A contingency perspective on learning and instruction in physical education

Introduction

The heterogeneity of learning groups presents a major challenge for physical education (PE) teachers. Concerning the ways in which teachers instruct, academics in general education (Hardy et al., 2019; Parsons et al., 2018; Van de Pol et al., 2010) as well as in the areas of PE and sport pedagogy (Dyson, 2014; Goodyear and Dudley, 2015; Grenier and Yeaton, 2012; Mosston and Ashworth, 2008; Wibowo, 2020) report that it is essential for teachers to adapt their instruction to the diverse needs of their learning groups. This is especially important for student-centred learning environments such as Cooperative Learning (Dyson and Casey, 2012; Wibowo and Bähr, 2018). From a socio-constructivist view, heterogeneity is a reality of every learning group, since each person has their own history and view of the world (Rovegno, 2006; Wibowo et al., 2014). Supporting learning in a heterogeneous classroom does not mean finding a one-size-fits-all strategy but finding a solution for ‘the dilemma of teaching individuals within groups’ (Corno, 2008: 162). To develop such solutions, teachers must have multiple instructional strategies at their command (Darnis-Paraboschi et al., 2005; Metzler, 2017; Mosston and Ashworth, 2008). This seems to be challenging for teachers, since research indicates that teachers tend to prefer strongly controlling their students’ learning processes (Bähr and Wibowo, 2012; Seidel and Shavelson, 2007). Although a controlling teaching style might be conducive to learning in certain situations, it must be assumed that the students’ differences are not adequately considered in long-term controlling teaching. This suggests that there is still potential for development on the part of teachers in adaptive teaching.

In the present article, we follow a contingency perspective that is used in research on scaffolding (Pea, 2004; Van de Pol et al., 2010; Wood et al., 1976) and research on adaptive teaching (Hardy et al., 2019; Parsons et al., 2018) to foreground features that are important to investigate contingency in PE. To pursue this aim, we first consider contingency from a socio-constructivist perspective. Second, we review empirical research from the field of PE that considers a contingency perspective on learning and instruction. Third, we depict contingency dimensions, instructional continua, and contingency rules as central features of investigating contingency in PE. Since the construct of autonomy plays a major role in a contingency perspective on learning and instruction, it will be illustrated in depth. The methodological issues and future research questions of this contingency perspective are discussed in the final sections.

Contingency from a socio-constructivist perspective

The basic idea of a contingency perspective on learning and instruction is that different forms of instruction (e.g. question prompts, feedback, and peer instructions) have different effects on different students. Therefore, an instructional strategy needs to be weighted depending on learners’ needs. In this sense, contingency expresses the quality of the relationship between learning and instruction. In other words, contingency expresses the extent to which an instructional strategy fits learners’ needs and supports their learning processes.

On a theoretical level, contingency can be conceptualised by the socio-cultural approach and the model of the zone of proximal development (Vygotsky, 1978) as has been demonstrated by Wood et al. (1976) and Wood et al. (1978) in their research on scaffolding. Vygotsky (1978) assumes that everyone has an optimal learning zone, which is slightly above their actual skill level (the zone of proximal development). Wood et al. (1978: 133) call this zone the ‘region of sensitivity’. Support that is located outside of the zone of proximal development (above or below the developmental level) is therefore not likely to foster learning because it demands too little or too much from the students.

How does this inform our understanding of heterogeneity in a class? In a PE class, a teacher needs to consider multiple zones of proximal development. Furthermore, the zone of proximal development is an intra-individual dynamic, sometimes termed a non-linear approach to instruction and learning (Farias et al., 2018). This means that the zone shifts within the continuous learning processes of the students. Teachers need to take both sources of heterogeneity into account when planning and supporting learning processes – inter-individual differences and intra-individual dynamics. Since the aim of teaching, adopted by this study, is that students can solve a task on their own, autonomy is at the core of contingency (Van de Pol et al., 2010). This is highlighted by the characteristic of the contingent transfer of responsibility in research on scaffolding: ‘Via contingent fading (…), responsibility for the performance of a task is gradually transferred to the learner. Responsibility (…) can refer to students’ cognitive or metacognitive activities or to students’ affect’ (Van de Pol et al., 2010: 275).

