Investigating the relation of higher education students’ situational self-efficacy beliefs to participation in group level regulation of learning during a collaborative task

Abstract

Understanding the role individual beliefs play when the group faces challenge is key in understanding the shared regulation processes and participation that lead to collaborative learning success. As of now, there is not much research focusing on how self-efficacy plays a role in regulation taking place in collaborative group settings. Therefore, the aim of this study is to explore how situational self-efficacy beliefs relate to students’ participation in group level regulation during a collaborative task. The study involved 18 university students working in groups on a computer-based collaborative task. Repeated self-reports measuring group members’ self-efficacy were related to performance feedback from the task as well as participation in group level regulation identified from videotaped collaborative working. The results showed that self-efficacy varied depending on the nature of performance feedback. In addition, the way students participated in regulation was connected with their level of self-efficacy: low self-efficacy was associated with taking a passive role in regulation whereas high self-efficacy was associated with taking an active role. The study suggests that situational self-efficacy beliefs are associated with the participation roles during group level regulation, thus being of practical concern for educators seeking to support learners’ self-efficacy and active participation in collaborative learning.

Keywords:

• collaborative learning
• group level regulation of learning
• self-efficacy
• video analysis

Public interest statement

This article provides insight into the role of self-efficacy beliefs in collaborative learning. Individual learners have different beliefs about their capability and as group members they might act in different ways when their group faces challenge. In those challenging moments, the group must regulate their shared learning process. The main purpose of this article is to explore how situational self-efficacy beliefs relate to students’ participation in group level regulation during a collaborative task. The results from combining self-reports and qualitative video analysis showed that low self-efficacy was associated with passive participation in regulation whereas high self-efficacy was associated with active participation. These results are of practical concern for educators seeking to understand the more “invisible” processes of the human mind and to support learners’ active participation in collaborative learning.

1. Introduction

Interpersonal collaborative skills have been identified as crucial aptitudes in future working life (McKinsey Global Institute, 2017). Simultaneously, educational practices need to answer to the growing requirement towards strengthening such human capabilities as regulation of learning to tackle challenging problems in our rapidly changing world (Haddington et al., 2021). Much like in the common workplaces, in collaborative learning within educational contexts, no individual is solely responsible for the success or failure of the group; each plays a part in achieving a common goal. Thus, members of the group must collectively regulate their shared engagement towards that goal (Hadwin et al., 2017; Järvelä et al., 2010; Rogat & Linnenbrink-Garcia, 2011). This joint engagement requires active participation from the individuals of the group (Volet et al., 2017) as they contribute to joint regulation attempts whenever the group faces challenges.

However, individual group members possess different beliefs about their capabilities (Bandura, 1997) to participate in the collaborative tasks they engage in. According to Bandura (1997), an individual’s self-efficacy is closely tied to the domain of the task at hand but moreover to the context and social situation where the learning takes place (Bernacki et al., 2015; Pintrich et al., 1993; Schunk, 1984; Zimmerman, 2000), which emphasize the need to address individuals’ self-efficacy beliefs also when studying regulated learning in collaborative learning situations. Moreover, as is shown by previous research, these beliefs have the power to influence not only the learning outcomes but also the ways an individual regulates their own learning (Bandura, 2012; Winne & Hadwin, 2008; Zimmerman, 1989). What has not been researched enough, however, is the context of collaborative group work where the group members might act in different ways as their group face challenges, and where group level regulation is needed. So far, there are only a few studies looking at participation during group level regulation activities in collaborative learning context (e.g., Grau & Whitebread, 2012; Iiskala et al., 2015; Ucan, 2017; Volet et al., 2009), and there is lack in research that investigates how self-efficacy beliefs associate with this participation. In other words, there is a need for better understanding how situational self-efficacy beliefs relate to the way group level regulation unfolds in collaborative learning contexts. This is important, because understanding the role individual beliefs play when the group faces challenge is key in understanding the shared regulation processes that lead to collaborative learning success. Accordingly, this study focuses on higher education students’ situational self-efficacy and its role in regulating collaborative learning.

1.1. Self-efficacy in the learning situation

The belief of personal efficacy is defined as an individual’s perceived capability to perform a certain task (Bandura, 1977). Several studies have investigated self-efficacy in higher education. Many studies emphasize the role of self-efficacy by showing its impact on academic achievement (Honicke et al., 2020; McKenzie & Schweitzer, 2001; Vogel & Human-Vogel, 2016). Self-efficacy has also been linked to high level cognitions, such as elaboration (Stolk & Harari, 2014) and knowledge construction (García-Almeida & Cabrera-Nuez, 2020).

