Computational thinking (CT) is a problem-solving approach that uses computer science concepts such as decomposition, pattern recognition, abstraction, and algorithm design, and is widely acknowledged as a fundamental 21st-century skill (Grover, 2022; Newton et al., 2022; Wing, 2006) in an increasingly digitalized society (Barr and Stephenson, 2011). Technological advancements have transformed the structure of daily life, influencing domains far beyond computer science (Barr and Stephenson, 2011; Deguchi et al., 2020). The importance of CT competencies is reflected in both national and international policy documents, notably within the European Digital Competence Framework for citizens, where CT is identified as a core objective for primary and secondary education (Bocconi et al., 2022; Grover and Pea, 2013; Herrmann et al., 2020).

In particular, the integration of CT into K-12 educational contexts as a practical tool to achieve these objectives has received growing attention (Grover and Pea, 2013; Jocius et al., 2021; Sırakaya et al., 2020). Further research is needed to clarify how CT can foster mathematical thinking and learning in K-12 education, and vice versa (Hickmott et al., 2018; Ye et al., 2023). At the same time, teachers must be adequately prepared to cultivate CT competencies among their students, given the increasing emphasis on developing computationally competent learners and formally incorporating CT into curricula (Carpenter et al., 1989; Miller, 2019; Pei et al., 2018; Shumway et al., 2021). To meet these curricular demands, teachers require a solid foundation in CT content knowledge (Angeli et al., 2016; Caskurlu et al., 2022; Computer Science Teachers Association The International Society for Technology in Education, 2011). Delivering high-quality instruction necessitates a deep understanding of the resources and concepts to be taught, as this directly influences student learning outcomes, particularly in mathematics education, which is widely adopted in K-12 contexts (Baumert and Kunter, 2013a; Cui and Ng, 2021; Putnam et al., 1992).

To effectively integrate and develop computational thinking (CT) within K-12 mathematics education, selecting an appropriate computing device is essential (Barcelos et al., 2018; Hickmott et al., 2018; Lee et al., 2023). Physical computing devices offer opportunities to enhance mathematical learning by making it more engaging, accessible, and conducive to creativity and collaboration (Kalelioglu and Sentance, 2019; Pei et al., 2018; Shumway et al., 2021). These devices support a variety of CT-related activities, including programming tasks and instructional design, which can be embedded into mathematics learning (De Chenne and Lockwood, 2022; Kotsopoulos et al., 2017). Existing studies highlight pedagogical frameworks such as unplugged activities, tinkering, problem formulation, remixing designs, problem-solving, number theory, and mathematical modeling as effective approaches for integrating CT into K-12 mathematics education (Bråting and Kilhamn, 2021; Miller, 2019; Ng and Cui, 2021). Importantly, applying CT in mathematics can deepen conceptual understanding due to the reciprocal relationship between computer science and mathematical models, while disciplinary mathematical knowledge simultaneously strengthens CT skills (Aytekin and Topçu, 2024; Pei et al., 2018; Przybylla and Romeike, 2014).

Among physical computing devices, micro:bit stands out for its robust features and suitability for educational contexts (Ball et al., 2016). According to experienced teachers, micro:bit is user-friendly, affordable, and requires minimal preparation, addressing limitations found in earlier devices (Sentance et al., 2017; Kalelioglu and Sentance, 2019). Its interactive feedback mechanisms make learning more engaging, motivating, and stimulating for students (Lu et al., 2022; Sentance et al., 2017). Furthermore, as students become familiar with its components, they can leverage micro:bit to create increasingly complex and innovative projects (Fagin et al., 2001; Kalelioglu and Sentance, 2019).

CT abilities in K-12 mathematics education, such as problem-solving, remain relatively low due to several contributing factors (Nuraisa et al., 2021; Supiarmo et al., 2022; Wardani et al., 2021). In this context, teachers’ CT content knowledge within higher education institutions (HEIs) is still insufficiently developed, despite growing research interest in this area (Kong et al., 2023; Tang et al., 2020; Yadav et al., 2016). Baumert and Kunter (2013b) emphasize that content knowledge constitutes a core teaching competence, requiring teachers to possess a deep understanding of the subject matter to be taught (Mishra and Koehler, 2006; Shulman, 1986), as well as related pedagogical and domain-specific skills (Baumert and Kunter, 2013b; Depaepe et al., 2013; Star, 2005). Moreover, it is essential to examine the relationship between attitudes and emotions and CT content knowledge in K-12 mathematics education (Bieleke et al., 2023; González-Gómez et al., 2022; Lu et al., 2022). Previous research indicates that attitudes influence CT content knowledge and learning outcomes for both students and pre-service teachers (Cutumisu et al., 2021; Jeong and González-Gómez, 2025a; Sun et al., 2021). Learners’ attitudes and emotions toward activities involving micro:bit technology are closely linked to mathematics education (Humble, 2022; Voštinár and Knežník, 2020; Wadaani, 2023). Emotions, in particular, have gained prominence due to their significant impact on learning outcomes in CT (Bieleke et al., 2023; Herrero-Álvarez et al., 2023; Jeong et al., 2019). Mathematics education is a critical subject globally and is frequently associated with a wide range of emotions, both positive and negative, making it an important area of investigation (Frenzel et al., 2007; Pekrun et al., 2007; Schukajlow et al., 2017).

