Unlocking students’ creative potential in designing technological-enriched design solutions

Abstract

Creativity is an important competency for talents in the twenty first century. The study examines how integrating real-world community service challenges into the secondary school STEM curriculum impacts students’ creativity and STEM content knowledge. A total of 123 secondary school students learned the challenges faced by people with visual impairments and applied their STEM knowledge to design creative solutions. Students proposed diversified solutions, which were of three themes: safety, recreation, and convenience. The quality of the solutions indicated that the higher-performing groups outperformed the medium-performing and lower-performing groups in originality. The lower-performing groups needed further support on all dimensions, especially technical knowledge and skills, which measures proficiency in using integrated knowledge to realize designs.

Keywords:

• Interdisciplinary STEM
• community service learning
• creative problem-solving
• creativity
• technology
• design thinking

Introduction

Creativity is an important competency for talents in the twenty first century (Hennessey & Amabile, 2010; Organization for Economic Co-operation and Development [OECD], 2022). Education scholars believe combining science, technology, engineering, and mathematics (STEM) education with exposure to real-world challenges is beneficial for developing students’ creativity (Herro et al., 2019; Soomro et al., 2023). This approach increases the likelihood that students will connect their learning with real-world scenarios and see how their knowledge can be applied for the betterment of the world (Maiorca et al., 2021; Roberts et al., 2018). Furthermore, engagement in meaningful and intellectually challenging tasks can stimulate students’ interest in a subject (Hervani & Helms, 2004; Kampylis & Berki, 2014; Song, 2020) and thus stimulate their creative potential (Rahimi & Shute, 2021; Xiong et al., 2022; Zheng et al., 2021). Nonetheless, there is a dearth of research on how theoretical frameworks can guide students in developing their creative potential for solving real-world problems and measuring the impact of such approaches through students’ creative product (Aguilera & Ortiz-Revilla, 2021; Soomro et al., 2023). Some scholars have suggested that a supportive environment, provision of feedback, and a creative process are essential for cultivating students’ creativity (Amabile & Pratt, 2016; Sternberg, 2003; Zheng et al., 2021). Given the importance of creativity, steps need to be taken to provide students more room for open-ended explorations (Foster & Schleicher, 2022; Sternberg, 2003) and to examine whether they can devise creative and appropriate solutions for specific purposes. In the K-12 setting, researchers engaged students in pre-assigned tasks on developing toys (Zhou et al., 2017), improving the design of a watch through mind mapping (Sun et al., 2022), or designing personalized masks (Weng et al., 2022). Exploration needs to be elevated to a new level where students are able to utilize advanced technologies (e.g. application development, Internet of Things) and design solutions that are applicable to real life. There is also a need for a deeper understanding of student learning outcomes in integrated STEM programs and how to support students at different performance levels. We must be attuned to the problems that students encounter and provide a suitable level of support to them. Accordingly, the present study explores the effects of engaging students in an integrated STEM program, where students design technological-enriched solutions for community citizens, and its impacts on students’ creativity and STEM content knowledge. In the integrated STEM program, students would be introduced to a community service issue (i.e. the living conditions of visually impaired groups and the problems they face in their lives), learn a certain amount of STEM knowledge as a foundation (i.e. technological and engineering knowledge), and then work in groups to conduct open-ended explorations. The groups would design solutions (i.e. in the format of design proposals) to improve the quality of life of visually impaired groups. To examine the effectiveness of the integrated program, we intend to answer the following questions:

1. What is the impact of the integrated STEM program on students’ diversity of technological-enriched design solutions?

2. What is the impact of the integrated STEM program on students’ quality of technological-enriched design solutions?

3. What are the differences in the quality of technological-enriched design solutions among the higher-, medium-, and lower-performing groups?

4. What is the impact of the integrated STEM program on students’ STEM content knowledge?

Literature review

Theoretical foundations for fostering creativity

Creativity has been recognized as an essential factor in the progress of human civilization (Hennessey & Amabile, 2010). It is a key attribute within organizations necessary to survive in an increasingly competitive global market (Wigert et al., 2022). People use creativity to solve problems and address the myriad challenges faced by schools, cities, nations, and the global community (Hennessey & Amabile, 2010; OECD, 2022). Creativity is therefore considered a critical skill for twenty first century citizens (Cheng et al., 2022; Kampylis & Berki, 2014). Plucker et al. (2004) defined creativity as “the interaction among aptitude, process, and environment by which an individual or group produces a perceptible product that is both novel and useful as defined within a social context” (p. 90). The creative product can either be a response, an artifact, or a solution to an open-ended task (Amabile, 2012; Amabile & Pratt, 2016). The two essential features of a creative product are novelty and usefulness (Runco & Jaeger, 2012).

