Model for the Analysis of Social Regulation and Collaboration during the Development of Group Tasks

Abstract

This paper presents a model for the analysis and characterization of social regulation during collaborative task development. The structural part of the model is composed of three components which give rise to the generation of four phases of group interaction. The combination of these phases of group interaction and their components allows us to know whether or not a given group during the execution of tasks manages to develop mechanisms of collaboration and socially shared regulation. As for the conceptual section, the model is supported by three components that deal with task regulation, communication regulation, and collaborative work. Each of these components in turn presents aspects that can be identified in the interaction of the groups during the development of a task. The model was applied with the participation of five work groups made up of graduate students. The type of study was a descriptive quantitative approach. The results made it possible to corroborate the functionality of the model based on the identification of recurrences of events in the phases reached by each of the groups during the development of shared tasks, as well as aspects of collaboration and social regulation during the execution of group tasks in socially and collaboratively regulated learning processes, according to the analysis of the interactions recorded by the groups.

1. Introduction

Self-regulation is one of the factors related to learning and academic behavior that has shown the greatest importance over the years [1,2], and its significance lies in its absolute necessity for carrying out intentional and complex learning [3]. However, in recent decades, there has been an increasing interest in exploring the learning regulation process beyond an individual’s intrinsic perspective and independent content management. This interest focuses on how regulation occurs in group interactions and how groups socially regulate themselves during the development of collaborative tasks aimed at the shared construction of knowledge [4,5].

For authors such as Baker et al. [6], Baker [7], Chan [8], Gourgey [9], Järvelä [10], and Zhao and Zheng [11], the study of social regulation from a collaborative perspective involves a shift from the usual approach to studying regulatory learning processes. It moves from an individual-centered view focused on content management to an interaction that plays a prominent role among group members, ultimately revealing how communication and task execution are regulated and perceived at the group level.

From this perspective, the study of social regulation in collaborative learning processes demands not so much a focus on intra-subjective processes and content control but rather distinguishing episodes of communicative interaction that involve opinions, reactions, and interactions among different individuals related to the task [12]. This includes the perception of its management and handling, as well as the collaborative process itself, whether formed or not within the group during the development of a specific group activity [13].

While the aforementioned aspects have already been the subject of study and report in various investigations [14,15,16,17,18], to date, a general model has not been defined that includes research from recent years on social regulation and constitutes, in turn, a kind of synthesis of these studies on social regulation and collaboration. Such a model would especially serve as a basis for characterizing and analyzing social regulation and the shared construction of knowledge in the development of collaborative tasks systematically and relatively simply [19,20,21]. Accordingly, the objective of this article is to present a model for the analysis and characterization of social regulation and collaboration in postgraduate academic groups.

1.1. Background on Other Models

Although there are classic models that talk about the collaborative process and group dynamics, such as that proposed by Bruce Tuckman in 1965 [22] (based on five stages: formation, conflict, normalization, performance, and dissolution) and Douglas McGregor’s [23] model of the same year which deals with human nature at work and beliefs and management, known as theory X and theory Y, the truth is that in general terms and according to the tradition that covers research on the social regulation of learning, and specifically the study of collaboration as a process derived from social regulation, the proposal of models is a little more recent, with the oldest model found in 1998 and the most recent in 2013. One of these models is the one proposed by Wenger [24]. In this model, the author focuses on exploring learning as a social and contextualized process with communities of practice or shared regulated communities at its core.

Another model was proposed by Salomon and Perkins [25], which sought to account for the role of social mediation in learning processes. This model is based on the premise that social interactions and cultural tools are the basis for the study of the regulation of collaborative learning. Vauras et al. [26] propose a model of social regulation of learning. In this model, they particularly focus on the processes of social regulation of learning, analyzing how individuals within a group regulate their learning processes through social interactions. Another model proposed is that of Barron [27].

In this model, the author focuses on the analysis of the regulated behavior of individuals in relation to a given collaborative activity. He makes special reference to aspects related to communication, level of commitment, and participation. In 2006, another model came to light that includes the variable of social regulation: the model of emotional and social regulation of learning [28]. This model places emotions as the object of analysis in collaborative processes. The premise of Pekrun’s (2006) [28] model focuses on showing how both individual and group emotions influence group dynamics of learning and collaboration. Volet [29] and Thuman [22] propose another model in which the processes of synchrony in social regulation are explored. This model assumes that social regulation focuses on the alignment of efforts by the members of a given work group to achieve a collaborative learning process.