Thus, this study aims to integrate and develop CT through a micro:bit-based activity (CT-micro:bit) by examining pre-service teachers’ self-perceived CT content knowledge (measured via decomposition, pattern recognition, abstraction, and algorithm design), attitudes (measured via creativity, problem solving, algorithmic thinking, cooperation, and critical thinking), and emotions (measured via emotional stability, self-motivation, and emotional relations) in preparation for future K–12 mathematics teaching. The research is guided by the following questions:

The research results could be considered as a further suggestion, how pre-service teachers could improve in further K-12 mathematics learning to enable the development of self-perceived content knowledge, attitude, and emotion of CT through the activity of physical computing device like micro:bit.

This study investigates the self-perceived content knowledge, attitudes, and emotions associated with CT among pre-service teachers in HEIs through micro:bit-based activities. The research aims to potentially enhance preparation for K-12 mathematics education. A pre-experimental design was adopted, employing quantitative descriptive data to compare pre- and post-test results across various academic courses (Maxwell et al., 2017). Additionally, a correlational design was used to identify variables that may predict outcomes (Creswell and Creswell, 2018). The conceptual framework is illustrated in Figure 1.

To assess whether the implementation of an instruction methodology based on the use of micro:bit to integrate and develop the computational thinking on mathematics educations of pre-service teachers had a significant influence on pre-service teachers on the dependent variables, the mean values of the G-CTK, G-CTA, and G-CTE were compared before and after the intervention. For G-CTK, the mean score before the intervention was 2.50 (std dev = 0.66), which increased to 4.71 (std dev = 0.28) after the intervention. The Student’s t-test for paired samples indicated that this difference was statistically significant (t(226) = 32.4, p < 0.001, CI₉₅ = [2.06, 2.33], d = 4.39). This result indicates a statistically significant increase in self-reported confidence regarding computational thinking content knowledge from pre- to post-test instructional intervention. Similar results were observed for G-CTA, where the mean score increased from 2.79 (std dev = 0.60) before the intervention to 4.70 (std dev = 0.28) after it. The t-test confirmed that this difference was also statistically significant (t(226) = 30.8, p < 0.001, CI₉₅ = [1.78, 2.02], d = 4.08). Finally, for G-CTE, the mean score rose from 2.70 (std dev = 0.46) prior to instruction to 4.74 (std dev = 0.25) post-instruction (t(226) = 40.19, p < 0.001, CI₉₅ = [1.86, 2.04], d = 5.32). This difference was also statistically significant. Figure 4 represents the evolution of these variables before and after the intervention. In all cases, the effect size, measured as Cohen’s d, denotes that large pre-post differences were observed across all the studied variables, although these should be interpreted cautiously given the pre-experimental design. The large effect size values may indicate the presence of a ceiling effect, potentially introducing a limitation to the overall measurements. However, it is important to note that the effect sizes calculated for each individual variable fall within the conventional thresholds for Cohen’s d.

A one-way ANCOVA was conducted to assess the influence of the instructional methodology on the students’ G-CTK, G-CTA, and G-CTE whilst controlling students’ grade, gender, age, and educational background. ANCOVA results indicated that there was a significant difference in mean values of G-CTK [F(1,221) = 1054.4, p < 0.001], G-CTA [F(1,221) = 948, p < 0.001], and G-CTE [F(1,221) = 1609.8, p < 0.001] before and after the intervention, respectively. However, the results of the ANCOVA test also indicated that students’ grades, gender, age, and educational background did not influence the dependent variables assessed in this study. On the other hand, to have a better understanding of how each individual score sifted within the intervention, Figure 5 represents the mean values of each individual item before and after the intervention. As shown in Figure 5, all items in the “pre-perceived computational thinking content knowledge” domain exhibited significant pre–post improvement (p < 0.001). The largest mean differences were observed for abstraction (Item 1; ΔM = 2.25) and pattern recognition (Item 4; ΔM = 2.65). Comparable effects were found in the “computational thinking attitude” domain (p < 0.001), with problem solving (Item 2; ΔM = 2.18) and algorithmic thinking (Item 3; ΔM = 2.11) showing the greatest gains. Regarding the “computational thinking emotion,” in the case of the positive one, a significant positive influence was observed (p < 0.001). On the other hand, for the negative emotions the mean values of observed in “Nervousness” (CTEN_1) and “Disappointment” (CTEN_4) significantly decreased after the intervention (p < 0.001), while “Frustration” (CTEN_2) and “Boredom” (CTEN_3) did not experience any change during the intervention (p = 0.209 and p = 0.302, respectively).