Drawing on the definition of creativity as the ability to generate novel and useful ideas or solutions to open-ended tasks and supported by extensive studies of the stimulating and inhibiting factors of innovation, Amabile (1982, 2012) proposed the componential model of creativity. The componential model posits that motivation to innovate, knowledge and skills in the task domain, and creative processes are essential for incubating creativity. Meanwhile, a supportive social environment is considered vital in stimulating the motivation to innovate (Amabile & Pratt, 2016). Inspired by the componential model of creativity, Huang et al. (2022) proposed the integrated STEM-based creative problem-solving model (also known as the integrated STEM-based community service-learning model). The model proposes that real-life problems or challenges in the community can be used as a starting point to stimulate innovation in learners. Then, students can be taught the basic knowledge and skills in the domain (Csikszentmihalyi, 1996) to lay the foundation for innovation. Next, students can be guided through a creative process to explore solutions. In order to optimize students’ creative potential, schools should establish a supportive social environment, provide adequate resources and support, and encourage communication and collaboration among peers. Figure 1. As for the creative process, the model refers to the EDIPT (empathize-define-ideate-prototype-test) design thinking framework (Hasso Plattner Institute of Design, 2010) and suggests organizing creative exploration activities through the processes of empathizing, defining, ideating, designing, and collecting feedback for improvement, emphasizing more the role of feedback (Torrance, 1953) in developing quality solutions.

Figure 1. The integrated STEM-based creative problem-solving model (adapted from Huang et al., 2022).

Measurement of creativity and its backgrounds

In Guilford’s (1967, 1971) foundational theories of creativity, the cognitive process of creative thinking is divided, albeit not exhaustively, into two categories: divergent thinking and convergent thinking. Divergent thinking is oriented toward generating a variety of responses to a stimulus or solutions to a problem (Guilford, 1967; Reiter-Palmon et al., 2019). Convergent thinking, by contrast, is oriented toward arriving at a single best or correct answer to a defined problem (Cropley, 2006; Guilford, 1967) and involves the selection, evaluation, and transformation of the possibilities that lead to the emergence of new forms and patterns (Guilford, 1967; Sun et al., 2020; Ward et al., 1997). To assess divergent thinking, Guilford (1967, 1968) devised the Alternate Uses Task (AUT), which asks participants to list as many novel uses as possible for daily objects (e.g. Beaty et al., 2022; Dumas et al., 2021). To assess convergent thinking, Mednick (1962) developed the Remote Associate Test, which provides participants with a set of words and asks them to form compound associations with a keyword. This test has also been widely adopted (e.g. Chermahini et al., 2012; Lazonder et al., 2022). However, these tests are often conducted under manipulated conditions (or in artificial environments), and the test scores might not reflect real-world performance; the scores are therefore taken primarily as estimates of creative potential (Acar et al., 2020).

As reviewed in the next section, many scholars have suggested assessing creativity by having people engage in creative problem-solving in conditions that are more strongly connected to real-life contexts.

Measurement of creativity in STEM contexts

To strengthen the connection to real-life contexts and the evaluation of creativity, Nazzal and Kaufman (2020) recommended examining the quality of the solutions (i.e. an indicator of convergent thinking) arrived at when solving open-ended, ill-defined problems. In their study, Nazzal and Kaufman (2020) introduced engineering students to a real-life scenario. They asked them to identify an engineering-related problem, propose a solution to the problem, and explain their plans for implementing this solution. They evaluated responses to four design phases (problem identification, idea generation, idea evaluation, and solution validation), and the overall creativity demonstrated in the responses. The OECD (2019) suggested evaluating students’ creative thinking competencies from three dimensions: the ability to generate diverse ideas, the ability to generate creative ideas that are original and appropriate, and the ability to evaluate and improve ideas. For example, assign a real-life based task (e.g. improve the design of a bicycle) and examine students’ capabilities in providing diverse and creative design solutions and their capacity to improve existing design solutions. The OECD (2019) introduced a three-level system (full credit, partial credit, no credit) for scoring creative thinking according to the appropriateness and originality of students’ responses. In the present study, we focused on the diversity and quality of students’ technological-enriched design solutions in response to a community-service challenge for improving the living conditions of visually impaired groups. We aimed to understand:

1. whether students are able to generate diversified solutions to help the community citizens (i.e., the ideas proposed by students are similar or differ from each other);

2. whether the ideas proposed by students are of quality (i.e., the ideas are helpful and demonstrate their capacity in designing appropriate and user-centered solutions);

3. whether there are any differences among the higher-, medium-, and lower-level performing groups’ solutions, and what supports we can provide to the learner groups;

4. whether students’ STEM content knowledge would be improved after completing the interdisciplinary STEM program.

