There is a growing emphasis on formative assessment practices in science education to focus students on learning (Zhai 2021). However, despite this increased emphasis, designing and implementing assessments to support science teaching and learning remain challenging, representing a universal struggle rather than a localised issue. To establish this global significance, research has identified significant barriers to formative reform in diverse international settings; for instance, Yan and Brown (2021) noted the persistent dominance of summative assessment practices in Hong Kong schools, a challenge similarly observed in Ireland by Doyle et al. (2024). While these international studies demonstrate that the struggle to implement formative assessment is a widespread phenomenon, they also serve to contrast the unique hurdles within the South African context. Specifically, South African science education assessment practices are still struggling to align with ideal reform-based practices and policy aspirations (e.g. Kibirige & Teffo 2014). These issues are expected because South African assessment policy documents (e.g. the Curriculum and Assessment Policy Statement) lack pedagogical guidance on implementing formative assessment effectively (eds. Reddy et al. 2015). Consequently, South African preservice teachers must navigate a universal ‘apprenticeship of observation’ (Lortie 1975), which is further complicated by a national policy framework that provides insufficient direction for shifting towards formative practices.

While the apprenticeship of observation (Lortie 1975) is a general phenomenon affecting all student teachers regardless of their discipline, its implications are particularly acute for science teacher education in South Africa and internationally. Since preservice teachers do not necessarily enter training with subject-distinct beliefs, they hold general orientations shaped by their prior lived experiences as learners within traditional summative schooling environments. These imbalances, where the focus remains on summative testing over formative practice, are a cause of concern because they perpetuate the belief that grades are more significant than skills enhancement and learning (Xu & Brown 2016). Because science education is currently driven by reform-based aspirations that depend on formative engagement, these general entry-level beliefs must be interrogated and reconstructed specifically within the context of science pedagogy to prevent the reproduction of passive learning cultures (Zhai 2021).

The construction of assessment beliefs is a complex process rooted in a teacher’s personal history and professional context. According to Richardson (1996), there are three main sources of teacher beliefs: (1) personal experience; (2) experience with schooling and instruction; and (3) experience with formal knowledge. Within science teacher education, these sources are fundamentally mediated by the broader socio-pedagogical environment, as Mansour (2009) emphasises the importance of studying the social and cultural perspectives on science teachers’ beliefs as a foundation for understanding how teachers conceptualise their work. The significance of this socio-cultural shaping is evident in specific educational contexts; for example, Brown and Gao (2015) found that Chinese teachers who strongly believed in the importance of summative evaluation were less inclined to use formative assessment. Xu and Brown (2016) regard the assessment beliefs held by this group of Chinese teachers as a misconception, as they reflect traditional practices that are inconsistent with research evidence on effective practice or with current policy expectations.

Brown and Harris (2012) show that learners’ beliefs about assessment align with how assessment is used in their school environments and that these beliefs vary by level of schooling. Consequently, preservice teachers who are still students enter teacher education programmes with assessment beliefs shaped by prior schooling systems dominated by a summative assessment culture. These beliefs can be at odds with views that are adaptive for the promotion of learning (Brown & Remesal 2012; Xu & Brown 2016; Yough et al. 2023).

A noteworthy study by Daniels and Poth (2017) found that preservice teachers with assessment beliefs focused on improving learning were associated with mastery-oriented learning. Conversely, beliefs about assessment for accountability and meeting societal needs were associated with performance-oriented learning. Although their study provided useful information on the impact of assessment beliefs on learning, it was limited in that the regulation of such beliefs on learning was inferred indirectly through the achievement goal theory, using self-reports, and little evidence was known about the exact sources shaping those beliefs about assessment.

The study used a hermeneutic phenomenological design, which enabled participants’ subjective meanings to be established by exploring their lived experiences of assessment, and these experiences were then analysed inductively to identify emerging themes, as Miller and Barrio Minton (2016) explain. In the hermeneutic approach to phenomenology, the affordance of theories provides a way of grasping and understanding the meaning dimensions that correlate with lived experience, which can help interpret the findings (Neubauer, Witkop & Varpio 2019). Hence, the current study adopted a hermeneutic approach, enabling the research to capture participants’ assessment beliefs across three dimensions. These dimensions are the cognitive domain (what a student believes is true or false about assessment), the affective domain (emotional inclinations towards assessment) and the evaluative domain (the value placed on assessment) (Remesal 2011; Xu & Brown 2016).

