Bandura [12] defines self-efficacy as an individual’s belief in their ability to “successfully execute the behavior required to produce the outcomes” (page 193), with this belief determining “how much effort will be expended, and how long it will be sustained in the face of obstacles and aversive experiences” (page 191), thus influencing personal choices, motivation, patterns, and emotional reactions.

According to Bandura [12,13], there are four main factors that influence self-efficacy: (1) mastery experiences: individual’s prior experiences and performances, influenced by the success of previous experiences, which cumulatively increases their sense of self-efficacy; (2) vicarious experiences: the observation of activities performed by others, in which visualizing their accomplishments can increase their confidence in performing them as well; (3) verbal persuasion: individuals are encouraged to believe in their abilities through the use of suggestion, persuasion, or instruction; and (4) emotional arousal: emotions influence one’s performance, behavior, and perception of challenges, with positive emotions generally increasing confidence and vice-versa.

Self-efficacy is an important predictor of academic performance [3,14-19]. According to Rittmayer and Beier [20], individuals with high self-efficacy tend to set more challenging goals for themselves, envisioning a positive academic performance, thereby positively impacting STEM task performance and engagement. On the contrary, individuals who doubt their capacities (reflecting a low sense of self-efficacy) tend to avoid certain situations, believing that they are incapable of surpassing them and losing motivation [12,15,21], making them more likely to leave STEM education [22,23]. Self-efficacy may be a driving force in this process, helping shape career preferences [24] and aspiration [1,21,25-27].

Early exposure to STEM content in a child’s educational journey might be crucial for shaping STEM identity and self-efficacy [28-30]. Studies have proven the success of games in engaging students’ learning interest [24,31-35], with examples including how gamification has been applied in modern pedagogical strategies [32,36], and as a potential source for influencing self-efficacy [31]. Kinzie and Joseph [37] define a game “as an immersive, voluntary, and enjoyable activity (...) [that] provides a safe environment for taking chances and the opportunity to develop the knowledge and refine the skills required to succeed”. Using games in the classroom is becoming increasingly popular [31,33,34,38,39], facilitating learning through an enjoyable environment that fosters knowledge acquisition, learning achievements, and critical thinking, encouraging meaningful connections [31,33,34,40-42]. Reduced concern about failure, assumed within gameplay, contributes to a more positive learning experience [40,43], making games effective to introduce STEM-related scientific concepts [31].

For contemporary challenges, collaboration has become a promising mode of human engagement [44-46], defined as “a continued and conjoined effort toward elaborating a ‘joint problem space’ of shared representations of the problem to be solved” [47]. It is also seen as an educational approach that improves learning outcomes [7,48-51] through group activities, such as information sharing and knowledge development between peers [48-50,52]. With a bottom-up approach, it leads to joint problem-solving [44,45,53] by combining game-based learning with student-led game design [54,55]. These joint interactions reduce frustration and increase confidence through peer support [56-58], enabling competence development by sharing personal experiences, knowledge, and ideas [59-62]. Developing games that support collaborative interactions can thus enable positive outcomes regarding STEM achievement, persistence, engagement, and problem-solving skills [7,50,56,63-65].

Recent studies [41,51,66-70] show a positive influence of game-based collaboration in self-efficacy: (1) Sung and Hwang [41], through pre- and postexperience questionnaires, explored collaborative game-based learning through knowledge organizing and sharing to increase the students’ interest; (2) Pan et al [69] found that digital game-based learning significantly improved students’ social collaboration, problem-solving skills, and motivation; and (3) Yin et al [70] studied the effects of pair programming in CT education, with collaborative learning significantly improving students’ self-efficacy compared to individual learning, where students with higher self-efficacy were more engaged in CT learning, enabling better CT performance and higher learning satisfaction.

Bardram [71] proposed 3 levels of dynamics that a collaborative activity entails:

Co-ordination: individuals carry out tasks to achieve the overall objective of the activity. At this lowest hierarchical level (the “how”), individuals are coordinated to perform the assigned functions/tasks, which are carried out individually, leading to a routinization of the work.

Co-operation: individuals interact to decide on a common object, still sharing the objective of the activity. In this intermediate hierarchical level, the objective is still stable, but task distributions are actively discussed and adjusted between each other depending on the object of work (the “what”).

