1 Context

The crisis of COVID-19 brought important changes in education globally, including in higher-level education (Cahapay, 2020). Social distancing and other health measures led to an increase in the use of online platforms and hybrid learning environments (HLEs) (Bonfield et al., 2020). Some of these HLEs were adopted by certain institutions due to their benefits and their adoption continued even after the pandemic (Maddukelleng et al., 2023). The learning situations where students are distributed both onsite (where the teacher is also located) and remotely to attend a class or activity simultaneously can be referred to as synchronous HLEs or synchronous hybrid learning environments (SHLEs) (Raes et al., 2020). SHLEs offer students the opportunity to participate in real-time classes, both onsite and online, which increases the flexibility of where the student can learn from and access to educational resources from any location (Bülow, 2022). However, SHLEs present significant challenges that were observed during the pandemic (Carruana Martín et al., 2021). Among these challenges is the deployment of these environments, which involves substantial investments in infrastructure, teacher training, and technology (Raes et al., 2020). Although SHLEs have benefits, additional research is necessary to fully grasp their effects on student learning and achievement, and to determine optimal implementation practices (Carrruana Martín et al., 2022).

There are different models of SHLEs and all of them have in common the on-site part which takes place in a classroom. The online part has several alternatives: the students can connect from home, from a dedicated place (e.g., the library), from another classroom or combinations of the above. Some models had a semi-online component, such as the mirror classrooms (Carrillo-Peña et al., 2021). In this latter case, the teacher conducts the lesson in one room. This room is projected into nearby rooms the teacher can go to physically if necessary.

An alternative to this model is the use of telepresence (Kasuk, T., & Virkus, 2024; Handbook, 2010). The technology most commonly used in the literature for implementing telepresence is robotics (Wernbacher te al., 2022; Lei et al., 2022). This telepresence is achieved by substituting the teacher and/or one or several students with a robot remotely controlled by them (Cha et al., 2017). This substitution aims to mimic the real presence of the person to enable communication akin to face-to-face interaction. Another telepresence model utilises augmented reality for holoportation (Abdelfatah, 2016). This model replaces physical presence with augmented reality projections of the body and its movements. This model also seeks to replace online communication channels with an experience more akin to in-person interaction. Both models face significant barriers, including high costs and the need to implement technologies not commonly found in educational institutions. A more affordable alternative to these models is the telepresence classroom or multi-location classroom, as mentioned in some literature (Handbook, 2010). The telepresence classroom is a concept based on distributed physical space that allows students to be physically located in two distant telepresence classrooms (e.g., on different campuses) connected by specific technology; online students can also connect, allowing for multiple configurations (Nenonen et al., 2019). Therefore, some onsite students will be in the telepresence classroom with the teacher, some virtual students will be in another telepresence classroom (without the teacher), and online students could be located anywhere. The design and technology of a telepresence classroom may vary depending on the educational institution and its resources, but in general, these spaces have technologies that focus on providing quality and reliable communication and collaboration among participants (Nenonen et al., 2019).

Telepresence classrooms are a model that can support SHLEs and therefore share several of their characteristics. One of the most important characteristics is the implementation and use of technology in this environment. The literature has already addressed this issue in on-site, virtual and online environments, but there is very little work on SHLEs. Some papers address the challenges of various environments using Smart Learning Environments (SLEs) (Carruana Martín et al., 2019), some of these challenges are also present in SHLEs (Carruana Martín et al., 2023). In SLEs, technology performs three basic functions: sensing (obtaining data such as positioning or audio), analysing (processing that data), and reacting (using that data to support pedagogical activities) (Tabuenca et al., 2021). Nevertheless, SLEs have certain drawbacks, such as high implementation costs due to the technology and resources required, potential technical problems that can disrupt learning and cause frustration, and concerns about privacy and security of personal data (Carrruana Martín et al., 2022).