Although self-efficacy stems from individual perception, research showed decades ago that self-efficacy is very sensitive to variables and changes in the task context where the individual performance takes place (Schunk, 1984; Zimmerman, 2000). Therefore, self-efficacy is not static but changes dynamically within and across learning situations (Bernacki et al., 2015; DiBenedetto & Bembenutty, 2013; Geitz et al., 2016). Thus, self-efficacy is highly context specific. According to prior research, self-efficacy is closely tied to the domain of the task at hand and depends on learners’ prior experience with similar tasks (Bandura, 1997, 2012).

Previous performance, or mastery experience, has been thought to be the most reliable source of information for self-efficacy since performance concretely reflects capability (Bandura, 1997; Schunk & DiBenedetto, 2009; Usher & Pajares, 2008). A meta-analysis from Sitzmann and Yeo (2013) even suggests that self-efficacy is “primarily a product of past performance”. In general, self-efficacy increases when learners succeed in a task and falls when they fail in one (Pintrich & Schunk, 2002). Previous research has found relationships between past performance and self-efficacy in different learning contexts (e.g., Eseryel et al., 2014; Wilson & Narayan, 2016). In learning, knowledge of past performance can provide feedback for students (e.g., Wilson & Narayan, 2016) that helps them know how well they are doing. Thus, performance can provide cues for learners while a task is in progress in addition to when it is complete, contributing to the situational nature of self-efficacy beliefs. Based on the significant body of research showing that self-efficacy beliefs are a strong predictor of academic motivation and achievement of individuals (Bandura, 1977; Honicke et al., 2020; Linnenbrink & Pintrich, 2002; Pajares, 1996; Schunk, 1989; Schunk & DiBenedetto, 2009), it can be further assumed that these beliefs also play a role in collaborative learning processes by affecting individual group members’ choices and how they expend effort on collaboration (Schunk & DiBenedetto, 2009).

1.2. Self-efficacy and regulation of learning

Recent studies have emphasized the link between self-efficacy and self-regulated learning (SRL; e.g., Chen, 2022; Graham et al., 2020). Previous research has shown that self-efficacy not only makes a difference in how individual students perform on academic tasks, but also makes a difference in how they activate SRL during task performance (Hong & Park, 2012; Pintrich & de Groot, 1990; Zimmerman, 2011; Winne & Hadwin, 2008; see also, De Backer et al., 2022). Research evidence suggests that high self-efficacy beliefs are linked to adaptive study skills in learning (Linnenbrink & Pintrich, 2002) and learning satisfaction (Aldhahi et al., 2022). Students with high self-efficacy are, for example, found to be more engaged with learning activities (Alemayehu & Chen, 2021) and to use more self-regulation strategies in their academic performance (Lee et al., 2021, 2020; Pintrich & de Groot, 1990). Self-efficacy has been found to be a significant predictor of such SRL strategies as elaboration, critical thinking, and time management, for instance, (Lee et al., 2021). Self-efficacy has also been linked to factors such as high-level cognitive skills (Wang & Lin, 2007), mastery-orientation (Geitz et al., 2016), and self-monitoring (Alemayehu & Chen, 2021; Zimmerman, 1989). In addition, individuals with high self-efficacy are more persistent when faced with challenges (Komarraju & Nadler, 2013; Pintrich & de Groot, 1990; Usher & Pajares, 2008). While prior research has focused on the relation between self-efficacy and SRL, researchers have only recently started to investigate self-efficacy in relation to regulation happening in group level collaborative settings (De Backer et al., 2022).

1.2.1. Group level regulation and individual participation

As collaborative learning is a collective effort, mere self-regulation in the face of challenges is not enough: group level regulation is needed. In collaborative learning contexts, an individual is not solving a task alone; all group members are taking part in joint activity with a common goal. Thus, collaborative situations require students to also regulate their shared engagement towards a task (Järvelä et al., 2010). This is referred to as group level regulation of learning, which can be defined as a group’s joint regulation of their shared understanding of the task, enactment on the task, and engagement during the task (Rogat & Linnenbrink-Garcia, 2011). It is commonly perceived to cover co-regulation and socially shared regulation of learning (e.g., Hadwin et al., 2017).