According to this understanding, contingency expresses the extent to which a teacher’s instruction supports students to take as much responsibility for learning activities as possible (Van de Pol et al., 2010). Therefore, learning in the sense of progress can be tied to increasing students’ responsibility-taking for certain learning activities, which in turn results in students’ increased autonomy. Moreover, contingency can be differentiated into higher (support within the zone of proximal development) or lower (support outside of the zone of proximal development) levels.

A contingency perspective is distinctly different from an instructional perspective (Hardy et al., 2019). While instructional perspectives focus on specific instructional strategies, such as question prompts or peer learning (Ge and Land, 2004), a contingency perspective focuses on the relationship between learning and instruction. A question prompt might be valued as contingent for one student but not for another, or it might be contingent for one student in one task but not in the following one. Therefore, contingency is seen in research on scaffolding (Van de Pol et al., 2010) and adaptive teaching (Hardy et al., 2019) as a relational unit to which both students and teachers contribute. The idea that students influence the flow of the lesson in terms of their learning processes and instruction is further investigated in research on self-regulation (Boekarts and Corno, 2005), which also emphasises autonomy as a central aspect of learning. We must ask how a teacher can influence contingency in the classroom, but it is also crucial to regard the contribution of teacher and students from a contingency perspective. Research on how teachers adapt their instructional strategies to learners’ needs (Parsons et al., 2018) is important, but this research does not account for a contingency perspective if teachers do not explicitly regard students’ contributions. If an investigation only regards the frequencies of scaffolding or adaption means (Hardy et al., 2019; Van de Pol et al., 2010), an instructional perspective, not a contingency perspective, is presented.

The following section illustrates how contingency has been investigated in PE research; however, it is not an exhaustive review of empirical studies on contingency in PE. The findings point to important future challenges for contingency research in PE. For more general reviews on contingency in the area of scaffolding and adaptive teaching, see Hardy et al. (2019), Parsons et al. (2018), and Van de Pol et al. (2010).

Empirical research on contingency in PE

There is a paucity of research on contingency in PE. This section aims to provide some specific examples of empirical approaches from PE literature that provide direction for investigating contingency in PE in the future.

Casey and Dyson (2009) have identified several instructional strategies that aim at contingency in the classroom in terms of planning processes in a Cooperative Learning unit in tennis. The study focused on a teacher’s practice that involved implementing innovations (Cooperative Learning and Tactical Games) in his lessons through action research. The findings provide insight into the teacher’s perspective on contingency and suggest that different sources influenced contingency on the planning level. The first source for contingency was general knowledge about students (e.g. about social conditions and relations within the class). Experience from prior lessons was the second source of contingency in planning decisions. Casey and Dyson (2009) observed the teacher’s different foci on the students’ learning processes; for example, to support the students’ autonomy/independence, the teacher planned additional learning material/worksheets. Another dimension of contingency can be seen in the students’ habits as learners and their knowledge about the processes of the two different model-based practices used. In this study, the teacher planned an extra lesson, a crash course on ‘how to be taught this way’, so that the students could work according to the prepared tasks (Casey and Dyson, 2009: 185). Casey and Dyson took a holistic viewpoint and suggested looking at instruction more broadly, considering social processes while teaching and learning physical skills in PE. The results indicate an interdependent relationship of different dimensions, such as social aspects and subject matter development. For example, the authors report that the teacher in this school-based study structured the learning processes more clearly after the first lessons, which influenced the autonomy and social processes of the students.