The implementation of teaching and learning activities in telepresence classrooms also presents certain advantageous characteristics. Precisely because of its characteristics, complex scenarios can be initiated, harnessing the potential of the telepresence classroom. One such characteristics is the ability to work in groups with students who are either co-located or in different places. Therefore, it is essential to assess whether collaborative learning situations can be implemented in telepresence classrooms and what the effect of these situations might be on teachers and students (Echeverria et al., 2019). Among these collaborative learning situations, some are more complex to implement, such as those employing Collaborative Learning Flow Pattern (CLFP) (Hernández-Leo et al., 2006). One of these CLFPs is the jigsaw pattern (Carruana Martín et al., 2022), although there are others, such as pyramid, think-pair-share, etc.

Efficient communication, thorough planning, social dynamics, and individual student needs are crucial aspects of student collaboration (Echeverria et al., 2019). Collaborative learning, when well-organized, can foster teamwork and enhance the learning experience for students. However, collaborative learning also demands meticulous planning and coordination by teachers to be effective (Herrera-Pavo, 2021). A heavy workload can affect the effectiveness of collaborative learning; this workload can be influenced by various elements, including group size, activity type, and the experience of the teachers with this pedagogical approach used (Al-Samarraie & Saeed, 2018). This workload refers to the effort needed by the teacher and other stakeholders to conduct the intended activities (Prieto et al., 2015). Thus, proper training and support for teachers are essential (Hämäläinen & Oksanen, 2012). Technology can also aid collaborative learning through tools and resources for communication (Shen & Ho, 2020).

Another challenge of SSHLEs in general, and of telepresence classrooms in particular is the effect that the presence of technology can have on teacher agency (Kayi-Aydar, 2015). Teacher agency comprises various factors like experiences, professional training, culture, resources, environment, and social structure, which impact the decision-making process of a teacher (Biesta et al., 2015). Consequently, any constraints in a teacher agency can not only impair their capacity to make effective decisions but could also adversely impact the performance of student (Sammons et al., 2007). Given the issues previously discussed, it is crucial to analyse the telepresence classroom thoughtfully and strategically, considering the technological advantages as well as the implications for teacher agency and the learning process.

This research seeks to examine the factors impacting workload and teacher agency within collaborative learning in telepresence classrooms that support SSHLEs. To achieve this, two research questions are proposed:

  • RQ1: What factors influence teacher and students workload in a telepresence classroom during collaborative learning situations?

  • RQ2: What factors influence teacher agency in a telepresence classroom during collaborative learning situations?

2 Methods

2.1 Design

Two experiments centred on the use of the telepresence classroom were designed incorporating collaborative learning situations, specifically using a CLFP known as jigsaw (Hernández-Leo et al., 2006). This pattern has been customised in the case of hybrid scenarios in (Carruana Martín et al., 2022), these experiments aim to quantify the workload and teacher agency in telepresence classroom. The jigsaw model divides a topic into subtopics, designating students to master individual subtopics, and then groups them back together to share their expertise. This model comprises three Jigsaw Phases (JP), as described below. These three phases will be implemented in the two above mentioned experiments.

  1. 1.

    In the initial phase (JP1), the teacher selects a topic, divides it into 3 subtopics, and assigns each student a different subtopic, ensuring an even distribution of students per subtopic. Subsequently, students study their respective subtopic and undertake a test to validate their knowledge. This phase is carried out by the students outside of the course’s scheduled hours.

  2. 2.

    During the expert phase (JP2), students convene in groups according to their subtopics, potentially with multiple groups addressing the same subtopic. In this phase, all group members operate within a singular environment, either online or with peers who are in the same telepresence classroom. Groups tackle challenges pertinent to their subtopic.

  3. 3.

    The concluding jigsaw phase (JP3) sees the assembly of diverse groups, each including experts from every subtopic. These groups are composed of a mix of students from different telepresence classrooms, and some of the groups may also include online students. They confront challenges necessitating an understanding of all subtopics.