Perry and Winne (2013) refer to co-regulation of learning (CoRL) as the provision and reception of regulatory support among peers during collaboration. It includes the metacognitive processes that support and prompt the regulation of cognition, behaviour, motivation, and emotions (Hadwin et al., 2017). In turn, socially shared regulation of learning (SSRL) refers to groups taking metacognitive control of the task together through the group’s deliberate and negotiated effort to fine-tune the above-mentioned conditions (Hadwin et al., 2017; Järvelä et al., 2017). Thus, SSRL should be understood as a more reciprocal process (Isohätälä et al., 2017) that calls for “shared awareness of goals and joint monitoring of progress toward a shared outcome” (Perry & Winne, 2013, 47). In other words, during collaborative learning situations learners also need to support others (CoRL) and engage in regulating their learning process together with a group (SSRL; Hadwin et al., 2017).

Since group level regulation is a collective effort as well, contributions from multiple group members are needed for it to manifest between individuals, which highlights the importance of participation (e.g., Isohätälä et al., 2017; Volet et al., 2017; Vuorenmaa et al., 2022). For example, CoRL always includes at least one student with the role of “co-regulator”, which refers to an individual who has the required knowledge or expertise in each situation (Perry & Winne, 2013). In addition, SSRL has been associated with active participation during collaborative situations where interaction between two or more group members is needed to resolve challenges with the group (Isohätälä et al., 2017). Recent research findings highlight the importance of group members’ active and joint participation in social interactions for regulation in collaborative learning (Vuorenmaa et al., 2022).

To date, only a few studies have directly addressed participation in group level regulation activity (e.g., Iiskala et al., 2015). However, previous research indicates that learners working in groups tend to participate in the joint activity unequally and that the level and patterns of participation can vary depending on group type, context, and task (Isohätälä et al., 2020). Previous studies take different approaches to understanding participation during social learning situations; in many cases, roles have been used to decipher individual participation in the context of collaborative learning (e.g., Strijbos & De Laat, 2010; Volet et al., 2017). For example, Volet et al. (2017) studied participation during collaborative learning and considered it through different role categories, such as content focused or performance focused roles. Ucan (2017), in turn, categorized group members’ participation in regulated learning processes during collaborative tasks based on whether the group member was an initiator or responder for a regulation episode. While some research has been done on individual participation in group level regulation, little attention has so far been given to its motivational aspects (De Backer et al., 2022; Volet & Mansfield, 2006). For example, De Backer et al. (2022) investigated how motivation and self-efficacy related to students’ regulation profiles in an asynchronous computer-supported collaborative learning context. The results of the study showed that students who were more active regulators on the social level also had high beliefs of self-efficacy (De Backer et al., 2022). However, what remains unknown is how this relation might unfold in face-to-face collaborative situations that require reciprocal and synchronous participation from the members of a group. It is also unknown how situation-specific self-efficacy beliefs relate to participation when group level regulation takes place.

2. The current study

This study focuses on higher education students’ situational self-efficacy and its role in collaborative learning. In particular, the study aims, first, to investigate how situational self-efficacy beliefs relate to the performance feedback groups received during a collaborative task. Second, it investigates how students’ situational self-efficacy beliefs relate to their participation in group level regulation between group members.

The research questions are:

RQ1: How do situational self-efficacy beliefs relate to performance feedback during a collaborative task?

RQ2: How does level of situational self-efficacy relate to participation during group level regulation?

3. Methods

3.1. Participants

Participants were 18 higher education students (61% male, aged 20–44). They came from a variety of educational backgrounds, including bachelor’s (7%), master’s (57%), and PhD level students (36%). Participants also represented over 10 different nationalities. Lastly, participants were from different disciplines, for example, education (43%), economics and business (21%), or other fields (engineering, marketing, pharmacology, wireless communication). Three of the participants did not specify their backgrounds.

Participants were recruited from the University of Oulu by handing out fliers on campus and posting on social media. A free lunch ticket was offered in exchange for participation. Taking part in the study was voluntary and each participant gave a written consent. Participants were randomly assigned into six groups of three students each to complete a collaborative task. The six groups in the study were versatile in terms of their performance. The groups included high performing, low performing and moderately performing groups in terms of their final score in their collaborative task. This provided a variety of situational contexts where groups were succeeding differently with their task. The sample of eighteen participants was deemed to be appropriate with the capacity to provide adequate means for conducting exploratory and in-depth qualitative analysis (Volet et al., 2009). Diversity of participants’ backgrounds and groups’ performance was expected to enrich the set of reported self-efficacy beliefs and observed regulatory activities.