Another area of inquiry that foregrounds the dynamic nature of contingency is based on French didactique research (Amade-Escot and Venturini, 2015; Wallhead and Dyson, 2016). The centre of this research is the joint action in didactics (JAD) framework, which focuses on three analytical units for investigating teacher–student and student–student interactions (Sensevy et al., 2005). First, the mesogenesis unit documents the learning environment such as material or symbolic and social objects and development over time. Second, the chronogenesis unit ‘documents the evolution of the knowledge proposed by the teacher and studied by the students, as it unfolds during the joint action’ (Amade-Escot and Venturini, 2015: 420). Third, the topogenesis unit documents how the responsibility for the learning activities is divided between the teacher and students. These three analytical units support a fine-grained understanding of student–teacher social interactions. Amade-Escot and Venturini (2015) used the JAD framework within a case study research design and reported that strong teacher control over the knowledge construction (chronogenesis) and the distribution of responsibility (topogenesis) might lead to a less holistic (tactical meaning) and more fragmented (technical) knowledge of the students. Wallhead and Dyson (2016) investigated the co-construction of three fifth-year students and their teacher within a quasi-ethnographic framework in a Cooperative Learning setting. The findings indicate that the breadth of knowledge development (chronogenesis) might be reduced as a result of the initiatives taken by student leaders. Furthermore, the authors observed that increased temporal responsibility-taking by the teacher (change in topogenesis) can compensate for such developments in the co-construction process. Knowledge development (chronogenesis) was also fostered through clear and consistent expectations of role responsibilities (topogenesis) and more broadly defined parameters for the intended content-elicited ‘students’ actions that move the chronogenesis forward’ (Wallhead and Dyson, 2016: 320).

Based on the US discourse on pedagogical content knowledge (PCK) ¹ , Ayvazo and Ward (2011: 675) represented PCK as ‘the teacher’s ability to pedagogically adapt content to students of diverse abilities’. It is important to note that these authors take a behaviourist perspective that is distinct from the socio-constructivist perspective on contingency outlined earlier. Nevertheless, from a methodological standpoint, similarities emerge in these perspectives’ focus on the relationship between learning and instruction (contingency in our terms and appropriateness in the terms of Ayvazo and Ward, 2011). In their study, the researchers aimed to investigate how PCK differed when teaching stronger and weaker instructional units. Following Cooper et al. (2007), Ayvazo and Ward (2011) used behaviour analysis procedures to code the students’ performances, teachers’ adaptations, and subsequent students’ performances. Furthermore, in the teaching analysis, they coded teacher behaviour as appropriate or inappropriate (Ayvazo and Ward, 2011). Appropriateness must be a central element in our understanding of contingency since it expresses the relationship between the teacher’s instruction and the students’ learning processes. The results indicate that contingency in the classroom is higher when teachers are more familiar with the content.

In two subsequent studies, within a mixed-methods design, Wibowo (2015, 2019, 2020) analysed teacher–student interactions in a Cooperative Learning setting from six fifth-year classes over six weeks. In the first qualitative study, autonomy and problem-solving were identified as processes relevant to contingency in PE (Wibowo, 2015). Autonomy was defined as the capability of students to execute certain problem-solving activities with different levels of support from the teacher (Deci and Ryan, 1985; Van de Pol et al., 2010). Here, problem-solving was defined as analysing, planning, and testing (Adden et al., 1978; Greiff et al., 2013). Furthermore, for autonomy and problem-solving, contingency rules were derived from literature to identify contingency units (student–teacher couples) as contingent or non-contingent. In a second quantitative study, a coding procedure was developed to measure contingency in PE in a standardised approach, which was piloted with the six chosen classes over six weeks. The results indicated that 60.9% of the contingency units were rated as contingent in the dimension of problem-solving compared with 50.1% contingent contingency units in the dimension of autonomy (Wibowo, 2019, 2020).

