studying TPACK-related areas before and after the intervention using both

PK
Knowledge of different teaching and learning approaches, theories of learning, and assessment methods without references to any specific content areas.
Knowledge of how to use inquiry-based learning method.

CK
Knowledge of subject matter, different discipline without considerations of teaching the subject matter.
Knowledge of mathematics, arts, literature, etc.

PCK
Knowledge of how to combine the CK and PK in order to make the learning of the subject matter easy, to make the content understandable.
Knowledge of examples and analogies to teach mathematics.

TPK
Knowledge of how to take advantage of appropriate ICT for supporting certain teaching and learning approaches without considering subject matter.
Knowledge of Kahoot-application to activate students or Padletapplication for brainstorming.

TCK
Knowledge of how to represent, research, and create the content with ICT without consideration of teaching. Knowledge of how ICT is used by content experts.
Knowledge of how to use content-specific simulations, navigation app in geography, or SPSS in statistics.

TPACK
Knowledge of how to combine different areas, how to use appropriate pedagogical approaches for certain content with appropriate ICT.
Knowledge of how to use the Padlet application for supporting students' brainstorming and sharing of ideas in a biology course. qualitative and quantitative methods (see Doering, Veletsianos, Scharber, & Miller, 2009;Graham, Borup, & Smith, 2012

| Purpose of the study
According to Voogt and Roblin (2012), there is a need for longitudinal studies conducted over a longer time period and with a larger target group. The aim of this study is to provide new insight into the nature of pre-service teachers' developing TPACK. To meet these aims, we measure all seven areas of TPACK: the foundational areas (TK, PK, and CK) and their confidence with integrating them (PCK, TPK, TCK, and TPACK). This way of studying pre-service teachers' TPACK aligns with the ideas of Gess-Newsome (1999), recognizing the importance of separate foundational elements (PK, CK, and TK) and the intermediate areas (PCK, TPK TCK, and TPACK). We assume these insights are important for a better understanding of the way pre-service teachers' TPACK evolves. According to Tondeur et al. (2012) and Gao, Wong, Choy, and Wu (2011), the majority of pre-service and beginning teachers have difficulties using ICT in education. With yearly assessments, we are able to identify the confident and weak TPACK areas and possibly better understand the developments that lead to the challenges suggested by Tondeur et al. (2012) and Gao et al. (2011).
Similarly, the visualization of separate TPACK areas provides perspectives for the discussion concerning the structure of TPACK, that is, whether the development is equal or if there are differences between TPACK areas over time and whether we need to acknowledge these development trends more thoroughly when discussing the structure of TPACK.

| RESEARCH METHOD
This longitudinal study focuses on the changes in pre-service teacher's TPACK assessments using three measurement points during the first 3 years in teacher education, that is, bachelor's degree studies. The following section outlines the main features of Finnish bachelor's degree studies in teacher education focusing on the elementary level, Grades 1 to 6.

| Context of the study-Finnish teacher education
Elementary-level teacher education in Finland includes a strong practical and research orientation providing teachers with competencies for continuous professional development. The aim is to educate reflective teachers who are capable of using research-based evidence in their everyday work (Kynäslahti et al., 2006). Teacher education studies focus on areas such as educational sciences, educational psychology, sociology, educational systems, and their meaning in society (Malinen, Väisänen, & Savolainen, 2012). Pre-service teachers are provided with skills to combine different content area knowledge with PK and become familiar with different technologies for teaching and learning. Pre-service teachers are expected to gain skills for critical and creative thinking and to conduct research activities independently.
In practice periods, pre-service teachers integrate theoretical pedagogy and multidisciplinary subject studies by using a variety of teaching methods, inspiring learning environments and ICT.
Teacher education in Finland consists of a Bachelor of Arts (Education) degree (180 ECTS) and a Master of Arts (Education) degree (120 ECTS). The bachelor's degree covers the first 3 years of teacher education, that is, the area under focus in this longitudinal research. In Finland, a teacher must have master's degree in order to serve as a qualified teacher. At the universities participating in this study, the bachelor's degree contains the study units listed in Table 2, which is based on the curricula of the participating teacher education units.
There are differences in the names of the units between universities demonstrations, and so forth, and vary from face-to-face courses to more blended courses and fully online courses. The role of ICT varies between universities and courses. All of the universities in this study provide their students with Internet access within campus buildings as well as cloud services (e.g., Office 365, GAFE, PedaNET) and personal online environments are provided for pre-service teachers.