Methods

Design of the interdisciplinary STEM program

The design of the interdisciplinary STEM program was based on the integrated STEM-based creative problem-solving model. As presented in Figure 2, the integrated learning experience was organized according to the following steps:

Figure 2. The flow of the integrated STEM program.

• First, the students were briefed on implementing the interdisciplinary STEM program and introduced to the community service issues, i.e., enhancing the living conditions of visually impaired groups, they would be working to address.

• Second, the students were taught the foundational STEM content knowledge, especially on circuits, sensors, and coding, that they would need to devise feasible technical solutions.

• Third, the students were provided with the community service knowledge necessary to understand the major challenges visually impaired people face in real life. Students were informed to conduct open-ended explorations to identify a specific need of visually impaired groups and work out a design solution to help them.

• Fourth, the students followed the creative design process to brainstorm solutions with their group members and produce a detailed proposal explaining their design rationales and the technical knowledge involved. A rubric was also introduced to guide their work (Appendix A).

• Fifth, the students collected feedback by presenting their proposals to their peers and experts in STEM and community service, e.g., staff of the social service organization Blind Union.

• Then, they refined their solutions over two weeks and submitted a final version on the Google Classroom learning platform.

The program ran for eight weeks. The first six weeks were conducted live online using the Zoom videoconferencing software. In the final two weeks, the students attended on-site sessions. See Figure 3.

Figure 3. Timeline of the program.

Participants

A total of 123 Grade 8 secondary school students consented to participate in this study in the spring semester of 2021. The students had some foundational knowledge of coding through Scratch prior to participating in the integrated program. The students were randomly divided into 32 groups to work on their creative problem-solving. They were taught by two senior STEM teachers with at least five years of experience in teaching. See Figures 4 and 5 for screenshots of the live online Zoom course. Each teacher used the same teaching slides and student workbooks to ensure consistency. The teachers received the same training from STEM and community service experts before the beginning of the program.

Figure 4. Students learning app development skills via Zoom videoconferencing.

Figure 5. Students learning social-technical products via Zoom videoconferencing.

Data collection and analysis

The study collected and analyzed the diversity and quality of the students’ design solutions developed through creative problem-solving. The design solutions were in the format of STEM-based design proposals, in which students needed to introduce the problem they intended to solve, the product they proposed, and the technologies and principles they would use to actualize the product. Meanwhile, students need to justify the importance and uniqueness of their design solutions. To evaluate diversity, the design solutions were analyzed according to the themes and sub-themes of products’ functionalities using thematic analysis (Braun & Clarke, 2006). First, a researcher read through all the proposals to gain an understanding of the main functions covered by the proposed products. Then, the researcher generated the initial codes and coding scheme, and revisited the design proposals one by one. In the coding process, the researcher constantly compared and refined the codes and coding scheme, and determined the core categories (i.e. overarching themes). To ascertain the reliability of the coded results, an independent reviewer randomly selected 30% of the proposals and coded them using the finalized coding scheme. The inter-rater agreement (Campbell et al., 2013) between the reviewer and the researcher was 82%.

To evaluate quality, the solutions were evaluated by two experienced STEM teachers following a pre-defined evaluation rubric. The rubric evaluated the creative design solutions from four dimensions, including (1) defining the design goal, (2) technical knowledge and skills, (3) reasoning, and (4) originality. The rubric (see Appendix A) was uploaded to the Google Classroom learning platform so that the students would be clear on the performance goals to achieve (Reynders et al., 2020). To enhance the internal validity of the scores, the teachers were also briefed on the rubric, including its rationale, some examples for different score ranges, and the criteria for assessing students’ solutions. After all the solution proposals were graded, an independent reviewer randomly selected 30% of the submissions for second grading. The inter-rater agreement (Campbell et al., 2013) between the reviewer and the first teacher was 84% and that between the reviewer and the second teacher was 88%.

To track changes in students’ factual knowledge of STEM, a pretest was administered before the commencement of the program, and a post-test was administered after its completion. The content of the test was developed by an experienced STEM teacher and reviewed by two STEM teachers to ensure content validity and readability (Moss, 1992). In the STEM content knowledge test, students were asked to answer factual knowledge questions based on what they had learned in class. For example, one question was, “Which of the following is NOT a commonly used open data format? A. CSV; B. INO; C. JSON; D. XML.” The same set of questions was used at the beginning and end of the program.