In more recent years, Hadwin and Järvelä [30] proposed a model in which individual regulation, shared regulation of cognition, and co-regulated processes of learning are accounted for. Another similar model is the proposed model in which they focus on the analysis of co-regulated learning processes. This model focuses on the study of co-regulation as a supportive phenomenon in collaborative processes.

From the above models, aspects related to cognition as a collaborative process were taken from the model of Grau and Whitebread [31], task regulation from the model of Hadwin and Järvelä [30] and Winne and Hadwin [32], and communication from the models of Järvelä and Hadwin [33] and Janssen [34]. The structure of the model is detailed below [5].

1.2. About the Proposed Model and Its Structure

The model presented here is based on considering the group exercise of task solving as the fruit that arises from the interaction that occurs between individuals in a regulated manner and is associated with collaborative processes of shared knowledge construction [5], aspects of social regulation of the task [32], aspects of social regulation of communication [34], and aspects of monitoring that the group generates during the collaborative task-solving process, taken from Järvenoja [35] and Rogat and Linnenbrink [36].

1.3. Conceptual Section of the Model

In the conceptual section, the model is supported by three components: task regulation, communication regulation, and collaborative work [31,37]. Each of these components will be briefly defined below for better understanding.

Group Collaboration: This first component assumes the collaborative process as a shared construction of knowledge [31], specifically referring to the cognitive process of discussion and review of ideas that leads to the advancement of knowledge generated by a particular group during the joint development of a task. This greatly defines the collaborative process of interaction during the development of a group activity [9]. For its identification, collaboration consists of four episodes: initiation, exploration, negotiation, and generation [31,38].

Group Regulation of Communication: The second component refers to a metacognitive aspect in group communication according to authors such as Winne and Hadwin [32], Järvelä and Hadwin [33], and Sobocinski [21]. They have researched the specific manner in which metacognitive interaction is generated among the members of a particular group and the understanding of four specific episodes: questions to the group, negotiation of objectives, adaptation of task perception, and adaptation of goals. Each of these episodes accounts for how the members of a given group strategically regulate their communication to achieve their objectives during the development of collaborative tasks [32].

Group Regulation of the Task: The third component of the model is concerned with group regulation of the task. According to Barenthien [39], Hadwin [40], and Janssen [34], this component is understood as the metacognitive interaction that arises among the various members of a group, facilitating the resolution of the task itself collectively. Task regulation comprises six episodes that enable the generation of conditions for both individual and distributed understanding and responsibility through roles among the group members. These episodes are judgments about the task, exchange and organization of information, goal planning, task execution, and reflection on the task [5].

1.4. Structural Section of the Model

In the structural aspect, the model is composed of three components and four phases. Each phase should be read vertically, as this arrangement reveals the hierarchical structure intended for its use. The components are arranged horizontally. The combination of phases and components allows for the identification of aspects of social regulation and collaboration within work groups.

To operationalize the model, one must start by individualizing the messages sent in group interactions during task development [41]. This individualization of messages is referred to as an episode of group interaction [14] and constitutes the components described in the conceptual section of the model [42].

1.5. Description of the Phases Comprising the Model of Social Regulation and Collaboration

The first phase consists of the combination of episodes from the three components described in the conceptual section of the model: initiation, questions to the group, judgments about the task, and information exchange [9,19]. The second phase comprises the combination of the following episodes: exploration, adaptation of task perception, and information organization [20,43]. The third phase includes the combination of the episodes of negotiation, negotiation of objectives, and goal planning [44,45]. The fourth phase consists of the episodes of generation, adaptation of goals, and task execution and reflection [15,46] (

Table 1. Phases, components, and episodes comprising the social regulation model.

Interaction Phases	Interaction Episodes in Collaboration	Interaction Episodes in Communication Regulation	Interaction Episodes in Task Regulation	Group monitoring
Phase 1	Initiation	Questions to the group	Judgments about the task and understanding of the task
Phase 2	Exploration	Adaptation to task perception	Information exchange and information organization
Phase 3	Negotiation	Negotiation of objectives	Goal planning
Phase 4	Generation	Adaptation of goals	Task execution and reflection

).

1.6. Inter-Coder Agreement (ICA)

The inter-coder agreement was carried out in four phases: a first phase of selection and training of coders that consisted of a selection sub-phase in which the coders were familiarized with the topic of study and the methodology from which the data collection would be carried out, followed by the training sub-phase in which the coders were given information on the theoretical basis, the research questions, and the coding system consisting of codes, categories, and definitions.