Considering the activating/deactivating classification of emotions (Pekrun et al., 2007), results indicate that positive activating emotions (CTEP_1 and CTEP_2) significantly increased after the intervention, and that could be related to an enhancement of learning and performance (Pekrun, 2010). On the other hand, the positive deactivating emotions (CTEP_3 and CTEP_4) increased after the intervention. Although limited research is available on their specific effects, fostering these emotions could potentially reduce cognitive stress and might offer a temporary respite from intense cognitive engagement (Dreisbach and Goschke, 2004). Regarding the negative activating emotions (CTEN_1 and CTEN_2), a significant decrease was observed in CTEN_1 (nervousness) while no difference was observed in CTEN_2 (frustration). The negative activating emotions, in general terms, affect students’ motivation and reduce self-regulating learning (Heckhausen, 1991). In the case of the negative deactivating emotions (CTEN_3 and CTEN_4), no significant difference was observed for CTEN_3 (boredom) while a significant decreased was observed for CTEN_4 (disappointment). Negative deactivating emotions negatively influence students’ learning process and outcomes, therefore the instruction methodology implemented might mitigate these effects. The large Cohen’s d values for the overall scores derived from the EFA results may indicate a potential ceiling effect, which could constrain the interpretation of these scores when assessing the intervention’s influence. At the item level, pre–post Cohen’s d values ranged from d = 0.67 to d = 0.88, all remaining below 0.90. This item-level analysis strengthens the validity of the study by complementing the interpretation of the global results, while factors such as the use of self-report measures and the absence of a control group may be also acknowledged.

Finally, Table 4 represents the plot of the Spearman correlations and network plot for the validated subscale scores (G-CTK, G-CTA, and G-CTE) and the two CTE factors (frustration and boredom) identified through exploratory factor analysis before and after the intervention. In this plot, the color of the lines denotes the positive (green color) or negative (red color) relationship between variables, and the line thickness the strength of this relation.

According to these results, prior to the intervention, G-CTK, G-CTA, and G-CTE exhibited significant positive correlations with each other (r_s > 0.8 and p < 0.001 in all cases), while no correlation was observed with the two CTE factors (frustration and boredom). After the intervention, significant correlations among G-CTK, G-CTA, and G-CTE persisted (r_s ranging between 0.4 to 0.6 and p < 0.001, as indicated in Table 4). Notably, a significant negative correlation also emerged between the CTE constructs (frustration and boredom) and the three subscales (G-CTK, G-CTA, and G-CTE) (r_s < −0.3 and p < 0.001 in all cases), with the strongest negative association observed between CTE and G-CTK. This finding may provide insight into the effectiveness of the intervention in enhancing pre-perceived CT content knowledge.