Results

Diversity of the design solutions

To understand the impact of the integrated STEM program on students’ diversity of technological-enriched design solutions, we conducted a descriptive analysis of students’ design solutions. Three themes were identified in the 32 design proposals according to the functionality of the proposed products: safety, convenience, and recreation. The safety-related proposals (n = 18) covered the four sub-themes of safety at home (n = 6), safety outdoors (n = 7), medical and health safety (n = 4), and food safety (n = 1). The convenience-related proposals (n = 12) covered a broad range of sub-themes, such as convenience for seeing things (e.g. smart glasses), convenience when shopping (e.g. a product and price recognizer), and transport convenience (e.g. a bus reminder). The recreation-related proposals (n = 2) addressed the sub-themes of the accessibility of reading (n = 1) and sport (n = 1). For instance, one group proposed a smart carpet that could improve safety at home by detecting falls and responding by emitting an alert sound and sending a message. Another group proposed an online shopping application that could improve convenience by speaking the names and prices of products and identifying which products were discounted. A third example was a proposed design for a multifunctional wheelchair that could improve safety outdoors by identifying obstacles and helping users avoid dangers on the road.

The themes and sub-themes of the students’ proposed designs are presented in Table 1. along with examples of the products they came up with through the creative problem-solving process. The solutions were spread across different categories of functionality, suggesting that the students could engage in divergent thinking and propose solutions from different angles.

Table 1. Themes and sub-themes of the technological-enriched design solutions.

Themes	Sub-themes	Examples of design solutions
Safety	Home safety	Visually Impaired Annunciator, Smart Door, Safety Smart Carpet, Cooking Flame Assistant
	Outdoor safety	Smart Crutch, Electric Guide Dog, Walk Out Safety App, Multifunction Wheelchair
	Food safety	Visual Impaired Drink
	Medicine safety & health	Health Checker App, Medicine Giver, Dan’s Virus Detector and Virus Cleaner for iPhone
Convenience & assistance	Convenience for seeing general information	Visually Impaired Glasses, Smart Glasses, Sense and Read Inductor
	Convenience for shopping	Price and Product Recognizer, Online Shopping Device
	Convenience for transportation	Easy Walk, AI Voice Guideline Earphone, Bus Reminder,
	Convenience for other purposes	Smart Hearing Aid, Social Distance Detector, Visually Impaired Auxiliary Watch
Recreation	Recreation for reading	Smart Pen
	Recreation for sports	Elderly Exercises APP

Quality of the design solutions

To understand the impact of the integrated STEM program on students’ quality of technological-enriched design solutions, we analyzed the overall rating and the sub-ratings (i.e. defining design goals, reasoning, technical knowledge and skills, and originality) of students’ design solutions. According to the rating scheme of the rubric, the design solutions were scored out of 30, with one of the higher-performing groups scoring as high as 28. See Figure 6. For this higher-performing group, the definition of the design goal and the presentation of the technological-enriched solution demonstrated completeness. One group scored merely 15 points because it did not adequately explain how the solution works and how it could be developed.

Figure 6. Distribution of the quality of the solution proposals.

Based on the scores that students received, we classified them into higher-performing, medium-performing, and lower-performing groups. The top one-third of the groups with the highest scores were classified into the higher-performing groups, the bottom third were classified into the lower-performing groups, and the remaining ones were classified into the medium-performing groups. Out of the 32 groups, the higher-performing groups (n = 10) received scores of 23–28 points for their proposals, the medium-performing groups (n = 12) received scores of 21–22 points, and the lower-performing groups (n = 10) received scores of 16–20 points.

We compared the scores in the four dimensions for the higher-, medium-, and lower-performing groups. As the scores were not normally distributed, Kruskal–Wallis tests were conducted to examine the differences between groups at these three levels, and the results indicated significant differences in all four dimensions of defining design goals, reasoning, technical knowledge and skills, and originality.