This last step was applied so that the coders could understand and apply the codes in a similar way. The second phase was the development of the coding scheme, which included the creation of the scheme based on the theoretical review of different existing models of social regulation and collaboration. Subsequently, a pilot test was generated in which the coders applied the scheme to a small set of transcript data to verify their applicability and comprehensibility. This was to check for discrepancies or doubts about the application of the scheme. This was followed by the third phase of independent coding, in which the coders independently applied the codes to a larger data set, avoiding the influence of the other coder. For the development of the fourth phase, inter-coder agreement was calculated by comparing the coding performed and measuring agreement by calculating Cohen’s Kappa index, which yielded a value of 60.

1.7. Observation Time of the Working Groups

The working time of the groups was carried out during the academic semester (8 weeks), during which time the groups had to meet every two weeks to discuss aspects of their research proposal, taking into account a thematic guide given by the professor.

Table 2. Working groups per session and recording times.

	Working Groups per Session and Recording Times
Session	Group One	Group Two	Group Three	Group Four	Group Five
1	50′,57″	21′,13″	22′,41″	45′,04″	4′,27″
2	42′,11″	11′,12″	25′,25″	41′,39″	60′,22″
3	60′,06″	12′,58″	19′,43″	17′,27″	14′,39″
4	180′,09″	13′,18″	12′,58″	11′,46″	69′,63″
Total time	332′,83″	58′,01″	79′,67″	115′,16″	148′,51″

below shows the list of working groups per session, and the recording times that gave rise to the data for this study.

1.8. Transcriptions of the Audio and Video Recordings

The recording of the synchronous meetings resulted in the generation of audio and video recordings. Each video transcript was compared with the recordings in order to correct possible transcription errors. A total of 1254 min and 16 s (20 h and 9 min of recording) was obtained, of which 734 min and 18 s of recording was analyzed, corresponding to a total of 12 h and 23 min of recording.

1.9. Procedure for the Identification of the Interaction Episodes of Regulation and Collaboration

The identification of interaction episodes was carried out on the basis of the analysis of each of the designed phases that make up the model on social regulation. Each episode of interaction in each of the components (collaboration, task regulation, and communication regulation) was carried out in a particular way in terms of the dialogs held between the different members of a work group during the development of collaborative activities. The protocol for the identification of the different interaction episodes is shown below. The protocol was adapted from Castellanos and Onrubia [36,47].

1.10. General Criteria for the Identification of Regulation Episodes

The beginning of an interaction episode is identified by the message fragment that triggers or activates a chain of contributions aimed at regulating the task, communication, or collaboration in the group. The end of an interaction episode is identified by the message fragment that no longer receives any contribution linked to the central theme manifested in the task resolution process, the way the group communicates, or the way the collaborative process is carried out.

1.11. Steps for the Identification of Regulatory Monitoring Episodes

Stage 1. Message fragmentation: Message fragments that characterize interaction episodes can be defined as parts of a message that refer to a certain thematic axis of the task or collaborative exercise and that have a complete meaning with respect to that axis in the context of the message in which they appear. A message may include one or more fragments of interaction episodes, so it is not unusual to find different interaction episodes in the same message or fragment.

Step 2. Criteria for message fragmentation:

−

Identify and distinguish the fragments that make up the same message while respecting the original structure of the turn.

−

Separate and assign a code to each message fragment according to the nature of the message fragment (interaction episode according to each component: collaboration, task regulation, and communication regulation).

−

Group the identified codes according to the nature of the fragments for analysis (shared knowledge construction, communication regulation, and task regulation).

−

Other considerations: For the interaction episodes, although they correspond to the axis through which the analysis of message fragments in a group communication is carried out, it should be noted that each fragment is part of a category of analysis linked to the interpersonal regulation of communication, task management, or the shared construction of knowledge itself.

−

When the content of a message begins with a word or sentence that refers to the group as a whole, the sentence is integrated into all the fragments that are on the same topic that is the focus of the group’s conversation.

−

When a fragment is composed of a sentence that refers to two or more different topics at the same time, the fragment may be part of two or more thematic axes.

−

When a message ends with a sentence whose content closes two or more thematically distinct fragments, the sentence is part of both fragments.

1.12. Coding of the Information Using MAXQDA Software

Once the transcriptions of all the sessions recorded by the collaborative groups had been made, they were exported to the licensed MAXQDA v24.5 software for mixed analysis. In this software, a separate folder was generated for each file related to each participating group.