Based on the results analyzed through various approaches, the study addressed the first research question raised for this study (Q1). Tang et al. (2020), along with other researchers (Kong et al., 2023; Yadav et al., 2016), indicated that the teachers’ content knowledge in this field of HEIs is not well-established for K-12 mathematics education. However, the findings revealed significant pre-post differences following the implementation of the CT-micro:bit activity sequence. In detail, as it was mentioned before, the mean scores of G-CTK increased 2.25 to be 4.71 (std dev = 0.28) after the intervention and this difference was statistically significant, which suggested a significant improvement. Therefore, these findings align with the previous research emphasizing the importance of developing teachers’ confidence and perceived readiness related to computational thinking (Baumert and Kunter, 2013b; Depaepe et al., 2013; Mishra and Koehler, 2006). Then, Lu et al. (2022) and Bieleke et al. (2023) specified that it is important to know the relationship of attitude and emotion to computational thinking content knowledge in K-12 mathematics education. To answer the second question raised for this study (Q2), Cutumisu et al. (2021) denoted that attitude had a strong relation with the content knowledge of computational thinking, which affected students’ learning achievements in K-12 mathematics education. In detail, the mean of G-CTA increased 1.91 to be 4.70 (std dev = 0.28) after the intervention, and this difference was statistically significant, which suggested a significant improvement. In all cases, the results of the ANCOVA test indicated that students’ grades, gender, age, and educational background did not influence the dependent variables assessed in this study. Therefore, this corresponds to the indications of previous research that, if learners showed positive attitude to a sequence of CT-micro:bit activity, they would more interconnect to K-12 mathematics education (Humble, 2022; Jeong and González-Gómez, 2025b; Sun et al., 2021; Wadaani, 2023). Moreover, to answer the third question raised for this study (Q3), Bieleke et al. (2023) and Herrero-Álvarez et al. (2023) expressed that emotion had a strong relation with the content knowledge of computational thinking, which affected students’ learning achievements in K-12 mathematics education. In detail, the mean of G-CTE increased 2.04 to be 4.74 (std dev = 0.25) after the intervention, and this difference was statistically significant, which suggested a substantial difference. However, “Frustration” (CTEN_2) and “Boredom” (CTEN_3) as negative emotion did not change during the intervention (p = 0.209 and p = 0.302, respectively). Considering the activating/deactivating classification of emotions (Pekrun et al., 2007), the results showed the same tendency although “Frustration” (CTEN_2) is an activating negative emotion and “Boredom” (CTEN_3) is a deactivating negative emotion regardless of participants’ personal characteristics. Particularly, in a different and challenging learning situation, these emotions are not considered negative since, without them, it would not have the fun component in the learning process (Clauson et al., 2019; Jeong, 2018; Jeong et al., 2016; Sierra and Fernández-Sánchez, 2019). In this context, therefore, this corresponds to the indications of previous research that, if learners showed emotion to a sequence of CT-micro:bit activity, they would more interconnect to K-12 mathematics education (Frenzel et al., 2007; Schukajlow et al., 2017). Thus, the results of this study indicate that the implementation of the micro:bit activity revealed significant pre-post differences in CT components. Specifically, the content knowledge components showed strong gains in concept generalization and recognition of structural regularities. The attitude components indicated a consistent increase in students’ confidence in tackling structured tasks and designing stepwise solutions. Finally, the emotion components suggest that the instructional approach may have mitigated detrimental affect related with computational thinking.

The proposed study aimed to potentially enhance pre-service teachers’ pre-perceived CT content knowledge, attitudes, and emotions for future K–12 mathematics instruction through the implementation of a CT–micro:bit activity. This intervention was integrated into the mandatory introductory course “Mathematics and its Didactics.” The CT–micro:bit activity followed a six-step sequence aligned with divisibility concepts, introducing participants to visual block-based programming and Python, and culminating in the construction of a physical calculator designed to identify prime factors and divisors.

Regarding the Research Question 1 (Q1), the intervention was associated with statistically significant pre–post improvements in self-perceived CT content knowledge (e.g., gains consistent with abstraction/concept generalization and pattern recognition/structural regularities). These findings indicate that mathematically grounded physical-computing tasks might strengthen pre-service teachers’ confidence in core CT content knowledge relevant to K–12 mathematics teaching. On the other hand, regarding Research Question 2 (Q2), CT attitudes improved significantly, reflecting increased confidence in tackling structured tasks and designing stepwise solutions (e.g., problem solving and algorithmic thinking). These attitudinal gains suggest that the activity fostered productive dispositions toward integrating CT in mathematics pedagogy. Finally, regarding Research Question 3 (Q3), positive emotions increased significantly. Among negative emotions, nervousness (activating) and disappointment (deactivating) decreased significantly, whereas frustration (activating) and boredom (deactivating) showed no detectable change. Interpreted within the activating/deactivating framework, these patterns suggest the instructional approach may have mitigated detrimental affect while enhancing emotions typically associated with learning and performance.

These results suggest that engaging pre-service teachers in physical computing activities may foster more positive self-perceptions of computational thinking content knowledge, attitudes, and emotional engagement, with potential implications for future K–12 mathematics education.

Nevertheless, despite its advantages, incorporating micro:bit activities into educational settings presents challenges, including alignment with curricular objectives and potential disruptions to established lesson schedules. This study also entails limitations that must be acknowledged. First, the pre-experimental design lacks a control group, and the measured effects are based on a single-site context. Second, the instruments relied on self-reported measures, which may introduce bias. Finally, as discussed in the manuscript, the possible existence of a ceiling effect on global scores should be considered. To fully leverage the benefits while mitigating these challenges, a deliberate and systematic approach to integrating CT–micro:bit activities into curricula is essential, incorporating appropriate comparison groups and conducting longitudinal follow-up studies, thereby informing promising educational strategies.