Defining design goals

The Kruskal–Wallis test results of this dimension showed a statistically significant difference in the scores of the three groups, χ2(2) = 13.624, p = 0.001, η2 = 0.401, with mean rank scores of 22.5 for the higher-performing groups, 17.33 for the medium-performing groups, and 9.5 for the lower-performing groups. According to Cohen (1988), effect sizes equal to or above η2 = 0.14 are considered large. Post-hoc Bonferroni tests showed that the higher-performing groups scored significantly higher than the lower-performing groups. However, there were no significant differences between the higher-performing and medium-performing groups. See Table 2. By comparing the submissions of the higher-performing and lower-performing groups, we found that the higher-performing groups were more expressive, as they not only introduced the name of a product, but also explained the background and importance of the product in one or two concise sentences. For example, one higher-performing group illustrated their design goals as “We hope to create a user-friendly product (i.e. social distance detector) to help the visually-impaired to be able to maintain social distancing and decrease the risk of being infected under the pandemic,” which made the purpose and reason behind clearly stated. The lower-performing groups tended to present their goals without much justification or elaboration, e.g. “Help the visually impaired to recognize words (i.e. smart pen).”

Table 2. Comparison of proposal scores on each dimension for the higher-, medium-, and lower-performing groups.

Dimension	Performance level	n	Mean rank	Kruskal–Wallis test result
Defining design goals	Higher	10	22.50	χ^2(2) = 13.624, p = 0.001
	Medium	12	17.33	Higher > Lower, p = 0.001
	Lower	10	9.50
Reasoning	Higher	10	21	χ^2(2) = 9.377, p = 0.009
	Medium	12	18.17	Higher > Lower, p = 0.01
	Lower	10	10
Technical knowledge and skills	Higher	10	23.85	χ^2(2) = 16.904, p < .001
	Medium	12	17.54	Higher > Lower, p < .001
	Lower	10	7.90	Medium > Lower, p = 0.03
Originality	Higher	10	23	χ^2(2) = 11.22, p = 0.004
	Medium	12	13.75	Higher > Lower, p = 0.01
	Lower	10	13.3	Higher > Medium, p = 0.011

Reasoning

The Kruskal–Wallis test results of this dimension indicated a statistically significant difference in the scores of the three groups, χ2(2) = 9.377, p = 0.009, η2 = 0.254, with mean rank scores of 21 for the higher-performing groups, 18.17 for the medium-performing groups, and 10 for the lower-performing groups. Post-hoc Bonferroni tests showed a significant difference only between the higher- and lower-performing groups. Further review of the proposals indicated that the higher-performing groups tended to summarize relevant data to support the importance of the product addressing the needs of the community. In contrast, the lower-performing groups might directly quote links to news reports or videos on this dimension without synthesizing the content.

Technical knowledge and skills

The Kruskal–Wallis test results of this dimension showed a statistically significant difference in the scores of the three groups, χ2(2) = 16.904, p < 0.001, η2 = 0.514, with a mean rank score of 23.85 for the higher-performing groups, 17.54 for the medium-performing groups, and 7.9 for the lower-performing groups. Post-hoc Bonferroni tests showed a significant difference between the higher- and lower-performing groups and between the medium- and lower-performing groups. There were no significant differences between the higher-performing and medium-performing groups. The student proposals showed that the higher-performing groups described the selection of input and output devices, processors, and explained how the components would be connected to achieve the desired effect. In contrast, the lower-performing groups skipped this section or listed a few related names (e.g. sensors or a processor platform) without explaining how the components work. Informal conversations with teachers indicated that one group did make significant progress in designing a price recognizer APP, but the students did not report on it in their proposal. Thus, we need to consider whether the lower-performing students were less technologically proficient or unfamiliar with describing the technological progress in the form of a proposal, and adjust our supporting strategies.

Originality

The Kruskal–Wallis test results of this dimension showed a statistically significant difference in the scores of the three groups, χ2(2) = 11.22, p = 0.004, η2 = 0.318, with mean rank scores of 23 for the higher-performing groups, 13.75 for the medium-performing groups, and 13.3 for the lower-performing groups. Post-hoc Bonferroni tests showed that the higher-performing groups significantly outperformed the medium- and lower-performing groups, but no significant differences were identified between the medium- and lower-performing groups. Higher-performing groups proposed novel products and features that were not available in the market or not mentioned by other students, e.g. the social distance detector app. Descriptions of the product’s uniqueness were usually consistent with the market reality and reflected students’ understanding of user needs. For the lower-performing groups, the proposed products may be similar to what were already available in the market or other groups’ ideas, e.g. a smart pen for translating English into Cantonese.

The overall results of the performance comparison indicate that the higher-performing groups scored significantly higher than the lower-performing groups in all four dimensions and significantly higher than the medium-performing groups in originality. The medium- and lower-performing groups received low scores in the originality dimension. The lower-performing groups received their lowest scores in the technological knowledge and skills dimension. There was not a great difference in scores between the higher- and medium-performing groups, and the difference was mainly in the dimension of originality. The medium-level students had the potential to produce higher-quality submissions if given a certain amount of support, especially to boost originality. However, the lower-performing groups needed further support in all dimensions, especially technological knowledge and skills.