| Respondents
The target group consists of three cohorts of pre-service teachers from three Finnish universities starting their studies in autumn 2014.
The purpose and aims of the research were explained to all participants. Participation was voluntary. Data was collected using an online questionnaire. The total size of the three cohorts, representing the annual intake of new students for the three universities, was 365 pre-service teachers.
Research data were collected from the pre-service teachers' courses at three measurement points during 2014, 2015, and 2016 as follows: (a) permission for collecting the data was acquired from each teacher education department, (b) courses for the whole cohort were selected for data collection in each university, (c) researchers explained the aims of the study to the participants in the selected courses, (d) informed consent was obtained from all participants.
The total number of respondents at each measurement point, including both male and female respondents, varied from 267 to 209 (Table 3).
Validity of the instrument has previously been tested using explor-  0.900 and the values of RMSEA and SRMR were lower than 0.08. We also tested the univariable CFA models. The models were well fit with the values of CFI, and TLI were equal to 1.000 and the values of RMSEA and SRMR were 0.000. Due to the limited tables and number of words in this paper, the details of the validity process, the model fit indices of the CFA models, and the factor loadings of the CFA models at each time point are not reported in this paper.
Internal consistency for each scale using Cronbach's alpha (α) values are listed in Table 4 with example items. All α values were adequate at above 0.80 (e.g., Nunnally & Bernstein, 1994). The instrument contains 38 items using a six-point Likert scale (1 = I need a lot of information about the measured area; 6 = I have strong knowledge of the measured area). The areas of TPACK related to PK, that is, PK, TPK, PCK, and TPACK are grounded in 21 ST century skills such as collaboration, problem solving, creative thinking, and critical thinking (see Voogt & Roblin, 2012). The aim of the instrument is that this way, we are able to better acknowledge different pedagogical approaches, for example, knowledge about how to support collaborative learning practices.

| Analysis of the data
Analysis of the data was conducted using SPSS and Mplus (version 7) software in the following phases. In the first phase, we tested the measurement invariance to assure the measurement is invariant across the time. We ran a set of increasingly constrained structural equation models (SEMs), and tested whether the difference between these models is significant. The model fit differences were determined by three goodness-of-fit indexes: the CFI (Bentler, 1990), the RMSEA (Steiger, 1990), and the SRMR (Bentler, 1995). The change criteria were −0.010 for CFI; 0.015 for RMSEA; and SRMR of 0.030 for metric invariance, 0.015 for scalar invariance (Chen, 2007;Cheung & Rensvold, 2002). The results of measurement invariance tests showed that although the full support of measurement invariance has not been found, the partial measurement invariance was supported.  (Table 4). In the third phase of the study, latent growth curve modelling (LGCM) was conducted using Mplus.
LGCM is a class of SEMs and it allows obtaining the developmental trajectories of variables in longitudinal data. It also determines significant individual differences in the developmental trajectories by statistical significances of variances. In this phase of the analysis, seven LGCMs were fit using five fit indices: (a) the chi-square goodness of fit test, (b) CFI (Bentler, 1990), (c) the TLI (Tucker & Lewis, 1973), (d) SRMR (Bentler, 1995), and (e) RMSEA (Steiger, 1990). According to Hu and Bentler (1999), the recommended cut-off values for a well-fitting model need to be greater than 0.90 (for CFI and TLI) and below 0.08 (for RMSEA and SRMR). In addition, the descriptive statistics (mean value) were calculated for each TPACK area for each year (more details in Appendix A).