3.2. The task

For their collaborative task, the participants ran a complex computer-based company management simulation called the Tailorshop (Danner et al., 2011). In Tailorshop the idea was for the participants to run their own t-shirt production. The task provided a collaborative setting where participants needed to work as a group and make joint decisions about how they could maximize the productivity of the company. The aim of the task was to produce t-shirts for sale with the goal of increasing the company value by as much as possible during 12 simulated “months”. The Tailorshop was run by manipulating 12 variables such as worker wage, machine-maintenance, raw material ordered, business location, and advertising, which allowed the participants to influence the company value. This value was updated by the simulation after every “month”. The simulation began with six exploratory “months” meant for participants to get to know the simulation and variables. After exploration, participants proceeded to start their actual 12-“month” performance phase with the task, which was assessed and logged by the simulation. The six groups took on average 1 hour 32 minutes to complete the task.

3.3. Measures

The study combines students’ self-reports about their self-efficacy beliefs with task performance feedback gained from the Tailorshop simulation program and group work that was videotaped to capture actual group level regulation. A repeated single item self-report measure (see, e.g., Goetz et al., 2016) was used to measure situation-specific self-efficacy beliefs. The simulation program prompted participants to evaluate their self-efficacy after the exploration phase and then after every third “month” during the task. In practice, the participants responded to the question, “How confident are you that your team is attaining the current task goal?” using a 5-point Likert scale ranging from “Very confident” (5), “Confident” (4), “Moderately” (3), “Slightly” (2) to “Not at all” (1). Altogether, the participants evaluated their self-efficacy five times during the task.

Performance feedback was based on the company value. The program provided feedback to the group after every “month”. This study used only the feedback provided before the situational self-efficacy ratings (every three “months”) for analysis. Positive feedback was recorded when the group managed to increase the company value in comparison to the preceding “month”. Negative feedback was recorded when company value declined from the previous “month”. Video recordings of the groups’ collaboration taken using 360°Cameras provided insights into actual group level regulatory activities and associated participation occurring during the collaborative task.

3.4. Method of analysis

3.4.1. Self-efficacy in relation to performance feedback

To investigate how situational self-efficacy beliefs are related to performance feedback, the self-efficacy ratings (f = 72) given during the task were paired up with the performance feedback (positive or negative) provided at the same point in the task. Altogether, there were four measurement points at which performance feedback was given and followed by provision of situation-specific self-efficacy ratings. The first self-efficacy rating was excluded from the analysis since it was not preceded by performance feedback. Due to limitations from the amount and normality of data, a nonparametric Mann-Whitney test (alpha level = .05) was performed to investigate differences between self-efficacy ratings after positive and negative feedback.

3.4.2. Identifying regulation and participation by using qualitative video analysis

Videos from groups’ collaborative work were analysed in two phases using qualitative content analysis. The analysis focused on participants’ verbalized behaviour (Chi, 1997). Phase 1 focused on locating episodes of regulation and phase 2 on classifying students’ participation during those episodes. The analysis was done using Observer XT12.5 qualitative analysing software (Noldus Information Technology), which is a tool for systematically organizing, coding and analysing observational data (Snell, 2011). With the Observer XT software it was possible to mark down and code the observed regulation activities at different time points in the videos and the observed participation of different individuals.

Phase 1. In the first phase of the analysis, episodes of regulation were identified from the video. Namely, the first phase located moments where groups confronted a challenge in their learning (Hadwin et al., 2011) which led to group level regulation activity. Thus, the unit of analysis was defined as a meaningful episode (e.g., Järvelä et al., 2016) that included a clear sign of a challenge to the task progress and a subsequent regulatory response from at least one group member. In the current data, a challenge was identified when a lack of task understanding (e.g., ‘I don’t get it ‘); poor task progress relative to the task goal (e.g., ‘We have very little money in our bank account ‘); constraints related to the task context, such as time, equipment, or difficulties using the simulation (e.g., ‘Why can’t I change this value? ‘); or negative emotional and motivational expressions (e.g., ‘The company is already destroyed ‘) were observed (e.g., Pintrich, 2004).