2.2 Data collection

Data collection for this study utilised multiple sources. Primary sources encompass logs from various applications deployed in the experiments, such as the online learning platform Engageli (Engageli Software, n.d.), coupled with the recording and transcription of lessons. These resources record the frequency and timing of teacher engagement with both onsite and online students. The interplay among teachers, students, and technology was analysed of the above data using Epistemic Network Analysis (ENA) (Csanadi et al., 2018). ENA facilitates the visual representation of connections between codes in flow data through dynamic network models. The coding has been equivalent to that used by the authors in previous studies (Carruana Martín et al., 2023) to facilitate comparisons with the results of these studies. Additionally, codes pertaining to the classroom where the teacher is not present have been added, which has been termed as ‘virtual’. These are in addition to the previously established codes of ‘onsite’, for students in the classroom with the teacher, and ‘online’, for students connecting from a different location. Activities carried out by the teacher associated with each code are documented in Table 1. Furthermore, a questionnaire based on the NASA-TLX model (Hart & Staveland, 1988) was utilised to measure workload (see Appendix A and Appendix B). This questionnaire comprises six questions rated on a scale from 1 to 100, 15 pairwise comparison questions to discern variation, demographic questions, and queries related to the activity to support correlation. Additionally, the teacher agency questionnaire, modelled after the research by Hull et al. (2021), one of the seminal works on teacher agency, was employed. The primary intent of the teacher agency questionnaire is to contrast teachers’ perceptions of their agency pre and post telepresence classroom implementation. The questionnaire consists of 17 questions, delves into various facets of teacher agency, asking teachers to express agreement on a 1 to 5 scale. The phrasing of questions, positive or negative, means a score of 5 could indicate heightened agency or its converse, depending on the context (this questionnaire can be found in the Appendix C). Finally, teacher interviews were conducted as part of the data collection process. These interviews sought to capture nuanced insights not apparent from questionnaires and enriched the understanding of teachers’ perceptions regarding workload and agency. The structure of this interview draws upon evaluation principles proposed by Stake & Jorrín-Abellán (2009).

Table 1 Codes of teacher actions for the ENA model
Full size table

The teacher agency questionnaire is another source of data and was administered both prior to the commencement of the jigsaw activity and upon its conclusion. Interviews were conducted either before the initiation of the jigsaw activity or after its completion. The recording and transcription of the class, as well as the accumulation of logs, took place during the execution of the jigsaw activity. Conclusively, at the termination of the jigsaw activity, both students and the teacher (with one exception in a specific experiment) filled out the workload questionnaire. The structured deployment of these sources throughout the experiments is delineated in Fig. 1.

Fig. 1
figure 1

The organisation of the data sources from teacher and students

Full size image

2.3 Experiments

Universidad Carlos III de Madrid has three telepresence classrooms across different campuses. In two of them, located in municipalities that are 3.74 km away from each other, called Leganés and Getafe, an introduction to programming course was delivered over an entire term. The connection between these two classrooms can be seen in Fig. 2. This course consisted of 19 one-hour sessions, of which 8 were conducted entirely online, while 11 were held in the telepresence classrooms. The sessions in the telepresence classrooms alternated the presence of the teacher between the two rooms. The students attending this course were divided into two groups: students from the Leganés campus and students from the Getafe campus. Within this context, two experiments were conducted.

Fig. 2
figure 2

Telepresence classroom at the Universidad Carlos III de Madrid

Full size image

The first experiment was conducted at Universidad Carlos III de Madrid in its telepresence classrooms of Leganés and Getafe. The aim of this experiment was to study the impact on both the teacher and the students of utilising the telepresence classroom as SSHLE. To achieve this, the experiment was conducted during one of the course sessions in the telepresence classroom. The class featured the teacher’s presence in the Leganés classroom, which had 10 participants; the Getafe classroom also had 10 participants, and there were 2 participants online. In JP2, two gropus of two students were formed, three groups of three students, along with two groups of four students. In the JP3 phase, four groups of four students were formed. Additionally, two further groups of three students were established (two in Leganés and one in Getafe) to explore one of the possibilities offered by the telepresence classroom. They were positioned side to side with the projection and instructed to attempt working as if they were in an onsite group setting, as this positioning simulated being in the same classroom. To support this SSHLE, the telepresence classroom synchronisation, that is managed by the company Streamplay was employed, and Engageli was used for the collaborative aspect (Engageli Software, n.d.). Engageli facilitates interaction between teachers and students, offering supports collaborative workspaces, virtual tables, and group resource management. Prior to initiating the experimental design, the teacher filled out an agency questionnaire. An initial interview was conducted with the teacher, followed by the execution of a jigsaw activity focused on user-centred design. Data for the activity was collected from the Engageli session recordings and transcript of an observer in the classroom. Subsequently, both the teacher and the students completed the workload questionnaire, and the teacher revisited the agency questionnaire to provide post-experiment insights.