Likewise, the categories described in the section on defining the categories of analysis of the research were generated. Once the file folder was generated for each collaborative group in a discriminative manner and the categories with their respective codes were created, the categorization and coding of each file were carried out, as illustrated in Figure 1, which refers to the categorization of data.

1.13. Educational Methodology for Data Collection

The formative activity in which the 1497 interactions of the five academic working groups were collected and which constitute the object of analysis of this article took place in the development of the research activities leading to the elaboration of the thesis project. This activity was divided into four deliverables: selection of the topic and problematization, background research and status of the topic, construction of the research question and objectives, and justification and methodological design. The working time was spread over eight weeks with synchronous group meetings every two weeks. After the meetings, each group had to send its recording to a shared folder on Google Drive. The group meetings were to be held via the Google Meet platform. In the recorded video, each group was to show the development of the academic activities. Each group was free to decide how long it would meet for. The only condition for each group was that each task took place during the re-recording and not before, in order to capture the interactions between the members of each group as naturally as possible.

1.14. Operationalization of the Model

The model, in its practical application, allows for the organization and analysis of the interaction episodes that occur among the members of a work group. The arrangement of the different interaction episodes that comprise the components is ordered hierarchically from top to bottom and inter-related to account for the phases that make up the model of social regulation and collaboration [10,47,48].

It is important to remember that the interaction episode, as the basic unit of the message, allows for the analysis of the collective messages of a given work group and their placement in a specific phase according to the component to which that episode corresponds [19,49]. In other words, interaction episodes are considered the basic unit of analysis, consisting of a sequence of verbal contributions among the various members of a work group during the development of collaborative tasks. Their duration is delimited by the framework that begins and ends with a message or contribution, or when contributions on the theme that generated the sequence of messages cease [50,51].

This operationalization of the model was carried out through the observation and description of aspects associated with social regulation and collaboration in five first-semester groups of a master’s program in education in the city of Manizales during a collaborative classroom exercise.

2. Methods

Based on a quantitative descriptive study in which dimension reduction techniques were applied in order to profile academic groups by means of principal component analysis with varimax rotation in IBMSPSS v.25, hierarchical cluster classification dendrograms with Euclidean distance and Ward’s linkage were applied to minimize the variance between groups, and covariances from the application of the Bias-Corrected bootstrap confidence interval (BCBCI) algorithm were calculated, we proceeded to the analysis of 1497 interaction episodes recorded in a time of 736′,11″ minutes by five groups of first-semester postgraduate students of a master’s degree in the private sector in the city of Manizales (Colombia).

The analysis was carried out in order to verify the functionality of a model of social regulation and collaboration during the development of group tasks.

2.1. Participants

Five groups of first-semester postgraduate students from a master’s degree in education at a private university in the city of Manizales (Colombia) participated. The working groups were formed during the first week of entry to their postgraduate studies (master’s degree in education at a private university in the city of Manizales). The formation of the groups was of free choice. The academic space in which the formation of the groups took place was the subject Research Seminar, in which the interests and topics that reflect the research needs and ideas of the participants, who are beginning their postgraduate training process, were initially explored. The free creation of the groups was carried out in order to check whether the heterogeneity in the groups influenced the collaborative and socially regulated process both at the level of tasks and at the level of communication.

2.2. Techniques and Instruments

Non-participant observation was used for data collection. To carry out the application of this technique, audio and video recordings were made of the activities carried out for five study groups. A total of approximately 736′,11″ min of recording was obtained. The interaction episodes were then coded according to component, phase, and frequency of occurrence. This process coding was carried out in the personal licensed software MAXQDA. The final product was a total of 1497 interaction episodes recorded by the five groups. With this information, we proceeded to establish and organize the following parts of data systematization:

• Part 1: Recording of interaction times and number of episodes per group.
• Part 2: Definition of interactions sub-grouped by adaptation to the objective, execution of the task, reflection on the task, and generation.
• Part 3: Dimension reduction from the above variables to identify agglomeration of groups (group profile analysis). Dimension reduction techniques were applied to profile academic groups using principal component analysis with varimax rotation in IBMSPSS v.25.
• Part 4: Hierarchical clustering between groups with cophenetic correlation coefficient. Application of hierarchical cluster classification dendrograms with Euclidean distance and Ward’s linkage to minimize the variance between groups. For this purpose, factor loadings were used for supervised classification using INFOSTAT V.2020 software with R interface.
• Part 5: Estimation of the homogeneity of the number of interactions per category using Pearson’s coefficient of variation: the calculation of the coefficients of variation, validated by bootstrapping from the application of the Bias-Corrected bootstrap confidence interval algorithm (BCBCI) programmed using IBMSPSS v.25 logic syntax. The analysis was carried out in order to determine 95% confidence intervals for this coefficient of variation; with this, we can verify the functionality of a model on social regulation and collaboration during the development of group tasks, according to the variations in the length per interval.