| RESULTS
The LGCM models were fitted separately for each variable. Well-fit was obtained for all variables except PCK and TPACK. The results indicate that the mean values for the different TPACK areas vary between 2.22 and 4.12, indicating variation in the pre-service teachers' perception of their knowledge, that is, from "need more knowledge" to "more confident perception". Moreover, none of the results indicated a strong knowledge of the TPACK areas. However, the tendency across the three measurement points was towards better confidence in the different TPACK areas. Still, we report the PCK and TPACK in order to outline the changes in all TPACK areas. Reasons for a not-acceptable fit are discussed in more detail in the discussion section.
The latent growth curve trajectories of all the TPACK areas ( Figure 1) show that assessments of all TPACK areas had increased, but the size of the changes still varied between TPACK areas (see Table 6). Table 6 shows that all the mean growth rates are significantly positive. In the baseline, PK and CK had the highest starting points compared with other areas of TPACK, whereas TCK was at the lowest level. The slopes showed that the developments of TPACK areas were different over time. For example, PK, TPK, and TCK increased more than TK and CK (e.g., at the first measurement, the differences between PK and CK were minimal; PK M = 3.14, CK M = 3.03). At the third measurement, the difference was bigger, PK was perceived higher (M = 4.32) and CK was lower (M = 3.57).

| Assessments of TPACK areas over time
In order to provide more insight into three measurements, the mean values of the TPACK assessments are presented in Table 7.
At the beginning of teacher education, PK had the highest assessment score of all the TPACK areas (M = 3.14) and CK was scored at almost

| TPACK changes
The Overall, during the 3 years of teacher training, the greatest increases in assessment rating were in the areas related to PK, especially PCK where the gain was from 2.90 to 3.98 (mean change 1.08, growth rate = 0.531) and TPACK with a gain from 2.54 to 3.57 (mean change 1.03, growth rate = 0.504). In addition, PK and TPK showed a strong increase. At the beginning and at the end of the studies, PK was

| DISCUSSION AND CONCLUSION
The aim of the study was (a) to explore pre-service teachers' perceptions of their TPACK during the first 3 years of teacher education and (b) to investigate the changes of pre-service teachers' TPACK assessments over that period. The results indicate that teacher training seems to have a beneficial impact on TPACK; improvement was perceived to be gained in all of the TPACK areas studied, although the extent of the improvement differed between TPACK areas (see Table 7). Interestingly, the results indicate positive gains, especially in areas related to pedagogy. The strongest gains were in PCK, TPACK, and PK. The lowest gains were in CK and TK. These results indicate that teacher education in Finland provides strong support for the development of pre-service teachers' pedagogical thinking, providing pre-service teachers with stronger confidence in areas related to pedagogy than other TPACK areas.
Based on previous studies, it seems that pre-service teachers have difficulty finding ways to take advantage of ICT in teaching and learning in pedagogically meaningful ways (Lei, 2009;Valtonen et al., 2011).
This aligns with the results of the first measurement of this study. In the first year, TPACK was perceived to be low, that is, the ability to use ICT in pedagogically meaningful ways in science learning was assessed as weak. However, the results show that improvement was perceived to be gained, most of all in TPACK and PCK. According to Ertmer and Ottenbreit-Leftwich (2010), the role of concrete examples of pedagogically meaningful ways of using ICT in education is vital for pre-service teachers to gain the knowledge and confidence to use ICT in education. Also, Tondeur et al. (2012) proposed that authentic learning situations combining theoretical knowledge with practice are important elements for pre-service teachers to gain understanding of pedagogically meaningful use of ICT in education. Furthermore, the role of teacher educators as role models for pre-service teachers is important (Tondeur et al., 2012). Based on the results by Tondeur et al. (2012) and Ertmer and Ottenbreit-Leftwich (2010), the role of multidisciplinary courses within Finnish teacher education is assumed to be important in this respect, and that such courses focusing on combining the content areas of different disciplines with PK and, typically, with pedagogically sound ICT practices, are crucial for development within these two TPACK areas.
In contrast to the changes in TPACK, the changes in TK remained modest during first 3 years of teacher education, posing challenges for developing teacher education. TK, that is, knowledge focusing on plain technology (see    In addition, TCK is a challenging area for teacher education. In line with previous studies (see Graham et al., 2009), TCK was perceived as the lowest TPACK area within all measurements. The challenge with TCK may be the limited availability of content-specific technologies compared with more general-level technologies and software, such as tablet computers and office software. We assume that one reason for this is that content-specific technologies and software are typically rare and intended only for specific purposes, and the computers used in teacher education are not usually equipped with such technologies or software. Nevertheless, considering the rapidly increasing availability of a vast array of different applications for various purposes, this situation may well change in the coming years.