The second experiment was conducted in the same telepresence classroom. The aim of this experiment was to replicate the previous experiment while addressing issues that had been identified in the implementation of the Jigsaw pattern in the prior study. This experiment sought to analyse a real session with a teacher better prepared to manage the collaborative activity, as well as students who were also more familiar with such activities. To this end, the experiment was conducted during one of the course sessions in the telepresence classroom. The teacher was in Getafe telepresence classroom, which had 10 participants; the Leganés telepresence classroom had 11 participants, and 3 participants were online. Engageli was also employed to support this SSHLE, in the same manner as in the previous experiment. Data collection was the same as in the previous experiment. The two experiments conducted are summarised in Table 2.

Table 2 Detail of the two experiments conducted in the telepresence classroom
Full size table

3 Results

3.1 Experiment 1

The teacher’s responses to the NASA-TLX questionnaire suggested that mental demands and effort were the predominant factors influencing workload, with scores of 80 and 90 on their subscales, respectively. In particular, effort emerged as the most prominent factor amongst the variables in pairwise comparisons, standing out in all five. Further variations and subscale data are detailed in Table 3. The teacher’s overall workload was positioned at 65.33, on a scale from 0 to 100, placing it in the high workload bracket (60-80) (Hart & Staveland, 1988).

Table 3 Experiment 1 - NASA-TLX teacher scores (subscales in the range between 0 and 100 and pairwise comparisons in the range between 0 and 5)
Full size table

As for the number of participants, this experiment involved 22 students in a real scenario. Some authors establish a minimum of 20 participants in studies dedicated to detect most of the characteristics and problems associated with the use of technologies (Alroobaea & Mayhew, 2014). The number of students in this experiment allows obtaining results to cover most of the characteristics, although it would be desirable to compare the results obtained here with equivalent experiments in other real scenarios in which the telepresence classroom is used.

Regarding the student workload results, it is important to note a setback: for the hybrid student groups (in this case only with onsite and virtual students), who were instructed to position themselves in front of the projection and attempt to interact as an on-site group, they were unable to do so. This was due to communication issues (among which background noise and their own voices being too quiet were prominently mentioned), leading them to operate like the other hybrid groups. For this reason, groups with these characteristics were not formed in the second experiment, and are not referenced in the results. Pertaining to the student results (refer to Table 4), it can be observed that the mental demand is the highest. This is attributed to the challenges of coordinating with peers who were situated in a different environment during JP3. This issue arose because conversations from other groups filtered through the telepresence classroom’s microphones, hindering communication within each group and creating noise in both classes. To mitigate this, they had to speak more softly and prioritise text communication. Effort is the next highest value, and this, according to the students, is due to the complexity of the exercises and the limited time available to complete them.

Table 4 Experiment 1 - NASA-TLX student scores
Full size table

The ENA model displays the frequency of actions undertaken by the teacher and the correlations between these actions (see Fig. 3). In this experiment, the most frequent actions of the teacher were class announcements, tool usage (telepresence classroom support software or Engageli), and interactions with groups in their classroom. Furthermore, the strongest correlations occurred between class interactions and class announcements. Another robust correlation was between tool usage and monitoring the class status. The last notable correlation identified was between observing the class status and interacting with onsite groups. Additionally, it is worth highlighting the infrequent interactions with online students.

Fig. 3
figure 3

Experiment 1 - ENA model (the thickness of the lines corresponds to the number of times there was a transition from one action to another, and the size of the points corresponds to the number of times an action was performed)

Full size image

The results from the questionnaire regarding teacher agency revealed that 2 factors (11.76%) increased, 2 factors (11.76%) decreased, and 13 factors (76.47%) remained the same following the experiment. The factors that saw an increase pertained to the impact of their decisions on student learning and decision-making. Those that decreased were related to the capacity for analysis concerning the tools employed and the control they have over the class. These findings suggest that there were minimal changes in teacher agency, as approximately three-quarters of the factors remained unchanged. Furthermore, the factors that did alter shifted by only a single point, indicating a minimal overall impact.