The study was conducted at a South African university that offers a Bachelor of Education programme. Participants were purposively selected during the first semester from a cohort of first-year preservice science teachers and at the beginning of the second semester from a cohort of final-year preservice science teachers. Both cohorts were enrolled in a Bachelor of Education degree in Natural Sciences Education. The inclusion of first-year preservice science teachers was intended to maximise response variation and identify patterns within it, as first-year students are at the beginning of their training and final-year students are at the end of theirs. This variation in participants’ experiences enabled the researcher to make broader claims about the phenomenology at work, as Bordogna (2021) explains.

An independent person recruited 10 first-year participants on the basis of their willingness to reflect on and discuss their assessment experiences. Four of the 10 participants were male, and six were female, and their ages ranged from 18 years to 21 years. Nine final-year participants were recruited, comprising three male and six female. Thus, a total of 19 participants took part in the study. The participants came from different ethnic groups and socio-economic backgrounds.

Data were collected through semi-structured in-depth individual interviews, conducted by the end of the first semester for first-year participants and at the beginning of the second semester for final-year participants. These interviews were conducted at a time and place designated by the participants, with interviews lasting 20 min – 35 min. The interview protocol comprised six open-ended questions. These questions were formulated based on three assessment belief domains, which focus on students’ knowledge of: (1) assessment (cognitive domain), (2) their emotions towards assessment (affective domain), and (3) the importance they attach to assessment (evaluative domain). These three domains offered a convenient window through which to examine and probe regulatory responses to assessment beliefs and their influence on students’ learning. The following questions were posed:

Follow-up questions preceded these questions, which required the participants to describe their assessment experiences or the situations that led them to understand assessment as they did. This was followed by a question asking the participants to elaborate on assessment experiences or situations that led them to feel the way they did about assessment. Additional probing and prompting questions were used to reiterate the questions and further guide the interview discussion. For example:

The data were analysed inductively, categorising them into themes using an interpretive phenomenological analysis (IPA) approach. Interpretive phenomenological analysis involves analysing participants’ responses to understand how they make sense of their personal and social worlds (Smith, Jarman & Osborn 1999). This approach was appropriate, as the intent was to uncover the characteristics of regulatory responses to assessment beliefs and, from the sources of these beliefs, to assess how learning was influenced by these regulatory responses. The process involved iterative reading to identify key statements, which were clustered into meaning units and grouped to form subcategories associated with the cognitive, affective and evaluative assessment belief categories, as shown in Table 1 and Table 2.

The process unfolded by identifying meaning units, which are coding frames that reflect the coded accounts related to the characteristics of participants’ assessment beliefs. Similar meaning units were grouped together to form regulatory responses that represent sub-categories linked to a specific assessment belief domain.

Thereafter, the process was repeated to identify meaning units that reflected the coded accounts of participants’ sources of regulatory responses to assessment beliefs. Similar meaning units were grouped to form condensed meaning units that captured the core meaning of the participants’ learning accounts emerging from their lived assessment experiences. From these condensed meaning units, code labels were assigned that described the meaning of the condensed meaning units. Conceptually similar codes were then clustered and organised into themes anchored in a central notion to explain the impact of students’ regulatory responses to assessment beliefs on learning.

Specific strategies were used throughout the research process to enhance the study’s trustworthiness. Credibility, transferability, dependability and confirmability are strategies that establish the trustworthiness of qualitative studies (Guba 1981; Krefting 1990). The researchers set aside their personal biases during data gathering and analysis by focusing solely on coded frames that reflected the participants’ words and experiences. The main author made reflective notes (Shenton 2004) to guard against personal bias during the interpretation of the findings, which were subsequently shared with the co-author for verification. A thick description (Ponterotto 2015) of the research setting and participants was provided in order to further enhance the trustworthiness of the study. Ethical issues were addressed by obtaining ethical clearance from the relevant ethics committee and permission from the university gatekeepers. Informed consent was obtained from the participants, who were free to withdraw from the study at any time.