| Limitations and future studies
The LGCM provides an efficient method for assessing longitudinal data, the changes in pre-service teachers' TPACK assessments. However, the current study had some limitations, and as such provides opportunities for future research. The sample attrition in a longitudinal study is inevitable; in this study, the sample size remained acceptable (N = 148). Results indicated a poor fit for the PCK and TPACK. This poses questions pertaining to the reasons for these measures. Our assumption is that there are changes in the ways pre-service teachers assess their TPACK areas during the first 3 years in teacher education and the assessments may not be consistent during that period. Two areas with poor fit indexes were those with the highest gains, and this poses questions concerning the methods used, that is, there is a need for using different approaches for modelling the seven elements of TPACK for longitudinal settings. In addition, an interesting approach would be adding a fourth measurement point, to study whether that affects the fit levels. Also, because of limited space, the more detailed descriptions and tables focusing on the TPACK-21 instrument and especially measurement invariance will be further studied and reported in future articles.
Altogether, these challenges highlight the complex nature of the TPACK framework with seven areas. Within previous studies, the measuring and modelling of TPACK using SEM has posed difficulties (Archambault & Barnett, 2010;Chai, Koh, Tsai, & Tan, 2011). Still, we assume that five out of seven elements with good fit is a good result and a good starting point for future studies modelling the growth trajectories on TPACK within teacher education. We assume that in the future, the longitudinal research approaches will pose new challenges for the instruments. In this case, in the future there will be a need for further development of the TPACK-21 instrument.
For a specific future addition, an important area would be the introduction of similar longitudinal methods for in-service teachers, to study whether in-service teachers' TPACK changes over time.
The results outlining changes in TPACK areas and in differences between TPACK areas led to further questions concerning the factors affecting the changes and differences betweenTPACK elements. These results call for further research into the effects of different courses on changes in pre-service teachers' TPACK. Several studies have been conducted on the effects of different courses on technology in education (e.g., Doering, Veletsianos, Scharber, & Miller, 2009;Graham, Borup, & Smith, 2012). However, instead of focusing specifically on educational technology, we consider it important to study the effects of other "normal" courses in teacher education, such as those focusing on educational science, educational psychology, and learning ethics. It would be beneficial to identify what areas pre-service teachers specifically "grasp" within these courses taught using normal technology in pedagogically meaningful ways, what areas pre-service teachers pay attention to, and how they affect TPACK. This would provide important information for designing teacher education courses and curricula to better meet the needs of pre-service teachers in developing TPACK.
In addition, it is important to outline the differences amongst preservice teachers based on their TPACK assessments and what kinds of subgroups we can identify based on their TPACK. This would again provide us with more information for seeking ways to support the development of pre-service teachers' TPACK and meeting the needs of pre-service teachers within different TPACK profiles. Finally, the current study is limited to the data gained using the TPACK-21 selfassessment questionnaire. An important aspect for future investigation would be to expand the methods used for assessing TPACK (Krauskopf & Forssell, 2018) and to deepen these results using qualitative methods, such as lesson plans and interviews that would provide deeper insight into the nature and development of pre-service teachers' TPACK.

| CONCLUSION
The key conclusions arising from this study are (a) that teachertraining institutions in Finland have a beneficial impact on the development of pre-service teachers' TPACK and (b) that the TPACK model has potential as a vehicle for improvement in the complex process of preparing future students for ICT integration. Specifically, this longitudinal approach provides an insight into the progress of the different TPACK areas and possible changes occurring in relation to each other.