3.2 Experiment 2

The responses of the teacher to the NASA-TLX questionnaire suggested that temporal demands and effort were the predominant factors influencing workload, with scores of 90 in both sub-scales. Notably, mental demand emerged as the most prominent factor amongst the most significant variations in pairwise comparisons, standing out in all five. Table 5 details other variations and the data from the sub-scales. The overall workload for the teacher was positioned at 67.33, on a scale from 0 to 100, placing it within the high workload range (60-80) (Hart & Staveland, 1988).

Table 5 Experiment 2 - NASA-TLX teacher scores (subscales in the range between 0 and 100, and pairwise comparisons in the range between 0 and 5)
Full size table
Table 6 Experiment 2 - NASA-TLX student scores
Full size table

24 students participated in the second experiment. As in the first experiment, this number is higher than the number indicated as sufficient in some studies to detect most of the characteristics and problems associated with the use of technologies (Alroobaea & Mayhew, 2014).

The students’ workload results can be viewed in Table 6. Once again, mental demand is the highest. This can be attributed, as in the previous experiment, to the coordination challenges with peers who were located in a different environment during JP3. The high value of this demand was also due to noise and, in one case, the lack of collaboration of one of the group members. This lack of collaboration is attributed by the rest of the group on the fact that the student in question was in the telepresence classroom, where the teacher was not present to supervise that student. The other highest sub-scale is again effort. On this occasion, students solely attributed it to a lack of time.

The ENA model of Experiment 2 can be seen in Fig. 4. The most frequent actions of the instructor were tool usage and class observation. Moreover, the strongest correlations occurred between class observation and interaction with the hybrid groups. Another significant correlation was found between class observation and interaction with the onsite groups. The last noteworthy correlation identified was between announcements to the class and interaction with it. Furthermore, it is worth noting the infrequency of individual interactions with the virtual students. This was attributed to the fact that the students from Leganés had a deeper understanding of the subject, and their uncertainties arose more as a group than on an individual basis.

Fig. 4
figure 4

Experiment 2 - ENA model (the thickness of the lines corresponds to the number of times there was a transition from one action to another, and the size of the points corresponds to the number of times an action was performed)

Full size image

The results of the questionnaire regarding teacher agency revealed that three (17.65%) factors increased, one (5.88%) decreased, and 13 (76.47%) remained the same after the experiment. The factors that increased pertained to the ability to analyse the tools employed, the use of the tool, and the impact of their decisions on student learning. The factor that decreased was related to the integration of other materials into the tools used. These findings suggest that only minimal changes occurred in teacher agency, as approximately three-quarters of the factors remained unchanged. Furthermore, the factors that did alter only shifted by a single point, indicating a minimal overall impact. However, it can be positively noted that the teacher’s capacity for analysing the tools employed improved compared to the previous experiment.

4 Discussion

This paper conducted two experiments using the telepresence classroom in collaborative learning situations to address the proposed research questions. The research question “What factors influence teacher and students workload in a telepresence classroom during collaborative learning situations?” was analysed using the NASA-TLX questionnaire, the ENA model, and was complemented by interviews with the teacher.

The NASA-TXL questionnaire indicates a teacher workload of 65.33 and 67.33 in each experiment, respectively. Additionally, this questionnaire assists in identifying the various factors affecting the workload, further supported by the interviews. One of the primary factors resulting in such high workload for the teacher was the significant effort required for coordination and attention to the telepresence classroom in which the teacher was not present. While in the first experiment the teacher was limited by his effort, in the second experiment time was more important. One issue that impacted the teacher workload was the noise generated during JP2 and JP3, as conversations amongst group members, when too loud, were picked up by the microphones and transmitted to the other telepresence classroom. This is a well-known problem in the SHLEs (Carrruana Martín et al., 2022). Another challenge was related to some technical issues encountered at the beginning of the lessons when connecting the classrooms; these technical issues consumed part of the already limited available time.