STEM content knowledge

A total of 114 students completed both the pre- and post-tests that were administered to track changes in their STEM content knowledge. The results of a paired-sample t-test indicated a significant increase in students’ test scores from the beginning (M = 14.44, SD = 5.18) to the end of the program (M = 16.68, SD = 4.16), t (113) = −4.85, p < 0.001. The interdisciplinary STEM program appears to have effectively increased students’ STEM content knowledge.

Discussion

The interdisciplinary STEM program examined in this study was designed with reference to the integrated STEM-based creative problem-solving model and contextual factors of the school where the program was run. Overall, the students achieved significant growth in factual knowledge and were able to generate a diverse range of solutions through the creative problem-solving approach. After categorizing the student groups into three performance levels according to the scores they received for their design proposals, the higher-performing groups outperformed the lower-performing groups in all of the dimensions of the assessment rubric. The main difference between the higher-performing and medium-performing groups was in the originality dimension, and the medium-performing groups had the potential to produce higher-quality solutions with additional guidance. The main difference between the medium- and lower-performing groups was in their levels of technical knowledge and skills.

Diversity of solutions

Most groups generated unique design ideas within the themes of safety, recreation, and convenience. The fact that the students were able to develop such a variety of solutions might attest to their efforts to think from users’ perspectives and sense users’ multidimensional needs. Another reason for the demonstration of divergent thinking might be the efforts made by the teachers to guide students to approach their thinking from multiple angles. In the community service learning sessions, the teachers provided the students with examples of the difficulties faced by the visually impaired regarding clothing, food, housing, and transportation rather than emphasizing difficulties in only one area of their lives. This sharing of examples from multiple dimensions might have inspired a broader range of solutions, which would be consistent with the finding of Yi et al. (2015) that modeling positively affected divergent thinking and artistic creativity. Furthermore, we gave the students room to conceptualize their proposed product and design whatever made sense to them. Students are more likely to be creative when offered the space to explore various possibilities (English, 2019; Foster & Schleicher, 2022).

Quality of solutions

The overall quality of the students’ solutions was satisfactory, with 73% scoring above 21 points in the design solutions. This result might be related to adopting three quality assurance strategies in implementing a real-world design-based approach. First, the program was structured with reference to the integrated STEM-based creative problem-solving model, which accounts for the intrinsic and extrinsic factors that promote individual or group innovation, such as providing a supportive learning environment and giving students opportunities to learn relevant knowledge and skills (Amabile, 2012; Chiu et al., 2023). Second, to guide the students through the creative problem-solving process, we provided a comprehensive rubric to inform them of the main aspects that would be considered in assessing their solutions and how each of these aspects would be measured (Biggs, 2003; Jonsson, 2014; Peppler et al., 2023). Third, examples were provided with this rubric to give students a more concrete idea of each dimension.

The higher-performing groups outperformed the lower-performing groups on all rubric dimensions, whereas the main difference between the higher- and medium-performing groups was originality. This result suggests that strategies should be adopted to encourage the medium- and lower-performing groups to think outside the box and put more effort into generating different ideas. For example, attempts could be made to share additional examples of creative problem-solving to broaden these students’ horizons (Yi et al., 2015). The disparity in performance between the medium- and higher-performing groups was not remarkable, and it is likely that with targeted training, the overall performance of the medium-performing groups would improve.

It is noted that the lower-performing groups had difficulties in completing the technical knowledge and skills section. According to the teachers, some lower-performing groups completely skipped the technical knowledge and skills sections, signaling those students needed extra support at this stage. More scaffolding (Cui et al., 2023) is needed to strengthen students’ ability to express how their solutions can be actualized through STEM tools and devices (e.g. input and output devices, and the operating mechanism of the proposed product). Lower-performing groups could be encouraged to carefully work through the sample proposals and learn the technical expressions to tackle this situation. Moreover, students could be encouraged to seek one-on-one counseling when unsure of writing the technical section. Furthermore, efforts are needed to promote the lower-performing groups’ interest in STEM, as it is correlated with persistence and learning achievement (Crippen et al., 2016; Nuutila et al., 2020). As Sternberg (2006) noted, people rarely unlock their creative potential in a particular field unless they have a genuine passion for their work.