Group level regulation activity, in turn, was identified when groups regulated their joint understanding of the task, enactment on the task, and engagement (Rogat & Linnenbrink-Garcia, 2011). Regulatory behaviour was always a response to an articulated challenge, so it could include the group members clarifying their understanding of the task (such as task concepts, goals, or structure), adapting or reflecting on their strategy based on task progress, navigating task constraints, or productively reacting to negative emotional expressions. This regulatory response was labelled as either CoRL or SSRL (Malmberg et al., 2017; Ucan & Webb, 2015). CoRL responses were defined as moments when regulatory support was given or requested (Ucan & Webb, 2015). One group member had a clear role as the target of regulation or the regulator of one or more group member(s). Others that participated in regulation expressed mainly agreement (Malmberg et al., 2017). SSRL responses were defined as those in which at least two participants regulated their shared activity (Ucan & Webb, 2015) by complementing and bringing new and additional information into the discussion (Malmberg et al., 2017). Thus, participants’ input reflected the acknowledgement and integration of others’ contributions (Isohätälä et al., 2017). This usually led to longer-lasting exchanges between group members, and these episodes often involved all three group members taking part in discussion.

During the collaborative task, a total of 92 group level regulation episodes were identified using qualitative analysis. Of these episodes, 63% (f = 58) were considered CoRL and 37% (f = 34) SSRL episodes. Another researcher coded 30% of the same video data to ensure the quality of the developed coding measure, resulting in Cohen’s kappa κ = .68. Categorizing of the type of regulation generated a kappa value κ = .70. Both these values indicate good agreement (Flight & Julious, 2015).

Phase 2. Next, students’ participation during group level regulation episodes was coded according to the nature of their verbal participation. Four different data-based role categories for participation were used (See, Table 1). The aim of this phase was to distinguish individuals’ contributions to the regulation activity (e.g., Iiskala et al., 2015). The role of Initiator referred to a participant who articulated a challenge in the task process. The role of Influencer referred to participation during regulation. The role of Dominator referred to dominating regulation or being the only regulator. The role of Observer (passive) meant a lack of contribution during the regulation episode.

Table 1. Coding scheme for participation roles in group level regulation

Role	Description	In the data
Initiator	Articulating a challenge	Acknowledging a situation involving a lack of task understanding, poor task progress, or negative emotional or motivational states
Influencer	Participating in regulation, but not clearly dominating	Expressing agreement during a regulation situation or participating when there is no clear dominator of regulation
Dominator	Being the only regulator or clearly dominating regulation	Co-regulating others or leading a substantial portion of the discussion during regulation
Observer (passive)	Not participating in regulation, but merely observing or rationalizing the challenge	Not participating in regulation. E.g., merely taking part in the discourse about a challenge but not giving input to adapting the strategy

In this coding scheme, individual roles were coded inside the identified regulation episodes, and they could be demonstrated by multiple individuals at the same time. Participants could also have more than one role in one regulation episode. For example, the role of Influencer or Dominator could be paired up with Initiator during regulation episodes, meaning that a participant could both articulate a challenge and participate in the regulatory activity occurring after it. As for the role of Observer (passive), this code could only be used if no other role was appropriate, meaning that an individual was not articulating a challenge or taking part in regulation. Therefore, the connection of these roles to the group level regulation activity was important for this coding. It should also be noted that the role of Initiator was automatically present in all episodes due to the coding scheme. In total, Influencer (f = 145) and Initiator (f = 105) were found to be the roles participants took most often. The Dominator (f = 59) and Observer (f = 42) roles were observed less frequently. Inter-rater reliability coding for 30% of the data resulted in a kappa value κ = .77.

3.6.1. Combining participation roles with self-efficacy ratings

Next, the analysis related students’ participation roles in group level regulation to their situational self-efficacy levels. For this analysis, self-efficacy ratings were reduced to three levels of self-efficacy (LSE): “Low”, “High”, and “Moderate”. Self-efficacy ratings 1 and 2 formed the category of “Low” (f = 29) and ratings 4 and 5 the category of “High” (f = 36) self-efficacy. Rating 3 represented a “Moderate” (f = 25) self-efficacy category.

To investigate how participation roles occurred in relation with situational LSEs, all detected participation roles in group level regulation (f = 351) were matched with the preceding situated LSEs. The analysis utilized self-efficacy self-reports 1-4 as these ratings were followed by collaborative group work phases. The last self-efficacy self-report was excluded from this part of analysis since it was done after the task was completed. Nonparametric test was deemed appropriate for investigating the relations because the variables involved categorical data and repeated observations of the groups’ collaborative efforts.

A chi-square test (for frequencies) was conducted to investigate the association between situational self-efficacy levels and subsequent participation roles. Significant associations between categories were examined further by exploring significant z-scores from adjusted residuals at a Bonferroni-corrected α -level, where the two-tailed p .05 is divided by the number of cells in the contingency table (MacDonald & Gardner, 2000).