Ethical clearance to conduct this study was obtained from the North-West University Education Sciences Research Ethics Committee (No. NWU/00297/22/A2) on 26 October 2023.

The assessment experiences of the participating first-year preservice science teachers revealed gaps in their learning and understanding. This recognition of personal learning limitations reflects the students’ self-awareness of their learning needs. The following responses provide evidence of this:

A common feature of these responses is that they show how the participants responded to regulatory responses to assessment, which is viewed as a means to assess understanding and support learning and engagement by identifying weaknesses in their learning. In this way, assessment functions as a mirror, helping students recognise their current level of understanding. This is further evident in the following responses from participants when asked which experiences had shaped their beliefs about assessment:

When considering Zimmerman’s (2002) SRL theory, the forethought SRL phase is associated with diagnosing learning needs and setting goals, aligning with this theme. This diagnostic regulatory response to assessment runs parallel to the characteristics of self-directed learning (SDL), which entails learners’ ability to diagnose their learning needs and then identify relevant resources to close the gap, with or without the help of others (Knowles 1975). This diagnostic response to assessment is also evident in the regulatory responses to the affective assessment belief domain, shown by the participants’ positive emotions towards assessment. This is evident from the following responses:

The theme ‘resourcefulness’ concerns identifying and accessing resources for learning. These resources were linked to reflective behaviours that relate to how students regulate their learning when confronted with assessment demands. This theme was evident in the following responses:

These responses show how the participants regulated their responses to their evaluative assessment belief domain, which values assessment as a means for deeper learning and engagement, by seeking information through research and engaging with peers. In this way, assessment prompted their resourcefulness to help close their learning gaps. For instance, the participants’ responses indicate that their resourcefulness was shaped by their self-awareness of a learning limitation that surfaced as a result of their assessment experiences, which led them to seek help from their peers and conduct research to bridge the shortfalls and enhance their understanding. The resourcefulness reflected by the participants aligns with the SDL processes of selecting appropriate learning resources (see Knowles 1975).

The theme ‘application of learning strategies’ highlights how the participants regulated their responses to the value placed on assessment, resulting in their use of specific strategies to improve their learning. For instance, one participant said:

Another participant supported the use of group work as a useful strategy that enables students to enhance their understanding through interactions with their peers. The participant said:

Within Zimmerman’s (2002) SRL theory, the performance phase of SRL involves task strategies such as time management, seeking help, structuring a learning environment and metacognitive monitoring, which are closely linked to the implementation of learning strategies in SDL. For example, the learning strategy of group work aligns with SDL processes that emphasise collaboration among students, providing opportunities for them to monitor and control the learning process (Garrison 1997).

In the following response, P5 demonstrated the use of structured approaches to address complex, higher-order questions during assessment:

The strategy used by this participant reflects a deliberate effort to manage assessment demands, moving beyond rote memorisation towards critical thinking and reasoning. The use of such a strategy is a hallmark of SDL, aligning with the self-management aspect, which entails students applying learning strategies to manage their learning process (Garrison 1997). This strategy is also evident in the following response from another participant:

When the students were asked to identify which assessment practices helped their learning, one participant mentioned watching YouTube videos and another mentioned using digital devices. Their responses were:

The application of these strategies reflects the students’ active role as agents in their learning and development. These responses show how they regulated their responses to their evaluative assessment belief domain, which values assessment as a means for deeper learning and engagement, and to their cognitive assessment belief domain, which holds that assessment is for assessing understanding further, indicative of their positive emotions towards assessment. Thus, the students’ application of various learning strategies, as influenced by their regulatory responses to their assessment beliefs, aligns with the foundational principles of SDL, which constitute an active learning process (Beckers, Dolmans & Van Merriënboer 2016).