Theoretical and practical implications

In terms of theoretical implications, this study demonstrates that the integrated STEM-based creative problem-solving model is proficient in guiding the design process and promoting creative design solutions. Students are able to address real-world challenges—in this case, designing products to improve the lives of visually impaired groups. The students not only proposed product concepts but also justified their beliefs that the product was important, used specific technical knowledge at an introductory level to actualize the solutions, and wrote up a proposal. These tasks are quite challenging for students in Grade 8, and more than 70% of the groups came up with reasonable solutions. Though other scholars have attempted to train students’ creativity in various ways, such as through the design of Christmas party posters (Lin et al., 2020) and the improvement of watches (Sun et al., 2022), the direct integration of community service challenges into STEM education by engaging students in the design of products that help underprivileged groups is a practice rarely documented in the literature.

In terms of practical implications, regarding the school context, we sought to offer more support to students at the school level by providing them with the assessment rubric. By empowering the students to understand what solutions were considered good and what solutions were considered poor, we guided them to pay more attention to their creative problem-solving process and the quality of their design solutions. This practice removed many barriers to setting clear goals and allowed the students to understand the performance expectations and to plan and self-assess accordingly (Dawson, 2017). The transparency of the link between the task and the assessment strategies may have facilitated the development of the students’ disciplinary knowledge and skills (Jonsson, 2014; Reynders et al., 2020).

It is also important to emphasize the combination of STEM technical content knowledge learning with community service learning. Several female students mentioned to our project staff that they had not been interested in STEM until participating in this program, which is consistent with recent research findings on interest in STEM among girls (e.g. Ünlü & Dökme, 2020). In this interdisciplinary STEM program, however, they were quite willing to participate in the learning activities. Perhaps this can be explained by people’s tendency to participate more actively in activities they recognize the value of (in the present case, making people’s lives easier) (Afzal & Husain, 2020). The opportunity to contribute to the community and to use their knowledge to work toward solving real issues might have led these students to a broader appreciation of the STEM disciplines (Bringle & Hatcher, 1996; Tijsma et al., 2020).

Recommendations, limitations, and future research

In reflecting on the current study, we make the following recommendations for research and practice. First, a theory-grounded interdisciplinary design model should guide the design process. Models serve as a bridge between theory and practice and help people visualize abstract concepts and understand processes (Chittleborough & Treagust, 2009; Gilbert, 2004). In this study, we adopted the integrated STEM-based creative problem-solving model, which elaborated the steps for designing an interdisciplinary program and offered a coherent understanding of the elements that foster creativity. Incorporating the model increased efficiency in designing the interdisciplinary program and saved us much time in the planning stage.

Second, contextualized support should be provided to students. As the integrated STEM-based creative problem-solving model indicates, support at the school level is important for students to develop their creative problem-solving skills. Amabile and Pratt (2016) also mentioned the importance of having access to the necessary resources for creative problem-solving. In the program evaluated in this study, we provided access to resources for understanding the needs of people with visual impairments. We made the assessment rubric, annotated with brief examples, available to students.

Third, the power of integrating community service learning with STEM learning should be leveraged. Values, empathy, and citizenship have always been key elements that educators seek to instill in their students. STEM fields are important for technological and economic development, but many students find these disciplines uninteresting or perceive them as difficult to learn. Community service learning and STEM learning promise to produce a positive synergy but have been the subject of few educational studies (Afzal & Husain, 2020). Making more attempts at this integration can open new educational possibilities.

Due to the constraint of resources, a single-group pre- and post-test design was adopted in this study. As with some other STEM studies, finding an equivalent comparison group in a natural STEM educational setting was difficult (e.g. Kuo et al., 2019). Where conditions permit, it would be ideal to recruit a comparison group to evaluate their effectiveness in the future. Another limitation of the study was that it was conducted in only one school, but student characteristics and school-level contexts may vary. Therefore, similar studies in different schools would be helpful for comparative purposes. In the future, we may also consider introducing a panel of creative product designers and using the consensual assessment technique (Amabile, 1982) to assess student solutions if resources allow. It would also be interesting to investigate how students of different ages respond to the integrated curriculum and how they might be best supported.

Conclusion

The study provided empirical evidence regarding the effectiveness of integrating real-world community service challenges into the learning and teaching process of STEM education. It demonstrated that the integrated real-world design-based approach stimulated students to provide diversified design solutions for helping people with visual impairments. The design solutions proposed by students were mostly of quality. However, further efforts are expected to provide differentiated support to the medium- and lower-performing groups and to help them progress to higher levels.

Acknowledgment

The work was substantially supported by The Hong Kong Jockey Club Charities Trust (Project Title: Jockey Club Community Care and STEM in Action [Project S/N Ref: JC 2019/0112]).