4. Results

4.1. RQ1: How do situational self-efficacy beliefs relate to performance feedback during a collaborative task?

RQ1 focused on exploring situational self-efficacy beliefs in relation to performance feedback. The results showed that there were differences in reported self-efficacy beliefs during the task depending on the nature of performance feedback received. The overall mean value of students’ situational self-efficacy ratings after receiving positive performance feedback was higher (M = 3.3, SD = 1.2) than the mean value after receiving negative performance feedback (M = 2.7, SD = 1.3). A Mann-Whitney test indicated that this difference was significant (U = 465, p = .032, 2-tailed). The difference was visible across the collaborative task, as can be seen from Figure 1.

Figure 1. Self-efficacy means (SE (M)) after group performance feedback.

4.2. RQ2: How does level of situational self-efficacy relate to participation during regulation?

RQ2 concentrated on investigating how participation roles relate to the preceding level of situational self-efficacy. The analysis show that different LSEs were associated with different participation roles. A low LSE was most often followed by the role of Initiator (33.9%) and least often by the role of Dominator (15.2%). A high LSE was most often followed by the Influencer role (59.6%) and least often by the Observer (passive) role (5.3%). In addition, moderate LSE was usually followed by the Influencer role (34.4%) and least often by the Observer role (10.4%). Figure 2 illustrates the frequency of the various roles after each LSE rating category. The X axis shows the three LSE categories and the Y axis shows the frequency with which each role occurred after those reports.

Figure 2. Roles during regulation after different LSEs.

To explore the relation between LSE and subsequent participation roles during regulation, including both CoRL and SSRL, chi-square statistics were used. A chi-square test of independence was significant, indicating that LSE is associated with the role(s) taken during regulation (X2 (df = 6, f = 351) = 32.16, p < .001). More specifically, there were significant associations between the Observer role and a low LSE (z = 3.4, p < .01) and the Influencer role and a high LSE (z = 4.8, p < .01). The significant associations were interpreted from adjusted residuals by using Bonferroni adjustment to the critical z value of 1.96 (MacDonald & Gardner, 2000). Applied to the present study, this procedure produced an adjusted alpha of .004. The critical z score associated with this adjusted alpha was 2.89 (or approximately 2.9), and cells with an adjusted residual greater than that were considered statistically significant. These scores are presented in Table 2 with the frequencies of participation roles after different LSE ratings.

Table 2. Frequencies and adjusted residuals (z) for participation roles after different LSE

Self-efficacy level	Participation role during group level regulation
	Observer			Initiator		Influencer		Dominator
	f	z		f	z	f	z	f	z
Low	23	3.4*		38	1.1	34	−2.9	17	−.6
Moderate	13	−.7		41	.9	43	−2.0	28	2.1
High	6	−2.7		26	−2.0	68	4.8*	14	−1.6

Note: *Categories with adjusted residuals significant at a Bonferroni-corrected α =  0.004 (z score > 2.9)

The same associations were also found while focusing solely on CoRL episodes. A chi-square test of independence was significant, indicating that LSE was associated with the role(s) taken during CoRL (X2 (df = 6, f = 210) = 19.05, p = .004). More specifically, there were significant associations between a low level of self-efficacy and the Observer role (z = 3.1, p < .01) and between a high level of self-efficacy and the Influencer role (z = 3.2, p < .01). When performing the analysis with only SSRL episodes, 50% of the expected cell counts in the cross-tabulation were less than 5, so a Fisher’s exact test was used. The Fisher’s exact test statistic resulted in a p value of .04. However, this did not reach the Bonferroni corrected alpha level and therefore the results were considered not to be significant. Thus, no association was found between LSE and roles during SSRL.

5. Discussion

The aim of this study was to investigate how higher education students’ situational self-efficacy beliefs relate to how they participate in group level regulation during collaborative learning. First, the relationship of situational self-efficacy beliefs to performance feedback received during the collaborative task was investigated. Second, it was investigated how these situational self-efficacy beliefs relate to students’ participation in group level regulation, including both CoRL and SSRL. In the study, participants engaged in a computer-based task by working in small collaborative groups. In the analysis, students’ situational self-efficacy ratings across the task were combined with performance feedback from the same point in time as well as students’ subsequent participation roles during group level regulation episodes identified on video.

The results of the first research question showed that the situational self-efficacy beliefs differed significantly depending on whether the group’s recent performance feedback was positive or negative. Similar results have been obtained in previous studies. For example, poor task performance has been linked to decreased task self-efficacy (Wilson & Narayan, 2016). Some studies suggest that these types of changes are due to self-efficacy being calibrated to align with current performance (DiBenedetto & Bembenutty, 2013). In other words, the results indicate that situational self-efficacy seems to be sensitive to performance feedback during collaboration so that higher situational self-efficacy beliefs often coincide with positive task performance feedback.