3. Results and Discussion

To conduct the profile analysis, the groups were treated as variables. Afterward, the episodes were organized by dimension and phase, and component reduction was performed using the principal component analysis (PCA) technique with varimax rotation in IBMSPSS v.25 [41]. This analysis resulted in the convergence of the groups into two dimensions, allowing for a more discriminant segmentation (AVE = Explained Variance > 60% for each phase), as proposed by Rogat and Adams [43].

To identify the groups that made up the two dimensions obtained in the analysis, dendrograms were designed, leading to classification by hierarchical clusters using Euclidean distance and Ward’s linkage [52]. The organization of the dendrograms was performed to minimize variance between groups and increase taxonomy efficiency within the cluster, as proposed by Rencher and Schimek [44] (Figure 2). INFOSTAT V.2020 was used to perform the classification.

Once the behaviors of the groups were established through hierarchical clustering, it was observed that groups 1, 2, and 4 formed one cluster, while groups 3 and 5 formed a second isolated cluster. This association between the clusters was consistent across the four phases of the model on social regulation and collaboration.

To analyze the behavior of each component that makes up the phases of the model on social regulation and collaboration, Pearson’s coefficient of variation analysis was performed using the bootstrapping technique with the Bias-Corrected algorithm. This technique calculated the lower and upper bounds associated with the mentioned variables. This analysis allowed us to understand the behaviors of the groups that make up each cluster in the different phases of the proposed model on social regulation and collaboration.

3.1. Analysis of Pearson’s Coefficient of Variation Using the Bias-Corrected Algorithm

The results were obtained by bootstrapping for confidence intervals, where 5000 interactions were sampled from the available data using the BCBCI algorithm in IBMSPSS v.25 syntax, with user-defined function programming. This was used to estimate Pearson’s coefficient of variation, which has been shown to be significant as a robust algorithm in failure rate estimation, particularly in statistical mediation models, as reported by Rogat and Linnenbrink [36] and more recently by Wang [47].

The results show how the groups had remarkable heterogeneity with respect to goal planning in the model, corresponding to goal setting (Figure 3).

The result of the analysis of the coefficient of variation obtained for the task regulation component reveals differences in the development of the third phase within the model associated with the regulation of task planning. This difference persists, albeit less markedly, in the fourth phase. This finding suggests that the clusters within the analyzed dimensions follow a pattern within the proposed model. It also indicates that the clusters of regulating aspects of task resolution, which show aspects of understanding, planning, executing, and evaluating the joint activity, can be improved [35,53,54].

Another aspect, as suggested by López [55] and Hadwin [40], is that understanding social regulation of the task as a metacognitive process allows us to observe how groups manage to generate mechanisms and strategies to collectively consolidate objectives and goals that lead to the successful completion of the assigned task. In this sense, the coefficient of variation analysis for task regulation showed distinctive characteristics, emphasizing the functionality of the model in identifying aspects of the social regulation of the task, in accord with the variability of intervals [31,56] (Figure 4).

The results obtained clearly demonstrate goal adaptation, allowing for improved management of group communication. The higher adaptation of goals in certain groups compared to others indicates an exercise in listening, negotiating viewpoints, and considering task execution. This aligns with Rogat and Linnenbrink’s proposition [36] of positive socio-emotional interactions within group communication. Moreover, the presence of high goal adaptation in certain groups highlights aspects of planning, execution, and reflection on the task at the inter-group communication level [1,57,58].

According to Järvenoja [35], collaborative task development requires clarity on the relevance of shared information among group members, strategic alignment on collective goals, and the ability to jointly discuss, analyze, and negotiate task aspects. The proposed model of social regulation and collaboration enabled covariance analysis to compare and verify consistency in the behavior of groups forming the two analyzed variables, thus demonstrating the model’s functionality in understanding group dynamics across each phase of the process.

Pearson’s coefficient of variation concerning collaboration revealed distinct trends in the behavior of the groups comprising this variable. Similarly, the behavior of groups within this variable was associated with the observed patterns in task regulation and communication regulation results in the exploration phase (Figure 5).