Disclosure statement

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

Additional information

Funding

This work was supported by The Hong Kong Jockey Club Charities Trust.

Notes on contributors

Biyun Huang

Biyun Huang, PhD, is an Assistant Professor in the School of Education, City University of Macau. Her research areas are transdisciplinary STEM, gamification, and motivation design in E-learning settings.

Morris Siu-Yung Jong

Morris Siu-Yung Jong, PhD, is a Professor in the Department of Curriculum and Instruction and the Director of the Centre for Learning Sciences and Technologies, The Chinese University of Hong Kong. His research areas are educational use of virtual reality, blended learning, gamified learning and STEM/AI education.

Chin-Chung Tsai

Chin-Chung Tsai, EdD, is a Chair Professor in the Program of Learning Sciences and the Dean of the School of Learning Informatics, National Taiwan Normal University. His research areas are science education, E-learning and educational technology.

Junjie Shang

Junjie Shang, PhD, is an Associate Professor in the Graduate School of Education and the Director of the Lab of Learning Sciences, Peking University. His research areas are game-based learning, learning sciences, educational technology, and E-leadership.

Appendix A.

Rubric for creative problem-solving design proposals

Table

Dimension	Sub-dimension	Score range	Criteria
Defining design goal	Goals (i.e. the effects that the product can achieve)	0–3	Level 3: Excellent. Very specific, accurate, and reasonable descriptions. Issues that can be addressed by the product are clearly listed. Level 2: Good. Specific, accurate, and reasonable descriptions. Issues that can be addressed by the product are listed. Minor mistakes. Level 1: Fair. Vague and unreasonable descriptions. Issues that can be addressed by the product are barely listed. Major mistakes. Level 0: No coverage.
Reasoning	Needs of society (i.e. the problem that has been identified)	0–3	Level 3: Excellent. Very specific, accurate, and reasonable descriptions. Displays empathy and a precise understanding of users’ difficulties. Reliable and relevant information cited to support the main points. Level 2: Good. Specific, accurate, and reasonable descriptions. Displays empathy and an understanding of users’ difficulties. Minor mistakes. Level 1: Fair. Vague and unreasonable descriptions. Displays a low level of empathy and understanding of users’ difficulties. Major mistakes. Level 0: No coverage.
	Product impact (i.e. potential impacts on the community or industry)	0–3	Level 3: Excellent. Very specific, accurate, and reasonable analysis of the potential influence on the community or industry. Level 2: Good. Specific, accurate, and reasonable analysis of the potential influence on the community or industry. Minor mistakes. Level 1: Fair. Vague analysis of the potential influence on the community or industry. Major mistakes. Level 0: No coverage.
Technical knowledge and skills	Technical features (i.e. the main functions and how they work)	0–6	Level 3: Excellent. Very specific, accurate, and reasonable descriptions of the product features and functions (2 or more items). Brief description of how each function works. Level 2: Good. Specific, accurate, and reasonable descriptions of the product features and functions (1 or more items). Brief description of how each function works. Minor mistakes. Level 1: Fair. Vague description of the product features and functions. Failure to describe how each function works. Major mistakes. Level 0: No coverage.
	Making process (i.e. materials, devices, and process for making the product)	0–6	Level 3: Excellent. Very specific, accurate, and reasonable descriptions of the materials required for making the product. Sufficient evidence for the feasibility of developing the product. Level 2: Good. Specific, accurate, and reasonable descriptions of the materials required for making the product. Partial evidence for the feasibility of developing the product. Minor mistakes. Level 1: Fair. Vague description of the materials required for making the product. Minimal evidence for the feasibility of developing the product. Major mistakes. Level 0: No coverage.
	Suggestions for future products	0–3	Level 3: Excellent. Very specific, accurate, and reasonable suggestions for future product improvements (2 or more items). Level 2: Good. Specific, accurate, and reasonable suggestions for future product improvements (1 or more items). Minor mistakes. Level 1: Fair. Vague suggestions for future product improvements. Major mistakes. Level 0: No coverage.
Originality	Uniqueness (i.e. the uniqueness of the product, or the advantages of the product over similar products)	0–6	Level 3: Excellent. Very specific, accurate, and reasonable descriptions of the product’s unique advantages (2 or more items) over others on the market. Level 2: Good. Specific, accurate, and reasonable descriptions of the product’s unique advantages (1 or more items) over others on the market. Minor mistakes. Level 1: Fair. Vague description of the product’s unique advantages over others on the market. Major mistakes. Level 0: No coverage.