In a recent study, Farias et al. (2018) investigated contingency in the context of Sport Education. The focus of the study was on three student coaches and their teacher within an action research approach over a year-long programme, comprising four seasons of Sport Education (basketball, handball, soccer, and volleyball). The data included lesson observations, semi-structured interviews, and the teacher/investigator’s field diary. Similar to Casey and Dyson (2009), Farias et al. (2018) focused on contingency on a planning level as well as an interaction level. Since peer coaching places a great demand on student coaches, the autonomy of the student coaches was a central focus of the study. Autonomy was also a focus in three of the four previously reviewed studies. Furthermore, in contrast to the other studies, Farias et al. (2018) focused on student coaches’ metacognitive processes in terms of how they instructed their peers. The findings indicated that teachers’ intense support in the first season upon the failure of the student coaches to support their classmates’ learning processes led to increased responsibility-taking by the student coaches in the following seasons. Despite responsibility-taking in quantitative terms (more or less intense support), the authors regarded the teacher’s instruction in a qualitative way, outlining which processes were initiated by the teacher. In the beginning, the students had trouble with single leadership activities (e.g. observation and diagnosing errors of other students) and the teacher focused his attention on supporting their development. After a more controlling support style at the beginning, the teacher reduced control over the student coaches’ instructional processes when they showed increased leadership skills. Concerning the development over the whole year, the researchers stated that scaffolding was ‘a highly complex, and far from linear process. It was not a question of simple gradual withdrawing and the replacement of procedural guidance with discovery-based strategies. Rather, the scaffolding was shown to be a dynamic process’ (Farias et al., 2018: 289).

In the following sections, we outline three central features that build a framework for investigating contingency in PE: contingency dimensions, the holistic nature of instruction, and contingency rules.

Features of a contingency framework

Contingency rules

To connect students’ learning processes and teachers’ instruction, research on contingency in PE should explicate how and why a concrete instructional action is seen as supportive of a concrete learning activity. This has been termed ‘contingency rules’ (Van de Pol and Elbers, 2013; Wood et al., 1978). One example of such principles is the contingency approach that stems from research on scaffolding (Van de Pol et al., 2011; Wood et al., 1976). According to this perspective, instruction should activate students within their ‘region of sensitivity’ (Wood et al., 1978: 133), that is, one step above the students’ level of success: ‘If the child succeeds, when next intervening offer less help. If the child fails, when next intervening take over more control’ (Wood et al., 1978: 133). In the field of PE, Wibowo (2015, 2020) applied this principle to build contingency rules in the contingency dimension of autonomy. For each of the four levels of students’ autonomy, instructional actions were identified as contingent or non-contingent. For example, when student activities were categorised as low autonomy, the contingency unit was coded as contingent if the teacher instruction was categorised as highly controlling. Farias et al. (2018) identified contingency from an empirical approach and identified high contingent situations based on the students’ learning progress (e.g. in the complexity of the domain-specific content). Contingency rules are at the heart of contingency. They are the key that relates certain instructional strategies to learning processes. Therefore, identifying contingency rules from literature and empirical research is a major concern for research on contingency in PE.

The above figure illustrates our contingency perspective in PE.

We illustrate these three features of contingency (contingency dimensions, holistic nature of instruction, and contingency rules) and how they are connected, as shown in Figure 1. In the next section, we address the dimension of autonomy to give one specific and concrete example of how contingency might be operationalised in PE. Autonomy was chosen because it has the character of a core dimension from a contingency perspective and is seen as a crosswise category.

OPEN IN VIEWER

Figure 1.

Three units of analysis determining contingency in PE and their dynamic development over time (own presentation).

Autonomy as a central dimension of contingency in PE

Autonomy is important for contingency in PE for several reasons. First, learning to act responsibly for oneself and with others is an important part of education within and towards a democratic society (Dewey, 1916; Prohl and Ratzmann, 2018). Second, drawing on the theory of Bildung from the German academic tradition of PE, humans are open beings and therefore must actively shape their relationship with the world (Gröben and Prohl, 2012; Klafki, 2000, 2001). Taking responsibility and acting autonomously is connected with forming a subjective relationship with the world through movement and sports (Wibowo, 2015).