This theme concerns students’ responses that demonstrate motivation to learn, influenced by their regulatory responses to the affective assessment belief domain, showing positive emotions towards assessment, driven by creative, engaging and stimulating assessment experiences. The following quotations provide evidence for this:

The responses above show that problem-based learning and practical, creative assessment can trigger intrinsic motivation to learn. However, according to the SRL theory, students’ motivation is intricately linked to the establishment of learning goals and to their efforts to monitor, regulate and control their cognition, motivation and behaviour (Pintrich 2000). Although the aspect of students setting goals was not evident from their responses, this finding supports the notion that intrinsic motivation can be enhanced by tasks with an appropriate level of novelty and difficulty, which are related to personal interests and allow for personal choice and control (Schunk 2012). In the context of SDL, intrinsic motivation is seen as a mediator between the process by which the student takes responsibility for constructing personal understanding and the application of learning strategies to meet desired educational outcomes (Garrison 1997).

However, from the students’ responses, it was also noted that the participants’ regulatory responses to the affective assessment belief domain, characterised by negative emotions towards assessment driven by testing and examinations, hampered their motivation to learn. This is evidenced in the following quotations:

From these responses, the following maladaptive behaviours could be identified: (1) fear of failure; (2) a lack of enjoyment in school; (3) anxiety; and (4) dissatisfaction with examinations. These maladaptive behaviours towards tests and examinations can hamper students’ meaningful engagement and motivation for deeper learning, such as SDL.

On the other hand, P6 expressed negative emotions towards testing and examinations, as shown in Table 2. The participant demonstrated extrinsic motivation to achieve high marks, as a regulatory response to negative affective assessment beliefs about tests and examinations. The participant said:

One explanation for the regulatory response expressed by P6 is that students’ effort, engagement and strategic regulation, as mediated by assessment beliefs, are inextricably linked and shaped by the learning environment that produced and sustained those beliefs. Therefore, it is reasonable to assume that P6’s extrinsic motivation to obtain grades is linked to how tests and examinations are used in the learning environment, namely for grading and measuring learning rather than for identifying learning needs and taking remedial action.

This assumption was supported by P10, who placed value on an assessment that was congruent with the view of assessment as a means to grade and demonstrate acquired knowledge. The participant said:

This participant’s regulatory response to the evaluative assessment belief domain aligns with the view of assessment as a means to grade and demonstrate acquired knowledge. This led to the participant’s satisfaction with the testing culture in higher education, resulting in effort being directed towards learning for the end-of-year examination. The participant said:

This value placed on assessment does not align with the reform-based assessment practices currently advocated, which align with assessment-for-learning and assessment-as-learning practices. As indicated by P6 and P10 responses, tests and examinations provide extrinsic motivation that encourages students to improve their marks or maintain their grades, as noted by Trotter (2006). This focus needs to be reconciled and changed, as it has the potential to diminish the role of assessment in influencing how students learn (process) by placing more emphasis on what they learn (product).

The theme of ‘passive learning’ depicts how the participants’ learning was influenced by their regulatory response to the cognitive assessment belief domain, which portrays assessment as testing, and by their response to the evaluative assessment belief domain, which portrays assessment as a means to demonstrate acquired knowledge. For instance, the following responses indicate passive learning attributes identified from the participants’ lived assessment experiences:

One of the adverse effects of these passive learning attributes was reported by P4, who expressed a fear of engaging during classroom learning and a dislike of passively receiving information. This is evident from the following quotation:

The study identifies that preservice teachers’ learning trajectories begin in the forethought phase, where their internal assessment beliefs serve as a primary lens for interpreting tasks. For participants holding reform-based beliefs, this phase is characterised by a diagnostic regulatory response. This aligns with Zimmerman’s (2002) proposition that the forethought phase involves diagnosing learning needs and setting goals; however, our findings extend this by showing that preservice teachers use assessment as a mirror to trigger self-awareness of learning limitations. Applying this mechanism to consider ways to enhance the full scope of SDL among students requires that we view the interplay between assessment belief domains, regulatory responses and learning as bidirectional. For instance, a student’s ability to become self-aware of their learning limitations is a learning response to the cognitive belief that assessment is a means to check understanding, positive emotions towards assessment and the value of assessment as a tool to support learning. This illustrates an interplay among psychological-behavioural aspects.