The cluster analysis and Pearson’s coefficient of variation analysis highlighted the differentiated trends of the two variables analyzed, particularly evident in groups 3 and 5, showing successful development in phases 3 and 4 of collaboration. The presence of a high level of episode generation in the collaborative process indicates deep learning [9,51,59]. In contrast, low numbers of negotiation and generation episodes in the collaborative cognitive process may indicate poor social regulation regarding goal planning, information exchange, and goal adaptation within the group [21,60]. These aspects were consistently observed as a characteristic pattern for groups 1, 2, and 4 in phases 3 and 4 of task regulation, communication regulation, and collaborative processes, where values were observed close to the lowest reference value.

The findings from the covariance analysis particularly highlighted that while all work groups appeared to maintain similar conditions during phases 1 and 2, this aspect varied notably during phases 3 and 4. During these phases, the groups dealt with metacognitive aspects related to goal negotiation and goal adaptation, as well as planning, execution, and reflection on the group task performed [13,20,41,61].

In the case of the proposed model, its effectiveness can be understood by addressing processes of social regulation and collaboration. It allows us to observe, during its application in specific cases of group interaction, how and in which phase a particular group specifically encounters difficulties. Moreover, it helps identify aspects related to task execution, communication, or collaboration that pose real challenges within the collaborative process itself [31,49,59].

Similar aspects have been explored in other studies such as those conducted by Borge et al. [37], Gutiérrez-Castillo [62], Hadwin [40], Isohätälä [56], Järvelä and Hadwin [33], Khosa and Volet [60], Kwon [19], Malmberg [41], Rogat and Linnenbrink [36], and Sobocinski [21]. The reports from these authors have focused solely on providing a detailed analysis of aspects related to the study of social regulation and/or collaboration based on the Winne and Hadwin model [32], without introducing any new developments beyond the inclusion of aspects related to cardiac monitoring and synchronicity [51], or combining factors comparing aspects related to task regulation and communication regulation with a particular stylistic trait [53,62,63].

3.2. Scope and Limitations of the Proposed Model

The consolidation of a model that makes it possible to account for collaborative processes based on the characterization of the social regulation of communication and the task is highlighted as an achievement [64]. It should also be noted that although the model takes elements from other previous proposals, its main contributions are to have been able to organize in a hierarchical way the phases that a group goes through when working together and to have been able to demonstrate how these phases are linked to the evidence of collaborative processes, as could be seen in the analysis of Pearson’s coefficient of variation [12,26].

One of the main limitations of the model is access to data, which are based on human interaction processes in which it is not always easy to identify and characterize episodes or fragments of conversations. In particular, this requires additional training on the part of the people who need to implement it; it is not really an impediment, but rather a technical aspect to be resolved by the person who wishes to implement it [2,36].

Another perceived limitation was the population with whom we worked. As they were professionals in the process of postgraduate training, it is possible that their levels of self-regulation were better than in other types of population, which would suggest the need to pilot the model with a university population in the process of professional training or young people in high school [21,31,59].

4. Conclusions

According to the results obtained, it was possible to verify that the arrangement of the classical elements through which social regulation has been studied in separate components (task regulation and communication regulation) facilitates the understanding of the collaborative process as an inter-regulated group dynamic, present in the way in which the episodes of interaction are structured.

Based on the above, it can be affirmed that there is systematic evidence to support the claim that the proposed model is reliable for analyzing group tasks oriented towards collaborative processes. The structured arrangement of the model in phases and components related to collaboration and social regulation facilitates the identification and localization of group interaction episodes in hierarchically ordered phases. This distribution allows the model to determine whether a group is progressing well in a collaborative process or, on the contrary, shows recurrent problems in a specific initial phase.

Likewise, the recurrence or stagnation of groups in certain initial phases of the model provides valuable information for developing strategies to improve group work dynamics, derived from the application of the model described here. The hierarchization of the phases makes it possible to identify patterns common to all groups that start or stop at a certain phase.

Finally, and according to the findings reported in the Pearson coefficient of variance analysis, it can be inferred that all groups present similar characteristics in the interaction episodes related to the social regulation of collaboration, suggesting that all groups present a similar starting point, allowing us to understand how cognitive strategies are developed in relation to collaborative processes as well as the implementation of metacognitive strategies associated with the regulation and management of group tasks. This shows that the proposed model not only identifies the processes of task regulation and communication during collaborative activities, but also makes it possible to see in an organized way the episodes of interaction hierarchized by components during the development and execution of collaborative tasks, an aspect that is very useful for the analysis of the collaboration or of the components that make it up discriminated by phases in order to generate improvement plans or strategies.