As depicted above, autonomy functions as a crosswise category. In other words, autonomy means ‘taking responsibility for something’, for example, ‘to solve a problem’ (Wibowo, 2019), ‘to support peers’ (Farias et al., 2018), or ‘to master content’ (Ayvazo and Ward, 2011; Wallhead and Dyson, 2016). To illustrate the crosswise function of autonomy, we draw on an example from our work on contingency in Cooperative Learning settings in PE (Bähr and Wibowo, 2012; Wibowo, 2015). In the example, the student (Jannika) starts with low autonomy in learning to do a handstand within a Group Tournament (Slavin, 1994). The teacher supports her with a low controlling instructional style and, subsequently, Jannika’s group manages to help her without the teacher’s further support (Wibowo, 2015). The low controlling teaching style enables the students to take responsibility for solving Jannika’s problem by analysing her movement and planning further steps to assist her.

Initially, Jannika approached the teacher with a very low level of autonomy. Neither she nor her group members managed to improve her handstand. She analysed the problem (‘I always have too much momentum’), but it did not help her find a solution; in other words, she was not able to take responsibility for learning how to do a handstand. The teacher took over some of the learning process (‘Well, let us try one and we will look carefully’). Here, the teacher initiated the analysing processes within the group without adding any content information on how to perform a handstand. Anne and Max adopted this information and added content. First, they analysed the handstand (‘You are going in the wrong direction. You are doing this’) and then gave her advice on how to perform a handstand correctly (‘But you need to do it like this’) – Max took responsibility for teaching Jannika to develop the appropriate skills to perform a handstand. Furthermore, Anne suggested she could assist Jannika in performing her handstand. Here, Anne planned future actions without support from the teacher. The teacher withdrew from the situation and asked the students if they could take responsibility for themselves. Jannika did not seem to be sure but signalled that they would try. This specific analysis of the situation connects autonomy and problem-solving.

Before we look into contingency rules in this situation, autonomy, as a core contingency dimension and an instructional control continuum for autonomy support, should be depicted in more detail. Several research strands have been used to operationalise autonomy as a dimension of contingency in PE (Wibowo, 2015). In research on self-directed learning (Bolhuis, 2003) and self-regulated learning (Boekarts, 1999), it is understood that pupils can regulate their learning processes. This perspective was used to operationalise which actions students use for their learning progress. While these two approaches take a rather process-oriented perspective, autonomy also encompasses content in terms of knowledge or skills (Bolhuis, 2003). We propose using both aspects (process and content) to conceptualise autonomy in small workgroups. Furthermore, the support of the teacher for both aspects was considered, and autonomy was defined by the extent of responsibility (for processes or content) taken by the students and by the teacher (Wibowo, 2015; see Table 1).

OPEN IN VIEWER

Table 1.

Autonomy as a contingency dimension and respective control continuum (Wibowo, 2015).

	Student Autonomy		Teacher Control
	Regulation	Content	Regulation	Content
4 – high responsibility-taking	Changing a problem-solving activity	Providing full content information	Activating problem-solving activity	Providing full content information
3	No regulation; following the teacher’s guidance	Providing full content information	Activating a problem-solving activity	Providing partial content information
2	No regulation; following the teacher’s guidance	Providing partial content information	Activating a problem-solving activity	No content provided
1 – low responsibility-taking	No regulation; following the teacher’s guidance	Copying the teacher’s support	No activation of a problem-solving activity	No content provided

A problem-solving perspective was used to operationalise the students’ processes because different academics in the field of PE suggest that problem-solving is a useful heuristic for learning processes in PE (Brodtmann and Landau, 1982; Messmer, 2018; Mosston, 1968; Wright et al., 2004). Following relevant research on problem-solving (Adden, et al., 1978; Ge and Land, 2004; Greiff et al., 2013), the process of problem-solving was defined as a connection of analysing, planning, and testing (Wibowo, 2019, 2020). According to his perspective on self-directed lifelong learning, Bolhuis (2003) regards processes and content as equally important. Therefore, high autonomy is present when students regulate their problem-solving processes and when they provide the full content information without prior support by the teacher. Here, students’ autonomy is tied to the teachers’ responsibility-taking (operationalised by the control continuum in Table 1), because if a teacher permanently provides full content information, the students’ potential to take responsibility is restricted. Table 1 summarises the different stages of student autonomy and teacher control. The different stages were defined inductively in a qualitative study based on the analysis of student–teacher interactions (Wibowo, 2015).