In the study by Carruana Martín et al. (2023), an analysis of workload using the NASA-TLX was conducted in another type of SSHLE during collaborative learning situations. In that paper, they achieved a workload score of 50 (out of 100) when collaborative patterns were not utilised, but this increased to 60.67 and 76 (out of 100) when they were implemented. These results are analogous to those obtained in this paper, with all falling within the 60-80 range, which is considered a high workload (Hart & Staveland, 1988). This might suggest that collaborative learning situations in SSHLEs entail a high workload for teachers, even in purpose-built spaces such as telepresence classrooms. However, there are studies (Crespi et al., 2022) suggesting that the use of CLFP can increase the workload and stress. Nonetheless, there are too few examples available to definitively ascertain which conclusion is accurate.

Continuing with the workload analysis, the average NASA-TLX score for the students was 49.03 and 58.65. These results fall within a medium workload range (40-60) (Hart & Staveland, 1988). Reference is again made to the findings of the study by Carruana Martín et al. (2023) to analyse these outcomes. In that paper, the students reported workloads of 50.66 and 50.94. These results are comparable to those obtained in this paper and also lie within the same range (40-60), denoting a medium workload (Hart & Staveland, 1988). This might suggest that collaborative learning situations in SSHLEs result in a medium workload for students, even in spaces specifically designed for this purpose, such as telepresence classrooms. Another study (Malschützky et al., 2023) examining the mental workload of students in various environments (videoconferencing, audio, and chat) while engaging in collaborative situations indicates that there is scarcely any significant difference in workload when collaborating across these diverse settings. Nevertheless, there are very few examples to confidently assert any of these conclusions.

Fig. 5
figure 5

ENA - Comparison between ENA Model of Experiment 1 (red) and Experiment 2 (blue)

Full size image

Firstly, the two experiments conducted in this study are compared for the analysis of the ENA model (Fig. 5). Notably in this comparison, there is a significant relationship between the classroom announcements and the interaction with onsite groups from the first experiment. This is because, during Experiment 1, after making an announcement in JP2, the groups in the classroom where the teacher was located tended to ask him questions. In contrast, in Experiment 2, the relationship between class observation and interaction with hybrid groups stands out. This was due to students facing more challenges in JP3, prompting the teacher to monitor the overall state of the class, assisting the groups he perceived as struggling the most. Generally, the activity most frequently undertaken by the teacher in both experiments was class observation. Since the telepresence classroom simulates a complete class environment, it was more convenient for the teacher to undertake a general observation than to resort to a tool. Thus, the ability to comfortably observe the status of the class emerged as a key factor influencing workload.

When comparing the ENA models obtained in this paper with those found in the related literature (Carruana Martín et al., 2023) the significance of observing the class status becomes apparent. Moreover, the telepresence classroom stands out from the other SSHLE model in that tool usage is less frequent. Another noteworthy feature of the telepresence classroom is that the teacher tends to make more announcements and engage in more iterations with the class.

A comprehensive analysis of the data obtained from these two experiments, the NASA-TLX questionnaire, the ENA model, and the interviews can help identify key factors influencing workload. One such factor was the available time. This issue has already been identified in environments with a strong technological presence (Prieto et al., 2015) and in similar settings with SSHLE during collaborative learning situations (Carruana Martín et al., 2023). This factor is significant because it is compromised whenever a problem arises. Unlike other environments, quick alternatives are usually lacking, especially in the face of issues like internet connection disruptions. This is particularly relevant in cases involving technical issues due to the inherent dependency of SSHLE on technology. Moreover, while calculating the duration of a collaborative activity is already a complex task (Saputra et al., 2019), the necessary reliance on technology to facilitate collaborative learning situations only adds to the complexity. This factor was discerned from the NASA-TLX questionnaires completed by both the students and the teacher and was further corroborated through interviews with the teacher. In these interviews, the teacher identified that having alternative solutions to some of the more common failures could prevent certain time losses. The teacher also emphasised that being able to view the majority of the students as a single classroom saved him time.