The results from answering the second research question showed that groups actively regulated their task progress. This group level regulation was studied by analysing students’ participation roles. The results indicate that the roles participants took in group level regulation were associated with their level of situated self-efficacy. In particular, low self-efficacy was associated with taking a passive role in regulation, while high self-efficacy was associated with taking an active role in regulation. The results confirm previous research suggesting that individuals with high self-efficacy participate more actively when faced with challenges (Bandura, 1997, 2012; De Backer et al., 2022; Schunk, 1990). That is, students with high self-efficacy beliefs are more confident in their own abilities, which impacts how eager they are to act when faced with complex task demands (Pajares, 2008). Similar results were found when focusing on the association between level of self-efficacy and roles during CoRL episodes.

From another perspective, even though it was not statistically significant, reports of low self-efficacy level were most often (33.9%) followed by the student taking the role of Initiator, signalling an active role in recognizing challenges in learning. This might have significance for identifying low self-efficacy beliefs in groups as “triggers” for regulation. Prior research on collaborative learning has emphasized that increased awareness of regulation by individual students can invite other group members to participate in the regulation (Järvelä et al., 2016). The relationship between self-efficacy and specific monitoring activities has been studied in the hypermedia learning context and found to be detectable (Moos & Azevedo, 2009).

As for the high self-efficacy level, it was found to be related to the Influencer role. This indicates students taking an active stand in controlling recognized challenges, and rarely merely observing or dominating during regulation. This might imply that individuals with high self-efficacy are more reciprocal when taking part in regulatory action. This factor is especially important for SSRL (Isohätälä et al., 2017). This might also partly explain why focusing solely on SSRL episodes revealed no association between level of self-efficacy and roles during regulation. In addition to SSRL being simply rarer than CoRL (e.g., Malmberg et al., 2017), the very essence of SSRL is reciprocal activity. Thus, it can to some extent be expected that, during SSRL, the nature of the activity itself presupposes a form of equilibrium in participation. Such fine-grained processes of social interaction were not captured in the participation roles analyzed in this study. However, there may also be a reason for researchers to cautiously consider whether self-efficacy beliefs become more prominent during CoRL episodes, which, by nature, involve situations in which one individual is the more knowledgeable and one is the less knowledgeable receiver of support (Perry & Winne, 2013).

5.1. Limitations

This study has some limitations that need to be discussed.

First, the results of the study are based on a small sample of six groups of participants and situated on a selected task. Thus, direct generalizations should not be formulated based on its current extent. Furthermore, the statistical approach does not provide the means for making causal connections or detecting effects. It instead reveals interesting indications of associations between variables which can hopefully be used to motivate further studies. In addition, the participants in this study were divided into random groups; no additional attention was given to the educational backgrounds of participants, which might have affected their engagement and collaboration during the task. For example, the groups included bachelor’s, master’s, and PhD level students, which could potentially result in the domination of PhD level students in group participation. However, the nature of the simulation and the task itself did not require a specific knowledge base that would inevitably lead to a dominant position of students with higher education levels. Adding to this, the Tailorshop simulation was an unfamiliar setting for the participants, presenting a highly complex task that required intense cognitive effort. Thus, the findings of this study might not be applicable to other contexts. According to Kreijns et al. (2003), the key to successful collaboration is social interaction, which often gets ignored in favour of emphasis on the cognitive aspects of learning. In this study, the quality or conditions of social interaction were not considered, despite their profound importance in collaborative learning.

Second, as it is the case with all qualitative analysis, the coding for identifying group level regulation and participation from the videos is, to some extent, interpretational. In this research, cognitive aspects of regulation, such as strategy use and task understanding, were emphasized in the coding. The video analysis also did not account for fine-grained subtleties in participation, thus contributing to the absence of distinguishable differences in individual participation during SSRL. Moreover, the division between more “active” and “passive” participation based on verbal contributions can be problematic in the collaborative learning context, as being silent does not automatically correspond with being disengaged (Remedios et al., 2008). Thus, further research could focus on the different functions of participation during regulation, such as the roles of initiator and responder (see, e.g., Ucan, 2017).

Lastly, due to the internal nature of motivational beliefs, empirical research often lies in participants’ own perceptions and interpretations. Repeated self-reports offer time-sensitive access to students’ self-efficacy, but the results from these reports should be interpreted with caution.