The study finds that the diagnostic task interpretation from the forethought phase directly mediates the performance phase, in which students enact specific task strategies to manage cognitive demands. For participants holding reform-based beliefs, this phase is characterised by an active and agentic regulatory response. As this is evident, as students moved from diagnosing needs to taking action, they demonstrated resourcefulness and the application of learning strategies. This aligns with Zimmerman’s (2002) performance processes, which run parallel to SDL processes of seeking help and implementing task strategies. However, our findings extend this by illustrating that strategy use, such as the ‘PEE’ structure or brainstorming, is a deliberate effort to manage assessment demands by moving beyond rote memorisation towards critical reasoning. Applying this mechanism to enhance the full scope of SDL involves viewing the interaction between evaluative belief domains and learning strategies as bidirectional. For instance, a student’s resourcefulness manifested through research and peer engagement is a learning response to the evaluative belief that assessment is a means for deeper engagement and a cognitive belief that it assesses understanding. This illustrates a clear interplay among psychological-behavioural aspects.

The self-regulatory cycle concludes with the self-reflection phase, where participants manage their affective reactions to performance and evaluate assessment outcomes. Our findings confirm the theory that participants experience satisfaction or dissatisfaction based on their performance. For participants influenced by a traditional schooling environment, this phase is often characterised by a maladaptive or extrinsic regulatory response. For instance, participants’ cognitive beliefs about assessment as testing and grading, and their negative emotions towards assessment and the value of assessment as a tool to demonstrate acquired knowledge, are, in part, shaped by the monotony of summative assessment experiences expressed by students, illustrating an interplay between environmental and psychological aspects. This aligns with Zimmerman’s (2002) proposition regarding students managing their emotions, such as satisfaction or dissatisfaction in response to performance. However, our findings extend this by documenting how students internalise extrinsic motivation to prioritise grades over skills when faced with high-stakes testing. Thus, enhancing SDL requires acknowledging that the link between these internal affective states and the broader learning environment is bidirectional. For instance, a participant’s extrinsic motivation and subsequent passive learning are a psychological-behavioural response to the environmental constraint of a dominant summative testing culture. However, the data challenge the assumption that summative assessment inherently produces maladaptive self-reflection. By documenting participants’ positive engagement with continuous summative assessment, the study reveals that graded tasks, when feedback-rich, can foster intrinsic motivation rather than the toxic anxiety typically predicted by the literature on summative cultures (see Atjonen 2014; Cimer & Cimer 2010; Klinger & Luce-Kapler 2008). Ultimately, this demonstrates that the self-reflection phase is a critical site for transformation, where shifting the environmental drivers from static grading to interactive, feedback-rich engagement can realign the environmental-psychological interplay to support more adaptive SDL processes.

Applying the mechanism illustrated in Figure 2, the full scope of students’ principles of SDL can be enhanced by strengthening either the psychological-behavioural aspects or the environmental-psychological aspects. For instance, environmental-psychological aspects can be strengthened by fostering students’ beliefs about assessment that support the pedagogical role of assessment, especially by emphasising and implementing reformed assessment approaches, such as assessment as learning and assessment for learning. These approaches have been shown to promote SDL by actively engaging students in monitoring, regulation and reflection on their learning (Harrison, O’Hara & McNamara 2015b; Nasri, Nasri & Abd Talib 2021; Sosibo 2019). Consequently, when students adopt beliefs about assessment that support the pedagogical role of assessments, their SDL processes are enhanced through corresponding regulatory responses to these beliefs, thereby strengthening the psychological-behavioural aspects. By applying these bidirectional mechanisms, future studies can continue to explore how the development of reformed assessment approaches can be fostered through SDL, as a personal student attribute, denoting behavioural-psychological interactions, and through SDL viewed as an environmental aspect, denoting environmental-psychological interactions. This is a much-needed inquiry in light of the findings that showed the influence of students’ regulatory responses to assessment beliefs on their SDL processes. Consequently, more attention should be paid to developing a framework to help us understand how SDL can, in turn, further support the development of beliefs about assessment that align with reformed assessment approaches.