Co-construction: individuals question the activity’s main objective. At this highest hierarchical level, reformulating the objective implies changes in the objects of work, tasks, and interactions. The overall objective or motivation is collectively reconstructed by the individuals (the “why”).

The hierarchy between the 3 levels is dynamic, reflecting transitions or breakdowns [71] between them upward and downward; we move upward when there is a reflection on the object of work (co-operation) or the entire objective of the activity (co-construction); or we move downward to implement the renewed objects from the new objective, implementing new tasks and routines.

CT is a problem-solving approach rooted in computer science, considered a vital contemporary skill and valuable in STEM education, as it addresses computational tools and problem-solving processes [72-75] for solving complex scenarios [76-78]. It involves conceptualizing problems in abstract ways, being taught to people of all ages [72-74,79,80], and expanded to various disciplines [81] (such as biology, geosciences, economics, and humanities [72]).

The CT approach is composed of four main characteristics [82-84]: (1) decomposition: “breaking down” the problem or task into smaller steps/parts; (2) abstraction: identifying critical information while ignoring irrelevant or unrelated details; (3) algorithms: developing a step-by-step solution or rules to solve the problem; and (4) pattern recognition: searching for similarities among and within problems, identifying patterns that enable information organization and help in problem-solving tasks.

CT is historically associated with programming skills [85], often approached through programming-oriented activities [86-91], resulting in a lack of valid measures that assess CT without using programming tools [84]. However, CT should not be reduced to formal coding, as it involves broader problem-solving processes, thinking at different levels of abstraction, and transferable competences across contexts [72,73,81,86,92] (including STEM education [75,81]). Games have also been identified as promising approaches to make CT more accessible and engaging [80,93], with game-based learning associated with active learning, motivation, engagement, and problem-solving, particularly in areas often perceived as difficult [94,95]. Playful engagement with CT, particularly in collaborative game-based contexts, may thus provide useful settings for supporting self-efficacy in CT learning [76,77,86], ultimately enhancing engagement in STEM fields.

Our research design is structured as an iterative Design Science Research (DSR) approach, which is characterized by using “design as a research method or technique” [105], and is constituted by five phases: (1) awareness, where a new problem is perceived and characterized, raising interest to discover the “missing knowledge” [105]; (2) suggestion, where design ideas are formulated to explore the problem through a design proposal; (3) development, where a proof-of-concept is created to demonstrate the feasibility of the proposal; (4) evaluation, testing the designed artifact following a set of evaluation criteria, where gained information can justify returning to the awareness, suggestion, or development stages for adjustments and improvements; and (5) conclusion, as the end of a research cycle, synthesizing knowledge contribution, innovation value, and application in other scenarios.

In our research, we started by studying the current low levels of self-efficacy and motivation leading to STEM avoidance and dropout (awareness of problem), as well as design inspirations and concepts to gather insights into how cooperative behaviors could enable the empowerment of the participants during the gameplay activity, by developing competences that would increase their ability to perform in-game. A game design proposal was then formulated (suggestion) following a playful probing approach [96], focused on using play activities for data collection to investigate how participants engage with the artifact in a playable context. For that, a low-resolution paper prototype of the game was produced and used as a proof-of-concept of the playful probing approach (development) to perform gameplay rehearsals. The rehearsals were video-recorded, and individual semistructured interviews were conducted at the end of each session to gain further insights into the player’s perception and analyze modes of engagement (evaluation).

This study reports the evaluation of one iteration of the DSR process. It presents a cross-sectional study involving 5 playtesting sessions, analyzing data from a sampling population at a single point in time, aiming to find game design insights on the relationship between collaboration and CT-related self-efficacy. A qualitative study with the target audience was conducted (14 children aged from 6 to 11 years), with 5 gameplay rehearsal sessions and individual semistructured interviews, from which we sought evidence of how collaborative activities can support self-efficacy toward CT.

The qualitative content analysis of player behavior was based on an axial coding of categories related to meaningful events of competence development, collaboration, and self-efficacy sources. The axial coding was cross-checked and discussed among the 3 authors for consensus and consistent coding. We interpreted transitions in which players initially expressed difficulty but, through collaborative assistance, overcame them and performed new gameplay actions as a potential gain in self-efficacy, which we discuss as evidence of a relationship between these categories. In the end, we drew design insights for collaborative gameplay activities, supporting self-efficacy development in CT. The process is schematically explained in Multimedia Appendix 1.