5.2. Conclusion and future directions

The results of the study add to the existing research by suggesting that situational self-efficacy beliefs may play an important role in higher education students’ group level regulatory processes during a collaborative task. First, this study reinforces previous research findings by showing that higher situational self-efficacy beliefs are often coincided with positive task performance feedback during a collaborative activity. Second, it provides a novel example of how, in collaborative learning, the level of self-efficacy is associated with the participation roles individuals take during group level regulation.

Bakhtiar and Hadwin (2020) maintain that the context in which regulation is triggered is important, and they emphasize the need to consider the conditions for regulation to occur. From both educational and societal perspectives, it would be important to know more about the “invisible” processes of the human mind, such as regulated learning and self-efficacy beliefs, that help humans to adapt to new and challenging situations in life and across academic arenas (Haddington et al., 2021). In educational contexts, self-efficacy may have the potential to affect learners’ choices to participate in or even activate group level regulation during collaborative learning. The current study appeals for more empirical research to be conducted in authentic educational settings. Based on this study, it would be interesting to further explore how individual self-efficacy beliefs and participation patterns vary within tasks and explain how regulation arises from individual minds to evolve into a group level process. Understanding the interplay between individual situational conditions (e.g., self-efficacy) and group level regulation is needed to explore how groups succeed in developing effective ways to engage in regulatory activities for collaborative learning success (Mänty et al., 2020). To that end, a replication study of the current experiment with a larger sample size, multilevel analyses, and with multiple collaborative sessions is needed. It would also be interesting to further investigate how different levels of self-efficacy relate to awareness of challenges in learning or socially shared regulation episodes characterized by reciprocity. Overall, more empirical studies are needed to investigate the relationships between these phenomena.

This study also has implications for educational practice concerning how to take account of learner persistence in challenging situations during collaborative activities. In conclusion, it would be beneficial for both teachers and learners to recognize the mechanisms of self-efficacy. This way, they can be better equipped to understand its situational nature and potential behavioural influence in collaborative learning contexts. Students taking part in group learning activities do so according to their perceived capabilities. Thus, self-efficacy may be linked to how students play their parts in the group. As suggested by Wang and Wu (2008), quality feedback could be provided by teachers to promote students’ self-efficacy in higher education. For example, students with low self-efficacy beliefs might benefit from learning environments and prompts that would encourage them to participate in regulation activity. Teachers could support self-efficacy in collaborative contexts by promoting relatedness and social interdependence among group members (Van Blankenstein et al., 2019). Stolk and Harari (2014) further discuss that if learners’ self-efficacy is enhanced by creating environments that support it, this will result in students’ personal development, engagement, and well-being. Aiming to attain these important academic goals through educational practices, such as supportive learning environments, appropriate feedback practices, and learner activation prompts, holds great promise for enhancing the quality of learning and education.

Summary of Main Research Activities

Our research is carried out in Learning and Educational Technology research Lab (LET) at the University of Oulu, Finland. LET is committed to research in the learning sciences and technology-enhanced learning. In LET we tackle the challenge that the 21st century sets for learning; learning as a process that go beyond knowledge. It is based on motivational competence and use of effective learning strategies in individual and in collaborative work to manage different demands of the everyday life. In practice LET aims for understanding the skill and will of learning in order to design future innovations for learning. Our research paper contributes to this understanding by bringing light into the “invisible” processes of the human mind, such as regulated learning and self-efficacy beliefs, that play a role in collaborative learning contexts and help humans to adapt to new and challenging situations during learning and life.

Acknowledgements

Data collection was carried out with the support of LeaF Research Infrastructure (https://www.oulu.fi/leaf-eng/), University of Oulu, Finland.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was supported by Academy of Finland [grant number 297686 and 308809].

Notes on contributors

Sara Ahola

Sara Ahola is a doctoral researcher at the Learning and Educational Technology Research Lab, Faculty of Education and Psychology, University of Oulu. Her research interests include self-efficacy beliefs, group level regulation and collaborative learning.

Jonna Malmberg

Jonna Malmberg in an assistant professor at the University of Oulu, Learning and Educational Technology Research Lab, Faculty of Education and Psychology. Her research focuses on investigating self- and socially shared regulation of learning in digital learning environments.

Hanna Järvenoja

Hanna Järvenoja is a professor at the University of Oulu, Learning and Educational Technology Research Lab, Faculty of Education and Psychology. Her research interests are self-regulated learning, motivation and emotion regulation and socially shared regulation processes in collaborative learning and technology enhanced learning.