Another identified factor was the high level of effort required from both the teacher and the students. In such scenarios, the teacher often faces heightened effort due to the intricacies of conducting the class in conjunction with managing the technology (Carruana Martín et al., 2023). However, it is particularly notable that students also report such a high level of effort. Furthermore, this factor becomes crucial because if the telepresence classroom imposes undue strain on its participants, it presents an urgent issue that requires resolution. Students justified these figures by referencing the challenges posed by background noise and communication issues. This factor was derived from the NASA-TLX questionnaires of both the students and the teacher, and was further corroborated through interviews with the teacher. A suggestion made by the students to mitigate this challenge involved muting the communication between telepresence classrooms during specific moments and the use of headphones by group members. As for the teacher, recommended solutions included support in group management and enhanced adaptability of tools to manage sound distribution between classrooms and groups.

Another factor identified that impacted the workload was class observation. This factor has already been recognised in literature related to SLEs (Tabuenca et al., 2021), in studies focused on collaborative learning (Amarasinghe et al., 2021), and in scenarios similar to those covered in this paper (Carruana Martín et al., 2023). This factor becomes significant as the teacher needs to understand the class status, its progress, or whether any student or group requires assistance. The challenge this factor presents in the telepresence classroom is determining the status of students located in different environments from the teacher. This factor was identified from the ENA models and further corroborated through interviews with the teacher. In these interviews, the teacher emphasised that the feature simulating two distant classrooms appearing as one greatly simplified this task. Moreover, the teacher pointed out that the Engageli tool provided the information that might be obtained at a mere glance, such as the status of online students.

The final factor identified was interaction with hybrid groups. This factor is unique to SHLEs (Raes et al., 2020), and thus also to SSHLEs. In the telepresence classroom, this becomes more challenging because, in addition to having onsite and online students, there are virtual students (those located in another telepresence classroom different from where the teacher is physically situated). Moreover, this factor becomes crucial if one wishes to achieve effective collaborative learning. This factor was identified from the ENA models and confirmed through interviews with the teacher. In these interviews, the teacher emphasised that being able to interact with groups in an “onsite” manner (being able to physically approach a group and engage with all its members, regardless of their environment) simplified this aspect. Additionally, the teacher indicated that the Engageli tool also supported this factor through its group management tools. A summary of all the identified factors can be seen in Table 7.

Table 7 RQ1: Factors influence the workload of the teacher and students in telepresence classroom during collaborative learning situations
Full size table

The second research question, “What factors influence teacher agency in a telepresence classroom during collaborative learning situations?”, was analysed using the teacher agency questionnaire and was complemented by an interview with the teacher.

The Teacher Agency questionnaire appears to indicate that the telepresence classroom, in collaborative learning situations, has little impact on teacher agency. However, certain specific factors were affected, suggesting that a targeted approach might enhance them. In Experiment 1, the factors that increased were decisions related to student learning and decision-making. Conversely, the factors that decreased in Experiment 1 were analysis concerning the tools used and the control exerted over the class. Experiment 2 presented a particular tendency. One of the factors that declined in Experiment 1 saw an increase in Experiment 2: the analysis related to the tools used. The other two factors that increased in Experiment 2 were the tool usage and decisions in student learning - the latter also increased in Experiment 1. On the other hand, the factor that decreased in Experiment 2 was the integration of other materials into the tools used. When comparing these results with related literature (Carruana Martín et al., 2023), it is apparent that the factor of tool usage also increased, suggesting that SSHLEs, such as the telepresence classroom, might indeed bolster this factor.

The factors identified in the questionnaire were cross-referenced with the interviews conducted with the teacher. One of the factors mentioned in the interview, and which appears in the questionnaire, was decision-making and the difficulty in making decisions when lacking experience. This factor increased in both experiments, and the teacher’s perception suggests it is due to having experience in similar situations. In the paper (Carruana Martín et al., 2023), this factor also emerged but was more associated with the ability to make decisions rather than the efficiency of those decisions. The teacher highlighted previous experience and training as characteristics that support the improvement of this factor. Another factor mentioned in the interview was the analysis of the tools and their overall usage. This factor becomes crucial when relying on these tools to conduct the class. Furthermore, this factor has already been identified in the related literature (Carruana Martín et al., 2023; Li & Ruppar, 2021). The literature also suggests characteristics that can support this factor. One such characteristic is the need for the tools to adapt to different situations. The teacher also emphasised this feature in the interview. Another characteristic highlighted by the teacher is the potential for integration with other tools.