The students in the example above demonstrated the highest level of autonomy when they began to plan support for their peer Jannika, after the teacher had initialised the students’ activities without providing content. Had the teacher provided partial or full content information (e.g. for partial content information, if the teacher had said ‘Well, let us try one and we will look carefully at the position of your hands’, or for full content information, ‘Your problem is that that your arms are not completely straight when you do the handstand, and then you buckle’), he would have taken more responsibility for the learning processes and thus left fewer opportunities for the students to take responsibility.

How can we estimate contingency between different levels of autonomy and the outlined control continuum for supporting autonomy? To define the relationship between students’ activities and teachers’ instruction in the dimension of autonomy, Wibowo (2015) used the contingency approach from research on scaffolding (Van de Pol et al., 2011; Wood et al., 1978). Applying the idea of activation in a ‘region of sensitivity’ (Wood et al., 1978: 133), the teacher is seen as contingent when he or she asks the students for a level of responsibility-taking one step higher than what they displayed earlier. Accordingly, when a student successfully executes a problem-solving activity after being supported by the teacher through a process of regulating instructions (autonomy level 3 in Table 2), the teacher would be contingent by giving no more support in the next intervention (control level 1, upper row and next to autonomy level 3 in Table 2).

OPEN IN VIEWER

Table 2.

Contingency rules for the dimension of autonomy (defined through the regulation of processes and content contribution, according to Wibowo, 2015).

Autonomy level of the students		Contingent, when the teacher’s subsequent control level is…
4 - problem-solving without support	successful	1 - no help with processes or content
	unsuccessful	2 - process regulation without content
3 - analysing, planning, or testing after being activated to do so	successful	1 - no help with processes or content
	unsuccessful	3 - supports students with partial content information
2 - analysing, planning, or testing after being activated through partial content information	successful	2 - process regulation without content
	unsuccessful	3 - supports students with partial content information
1- analysing, planning, or testing after being activated through full content information	successful	3 - supports students with partial content information
	unsuccessful	4 - supports students with full content information

According to this formulation, in the example, the teacher’s last intervention would be seen as contingent because the students acted previously on autonomy level four and the teacher responded with no support for process or content. If the teacher had tried to regulate the students’ learning processes or provided content, he would have reduced the possibility of responsibility-taking for the students, which would be categorised as less contingent.

This provides one perspective and a possible operationalisation of contingency in PE within teacher–student interactions. Through the contingency rules displayed in Table 2, which are based on the scales in Table 1, it becomes possible to rate teacher–student interactions in terms of contingency on a rational basis. This allows researchers to gain insight into how to deal with students’ heterogeneity, a central feature in PE. Our intention here is to present explicit examples of contingency in PE and illustrate the importance of elucidating which learning processes are focused as contingency dimensions and how different instructional strategies converge in an instructional continuum that is related to the learning processes of the students. Finally, the contingency rules are the heart of our understanding of contingency, since they express the quality of the support and helpfulness that affect the students’ learning processes.

After illustrating contingency in PE within the dimension of autonomy, we want to draw attention to some methodological issues of investigating contingency in PE.

Methodological issues

Investigating a complex phenomenon such as contingency in PE raises several methodological questions concerning the analytical units used, data collection, and research methods.

In the previous section, we presented a microanalytical view on contingency in PE, but there are different approaches to constructing contingency units. Where Van de Pol and Elbers (2013) used teacher–student–teacher triplets, Ayvazo and Ward (2011) used student–teacher–student triplets and Wibowo (2019) used student–teacher couples. Other approaches in general education advocate for less fine-grained units; for example, Hermkes et al. (2018) developed an approach in which the contingency unit was built from an interaction sequence starting from the approach of a teacher to a group or an individual and ending when the teacher leaves. They argued that the microanalytical approaches (interaction triplets or couples) do not grasp contingency in its full form. What are the consequences of the different resolutions concerning the analytical units? Since both perspectives are still new and barely investigated, the potential of each perspective must be uncovered. Both approaches are based on the formulation of contingency rules in advance of data collection and rest on collecting observational data.