We applied a low-resolution game prototype based on the work of Pinto et al [97,106], which proposed a cooperative design where players care for a meerkat colony, inspired by the collaborative social organization of this animal species. In this study, we provide an overview of the game, with a special focus on the design elements that support collaborative activities within the game.

Regarding the game’s theme, given children’s affinity with animals [107,108], and the intention of a collaborative environment, we sought species with a natural social organization and cooperative attitudes. Meerkats provided a suitable solution, as their cooperative social structure influenced both the game’s setting and the actions available for the players to perform. This led us to define 3 core functions [109,110], including territory exploration and development, defense against threats, and colony sustenance. As collaborative behavior was central to the design, the game was organized around a shared goal (the growth of the colony), which encouraged players to cooperate throughout gameplay.

The prototype was conceived as a collaborative environment for exploring CT through play, connecting its gameplay structure to STEM. Rather than introducing these competences through formal programming syntax, the game was designed so that children could engage with them implicitly by collaboratively defining the behavior of the meerkat entity through condition/action/result. In this way, the 4 CT dimensions (decomposition, abstraction, algorithmic thinking, and pattern recognition) were embedded in a playful and accessible activity [106].

The game design incorporated several activity modes [37,111] in a collaborative environment to promote mutual assistance in solving problems. Players participate in the game by defining the meerkat’s behaviors [106] as they move their characters on the board, roll the dice for action points, decide, and act to explore the space, gather resources, deal with dangerous encounters, and share action points with other players.

In the game, players need to define the behaviors of meerkats by coding action sheets with preconditions, actions, and postconditions (or results). The construction of the meerkat actions is based on Petri-Net models in games [106,112] by associating their structure with the programming components for constructing the game characters’ actions (Figure 1). The process is carried out by defining conditions (Figure 1A) that, when verified, trigger or enable actions (Figure 1B) that produce results (Figure 1C), in the form of information or resources that can possibly satisfy other conditions.

The game’s actions operate when the preconditions are satisfied, either through the existence of the element in the game (inserted in the condition) or the direct contact of the meerkat with it (eg, the player finds the element hidden in the main board). This process triggers the action associated with this precondition, thus producing the player-specified result on the main board. Through this action-construction process, players practice competencies associated with CT skills [106].

After defining the dynamics of the meerkat’s actions, the authors considered ways to design for collaborative gameplay with different levels of collaborative activity dynamics (Figure 2). For that, the 3 activity theory levels and their dynamics in collaborative activity [71] were considered.

Co-construction level: in order to facilitate social interaction and collaboration within the group, it is essential that its members share a common goal. For that, we proposed the shared purpose (the increase of the colony members).

Co-operation level: the game objective cannot be achieved alone, demanding interdependence among players. As such, the player’s panel for defining the meerkat behavior is shared, where players need to cooperate to discuss, distribute, and collaboratively define the actions/tasks that each meerkat will perform on the main board to achieve the overall objective of the collective activity.

Co-ordination level: task routinization and communication are important components for the game’s resolution, promoted by the game’s activities and the shared player panel. Players are required to communicate and agree on the definition of actions, coordinating the execution of these defined actions on the main board to achieve the game goal.

Each individual, after discussing and defining the player panels between all, will focus only on their function and the successful completion of their task on the main board. Considering that, we developed the following activities/tasks: (1) movement on the main board; (2) exploration of cells on the main board; (3) creation of burrows/tunnels on the main board; (4) collection of food on the main board; (5) unloading the meerkat food; (6) confronting predators/invaders/thieves on the main board; and (7) territorial control.

The main objective of players in this game is to grow the number of members in their colony. To accomplish this, co-operation is essential, as players must work together to define the meerkats’ behaviors, specifying them in each round and using the dice to take actions. Collaboration between players may manifest in the form of coordinating meerkat behavior coding, working toward shared goals, or sharing action points (dice values) between them. The game’s design is intended to encourage activities that promote communication and collaboration in order to develop effective strategies, learn how to code useful behaviors, and influence one another’s decisions to achieve common goals.

The game prototype consists of a player panel (Figure 3A) for coding the meerkats’ actions and a main board (Figure 3B), which defines the scenario where the meerkats enact the player’s choices.