Another factor mentioned in the teacher’s interview, was control over the class. This factor decreased in the first experiment but was unaffected in the second. The teacher had control at all times, except when implementing the jigsaw pattern. The interview revealed that in experiment 1, the teacher had less control because the groups with the most queries were in the telepresence classroom where the teacher was not present. This created communication challenges, and it took the teacher longer to address concerns. One of the features that can support this factor, as indicated by the teacher, is the enhancement of communication tools. Furthermore, the related literature (Carruana Martín et al., 2023) points to adaptability as a characteristic that also bolsters this factor. A summary of all the identified factors can be seen in Table 8.

Table 8 RQ2: Factors influence teacher agency in telepresence classroom during collaborative learning situations
Full size table

5 Conclusion

This paper examined factors affecting the workload of students and the teacher, as well as the teacher agency in telepresence classrooms during collaborative learning situations, were examined. The analysis was based on two experiments in which the telepresence classroom was used to support the SSHLEs. From these experiments, influencing factors related to the workload were identified: available time, high level of effort, class observation, and interaction with hybrid groups. Moreover, factors concerning teacher agency were determined: decision-making, analysis and use of tools, and control over the class. The experiments also revealed characteristics to alleviate the workload. Some of these characteristics encompassed: providing alternatives to the most frequent challenges, tools that aid in group management, and the primary advantage of the telepresence classroom - the perception of two remote classrooms as a singular entity. Features to enhance teacher agency were also identified. These include offering prior training about the environment, ensuring the availability of adaptable tools for various scenarios, and tools that foster communication across differing environments. The main innovations of this work include a detailed comparison and analysis of a variant of SSHLE, which still needs further exploration as there are no studies on the matter. Although the telepresence classroom does not introduce groundbreaking technology, for instance, the mirror classrooms (Carrillo-Peña et al., 2021) feature similar yet more basic technology. However, the telepresence classroom offers greater immersion. Moreover, another innovation is the data derived from the two learning experiences in the telepresence classroom.

A primary limitation encountered during the development of this paper was the integration of the jigsaw pattern within the already established course syllabus. While a course might anticipate the implementation of collaborative learning situations, executing them and adapting to guidelines like the jigsaw pattern is intricately complex. This demands a significant time investment from the instructor and entails preparing additional material for the class. Technical challenges were another limitation. The synchronisation software for the telepresence classroom is managed by the company Streamplay. This company has a designated support schedule for technical issues. However, the challenge arose when classes were scheduled outside this window, necessitating that technical issues be addressed without the company’s support, leading to extended resolution times. Also, it is worth noting the limitation of noise during teamwork phases. Although, by the time of the second experiment, this was recognised, and students were requested to bring headphones, with the audio muted between classes at certain moments. The complication arose when the teacher needed to communicate with all students, at which point sound from both classrooms would filter back in, leading to brief periods of disruption. Nevertheless, it is pertinent to note that this issue was considerably less prevalent in the second experiment compared to the first. Regarding the study limitations, the experiments were conducted with a single teacher, which might introduce a certain bias. Additionally, the small number of students and the fact that it was carried out within a single subject stand as other significant limitations.

Future work shall consider the introduction of the Jigsaw pattern from the outset of the subject’s planning to facilitate its use. Another line of future work would be to define alternatives for major issues such as network connection problems, lack of student materials, or when the telepresence classroom synchronisation software malfunctions. In addition, the analysis and design of a better layout and new tools to reduce noise that adversely impacts collaboration should also be considered. Finally, more experiments should be conducted in different SSHLEs with various teachers, distinct subjects, and a larger student body, given the limited experiments carried out thus far.