Other approaches have determined contingency according to the effectiveness of the instruction (Casey and Dyson, 2009; Farias et al., 2018) and draw on data other than observations. In these approaches, an instructional strategy is valued as more or less contingent when the students succeed or fail, respectively, in their learning process. Contingency units such as a mini lesson or learning materials (Casey and Dyson, 2009) or additional seminars on instructional leadership (Farias et al., 2018) are less fine-grained than contingency units from microanalytical approaches that use interaction triplets or couples. Research in other areas (social studies: Van de Pol et al., 2012; mathematics: Wischgoll et al., 2015) has combined these approaches (contingency measured by a priori defined contingency rules and its immediate effects in the learning situation) and seems to be promising since different contingency patterns show different effects. For instance, Van de Pol et al. (2018) found that high contingency is beneficial if students have a low initial content understanding.

A contingency perspective provides a more differentiated analysis of student–teacher interaction than research from an instructional perspective that focuses only on the teacher. One advantage of an instructional approach lies in its possibilities to analyse big datasets efficiently because it uses (only) one analytical unit compared with up to four analytical units in the microanalytical approaches to contingency.

Because there is little scholarly work on contingency research in PE, we suggest using both qualitative and quantitative research designs if possible, in mixed-methods approaches (Dyson, 2014; König, 2016). In our opinion, the reviewed research calls for exploratory research that focuses on deepening our understanding of relevant contingency dimensions and respective instructional continua or formulating contingency rules and investigating them empirically. For quantitative designs, it seems necessary to consider a multi-step coding procedure if using systematic observational data on the micro level, as described in the studies of Ayvazo and Ward (2011), Van de Pol et al. (2012, 2015), and Wibowo (2019). A systematic coding procedure should first focus on the activities of students and the instructions of teachers separately and then focus on contingency in a third step.

Concluding comments

Contingency is a complex construct that focuses explicitly on the dynamic relationship between the teacher’s instruction and students’ learning processes. A contingency perspective values the heterogeneity of learning groups as a challenge for teachers to support each student according to their needs. The purpose of this article was to discuss the value of a contingency perspective for empirical research in PE based on a relationship between the students’ learning activities and the teacher’s instructions. Utilising this contingency framework requires an explicit analysis of the students’ learning processes as well as the instruction by teachers to specify the relationships between them as the central unit of analysis. Furthermore, we separate different dimensions of contingency and foreground autonomy as a central dimension of learning in PE and for investigating contingency in PE.

The contingency framework in PE raises questions that lead to different research directions. (1) We see a need to conduct exploratory research and connect existing research to explore relevant dimensions of contingency in PE as ‘local theories’ (Shavelson et al., 2002), which are specific to a pedagogical perspective in PE. Autonomy was a central contingency dimension in connection with problem-solving processes and content development. However, what does autonomy mean in connection with other contingency dimensions (e.g. students’ social and emotional learning)? (2) A central direction of future research could be related to the effectiveness of contingency in PE. Studies on scaffolding have suggested taking a closer look at the construct of contingency (Van de Pol et al., 2015) and how it is connected to learning results. This seems necessary to gain a deeper understanding of how contingency influences processes within and the outcomes of PE. (3) To bridge research and practice, it is important to investigate how teachers can be best prepared within PE teacher education and professional development courses to implement contingency in PE.

A contingency perspective has the potential to investigate the interconnections between learning and instruction in detail. The central features of investigating contingency (specifying the contingency dimension, explicating a respective instructional continuum, and highlighting the contingency rules) and methodological issues were identified and presented. Future research can draw on these core characteristics and methodological aspects to advance our understanding of contingency in PE. Further, research on contingency in PE that draws on the central features presented in this paper contributes to a better understanding of learning and instruction in the field of PE.