The player panel was designed mimicking the Petri nets’ graphic structure [112]: it is formed by a set of cards, each with a condition area (with up to 3 possible conditions), one main action to be performed, and a result area (enabling up to 3 possible consequent results). This panel design gives the player freedom to set various conditions for the action to be enabled, combining different condition-action-outcome pairings to implement their strategy.

Considering these mechanics, we developed actions, based on the meerkat’s natural behaviors (Figure 3D), as well as elements to signify conditions or outcomes (Figure 3E). We also created visual components of the colony (Figure 3C), namely (1) a card with colony members, for keeping track of its size, (2) a card for storage, representing the players’ deposit of collected food, and (3) a meerkat inventory, to register the amount of food they carry, both in their hands and in the bag. Finally, enemies were also defined (Figure 3F).

The main board consisted of 2 overlapping A3 paper sheets, with the upper sheet divided into a 6×8 grid of cut cells to reveal hidden game elements. This design enables hiding resources and opponents on the main board, which are activated through actions defined in the player panel when discovered by the player. Trees and shrubs were also created to visualize food rewards, encouraging exploration.

Concerning game mechanics, players could use up to 8 behavior cards, which they collaboratively coded in each round to define the possible meerkats’ actions on the main board. Executing these actions depends on the discoveries made, as players explore the cells on the main board or by making appropriate changes in the game context (eg, to unload food requires excavating a burrow).

A round consists of three distinct phases (illustrated in Figure 4): (1) defining the actions on the player panel; (2) rolling the dice; and (3) the meerkat executing the actions (as defined on any players’ panels).

The game begins with 4 behavior cards on the player panel, 2 actions (walking and excavating), and 4 elements (directions, steps, favorable ground, and burrow). Additional elements and actions are unlocked gradually based on discoveries made during the game. Each player rolls a die, and the combined total determines the number of actions and movements the meerkats can perform in the round. Players can distribute these actions among themselves as desired.

The game concludes when all cells have been explored, all food (cells, trees, and bushes) has been collected, and all enemies are confronted on the main board.

Analyzing the set of evidence, we could observe how participants communicated and collaborated to overcome challenges and achieve the game’s goal of increasing the colony. Players engaged in a cooperative dialogue to determine the most beneficial actions and how to execute them (co-operation level [71]). This process allowed errors to be identified, which encouraged mutual assistance and reflection on how to solve them. This created multiple opportunities for supporting mastery experiences through verbal persuasion and vicarious experiences [12,13], influencing the learning process and confidence in subsequent activities. Interactions at this stage through the mutual support enabled the accomplishment of the game’s goal, leading to players sharing their available actions (represented by the dice roll each round) and eventually coordinating their tasks (co-ordination level [71]). In this illustrative case (Session 1), we can observe the influence of collaborative dialogue on the action construction in the player panel:

Later in the game, P2 encountered a more complex scenario than the one previously presented, when a series of events are meaningful as sources of self-efficacy:

While engaging in cooperative gameplay, we observed players discussing and clarifying their thinking about how to define creature behaviors to better address challenges posed in the game environment. The following evidence (Session 5) exemplifies this clarification regarding the dynamics of the condition-action-result (algorithms: CT [82-84]), highlighting the importance of dependencies in the condition area through verbal persuasion [12,13]:

A scenario from Session 3 demonstrates an illustrative case on how the collaborative activity created opportunities for self-efficacy. This influence leads to vicarious experiences and verbal persuasion, resulting in the mastery experience of the player [12,13]. In the following sequence of evidence, collaboration between players is observed in defining the task through discussion and signaling of elements for construction:

A last illustration from Session 2 demonstrates how the game challenged players to modify the initial rules (co-construction level [71]), fostering collaboration within the group:

Overall, collaborative behavior was observed across all gameplay rehearsals. However, in 2 groups, collaboration coexisted with competitive dynamics, which appeared to take different forms depending on the participants’ age. In Session 4, involving 9-year-old participants, competition took a more playful and friendly form and did not appear to affect collaboration. Rather, participants challenged one another while still working toward collaborative goals, with the idea that whoever completed the task first would be the “winner”. Gameplay was therefore characterized by social interaction that supported collaboration while also fostering beneficial competition. Playful attitudes, such as teasing and joking (emotional arousal [12,13]), reinforced both collaboration and competition. This balance created a dynamic gameplay experience, in which participants supported each other while also seeking individual recognition, blending mutual assistance with personal achievement. By contrast, in Session 1, involving participants aged 6-7 years, the competitive dynamic was less playful, observing a stronger sense of ownership over individual resources, which in some situations led to conflict. However, these difficulties were related less to the collaborative nature of the activity itself than to the sharing of resources (perceived as a personal decision). When the task focused on constructing actions on the player panel, players collaborated by supporting each other with guidance on what to do. They appeared willing to collaborate through ideas, encouragement, and procedural support, while showing greater reluctance to share actions and dice values. This became particularly evident when players were asked to share the value obtained from a dice roll, or when the dice was exchanged between them. Although all the dice used in the game were identical, P3 claimed that P2 had taken her dice and attempted to retrieve it from P2’s hands, which resulted in a conflict between the players, with P3 even beginning to cry. Textboxes 6-8 illustrate these interactions.

The previous example was the main situation in which age appeared to have influenced collaboration dynamics, particularly regarding difficulties in sharing objects or resources perceived as their own. However, the limited sample size does not allow stronger conclusions to be drawn regarding age-related effects, and no other meaningful event regarding age difference was observed beyond this case. Despite this conflict, players’ feedback remained positive, suggesting that these tensions did not affect their overall perception of the game. All players stated that they liked the game and considered it to have been designed for them, praising its fun nature, the challenges involved, and the fact that it was played together with others, which is especially relevant, given that they consistently expressed a preference for collaborative play. Overall, these findings suggest that the game was well received across the sample’s age range represented. This positive reception was further reflected in the interviews concerning collaboration and self-efficacy.

During interviews, players’ perceptions aligned with observations made during the sessions, indicating that collaboration was an important problem-solving factor that influenced their in-game behavior. When asked if observing another player helped them resolve game challenges (vicarious experience), all players (n=14) responded positively. Eight players mentioned that witnessing a fellow player’s actions inspired them to emulate those actions, with comments such as “Seeing others playing and building actions helped me understand how I could also do it” and “I saw my colleagues playing, and over time I tried to do the same”, further illustrating how observing others’ gameplay influenced their own strategies and actions.

Additionally, players mentioned that collaborative dialogues also played a significant role. Five players highlighted verbal suggestions and incentives during gameplay (verbal persuasion), with comments such as, “Yes, because she spoke with me” and “Yes, because that is a group game where everyone helps. They encouraged it”. The interviews highlight the potential impact of collaboration in supporting different forms of self-efficacy, facilitating effective gameplay and problem-solving.

To develop a collaborative design, it was necessary to investigate collaborative processes, particularly the 3-level dynamics of collaboration [71]: co-construction, co-operation, and co-ordination. Understanding the characteristics of each level and the notion of transitions between levels led to the need to develop mechanisms that support them. In our game, this was achieved by having the shared player panel and associated actions, which create opportunities for player mutual support and prevention of breakdowns of the collaborative activity.

For effective collaboration, it is essential to share a common goal that fosters social interaction, communication, and collaboration among group members [45,53], integrating individuals’ prior knowledge and skills developed toward that joint objective. In our game, the colony’s survival as a collaborative goal results in the development of collective comprehension and knowledge based on each other’s ideas and thoughts (through vicarious experiences and verbal persuasion [12,13]), ultimately leading to problem-solving success. Furthermore, the pursuit of this common goal creates interdependence among group members, as the objective of the activity cannot be achieved alone [45,53]. Thus, all members must contribute collaboratively, recognizing that their own success depends on their peers.

We suggest a game goal focused on a meerkat colony theme, in which players must collaborate to achieve it (co-construction level [71] - increase of the colony), establishing social norms, assuming diverse roles and actions in the game context (co-operation level [71]), cooperating to define them in the player panel, and distributing their coordinated execution among them (co-ordination level [71]). Such a collaboration context fosters a sense of community between the players, which is further supported by the following shared game elements and mechanics (inspired by the shared elements of [45,53]):

These collaborative design elements and gameplay mechanics interconnected player activities and encouraged them to think together, communicate effectively, and coordinate their efforts, and above all, support each other’s experiences, subsequently creating opportunities for self-efficacy development regarding the skills required for in-game success.