A pedagogical study on promoting students' deep learning through design-based learning

DBL originated from the “learning by design” model proposed by Kolodner and was later followed by the “design-based scientific inquiry learning cycle” model (Kolodner, 2002). The two main elements of this model are design and redesign, and investigation and inquiry. Later, Nelson created and put into practice and promoted “design-based learning”with considerable success in K12 interactive classrooms (Zhu et al., 2017). Nelson's “Backward Thinking” is one of the well-known DBL learning models. The model is roughly divided into six steps. That is to clarify the theme, determine the problem, establish evaluation standards, establish models, teaching guidance, evaluation and modification. Compared with the traditional teaching model, the “Backward Thinking” model highlights the iterative nature of the learning process and emphasizes the comprehensiveness of various disciplines in the learning challenge.

The “Backward Thinking” model is derived from Bloom's classification of cognitive goals, which classifies cognitive learning from low to high into six levels: knowledge, comprehension, application, analysis, synthesis and evaluation. This is one of the most common ways of evaluating the results of deep learning in terms of cognitive dimensions. Traditional education and learning focus on the memory and understanding of knowledge. However, true learning requires students to actively analyze, synthesize, apply, and evaluate facts and ideas. According to Nelsen, traditional “forward” teaching begins with basic facts, while “Backward Thinking” begins with the highest level of reasoning. Students' higher-order thinking and higher-order capabilities can be developed through the “Backward Thinking” model. In addition, Fortus et al. (2004) proposed the “Design-Based Science learning cycle”. The model consists of five steps: (1) Identify and Define Context; (2) Background Research; (3) Develop Personal and Group Ideas; (4) Construct 2D and 3D Artifacts; (5) Feedback.

From the above model, it can be seen that the DBL approach has the following characteristics. (1) Situational. The DBL approach emphasizes that learners can carry out inquiry activities in a task situation, so that learners can “learn by doing”. (2) Design. This is the core of DBL (Huang et al., 2016; Li et al., 2015). Design is an important tool for problem solving. Design thinking is always present throughout the learning process. (3) Integrative. In the DBL process, students need to participate in small groups and collaborate with others to design. Students tackle design challenges by integrating the knowledge and skills of different disciplines. (4) Iterative. The DBL process is an iterative cycle. Students' design solutions need to be revised and optimized several times. (5) Reflective. Students continue to learn from their experiences through reflection in the design practice process. Teachers provide students with timely feedback, making them clear about their learning and constantly adjust their learning plans.

Deep learning has been a popular research topic of interest in education in recent years. In 1976, Marton and Saljo at the University of Gothenburg introduced deep learning and shallow learning based on differences in the way information was processed (Marton & Saljo, 1976). Deep learning is high level or active cognitive processing, while shallow learning is low-level cognitive processing (Biggs, 1979). Biggs (1999) argued that deep learning was the process of making connections between old and new knowledge through critical thinking. Entwistle (2000) argued that deep learning was an active way of learning with the aim of understanding meaning. Zhang et al. (2012) argued that deep learning advocated active, critical and meaningful learning. From the above discussion, it is clear that deep learning refers to learners critically learning and reflecting in authentic situations, actively making connections between old and new knowledge systems, and solving complex problems through deep information processing.

Currently, research on deep learning has been conducted in three main types. In the first type, deep learning is used as a learning approach. Wang et al. (2021) applied deep learning theory to instructional design, and the results showed that the new teaching approach achieved high satisfaction levels, as well as improved student performance. He et al. (2021) combined deep learning with college ideological and political courses to teach. The results showed that the improved curriculum was able to attract students and led to progress in terms of values, ideology and knowledge base. In the second type, deep learning is used as a learning process. Among these, the 3P (Presage-Process-Product) model, developed by Biggs, has been influential. This model measures deep learning in terms of two dimensions: motivation and strategy. In the third type, deep learning is used as a learning outcome. By exploring effective teaching models to promote deep learning among students. Chen et al. (2016) designed a flipped classroom that led to an increase in graduate students' cognitive level and significantly enhanced deep motivation and engagement in learning. Terrenghi et al. (2019) investigated in a high school classroom that “Episode of Situated Learning” in a high school classroom and showed that it could increase student engagement and promote deep learning.

Although scholars have conducted research in classrooms across different disciplines, there is less theoretical and practical research on the assessment of students' deep learning. The National Research Council (NRC) has classified the competencies developed by learners in deep learning into three main dimensions: cognitive, interpersonal and personal (Trigwell & Prosser, 2001). This is highly consistent with the six competencies for deep learning proposed by the Hewlett Foundation (Pu et al., 2016). As shown in Table 1, the framework breaks away from traditional cognitive boundaries to include the interpersonal and self domains in the evaluation of deep learning. It provides an important reference for constructing specific evaluation dimensions and metrics.

The DBL approach emphasizes responding to challenges, integrating knowledge and applying skills to solution design in real-life situations. Students use established evaluation criteria to continually optimize their schemes, proactively add the required knowledge and ultimately achieve a problem-solving outcome. The approach is in line with Bloom's classification of cognitive goals. Learners achieve the cognitive level by applying knowledge, analyzing problems, synthesizing solutions and evaluating designs. At the same time, DBL also corresponds to the highest level of learning in Gagne's hierarchy system of learning. It means that learning takes place through problem solving. He et al. (2005) identified problem-based learning, task-based learning and process assessment as teaching strategies that effectively promoted deep learning. In 2015, the Horizon Report Basic Education Edition also identified project-based learning, problem-based learning, inquiry-based learning, and challenge-based learning as deep learning approaches that help students gain more active learning experiences (Jiao, 2015). Students learn more deeply when they are asked to design and produce work that requires understanding and application of knowledge (Feng & Miao, 2009). In the process of solving design problems, DBL enables learners to master the core content of the subject and apply their knowledge. It is a deeper approach to learning that enables individual knowledge construction (Zhu et al., 2017) and promotes the integration of multidisciplinary knowledge (Puente et al., 2011).

Many scholars have studied DBL to enhance learners from the empirical level. Kafai (2006) used DBL in teaching game programming, and the investigation showed that students' knowledge integration ability as well as their expressive ability were improved. Rao et al. (2018) studied that DBL could effectively promote students' creativity development. In a “Bridging Program” project conducted by Ayub et al. (2015), DBL greatly improved students' critical thinking ability.

In general, DBL incorporates the concept of deep learning to a certain extent (Du et al., 2013). The method of DBL can effectively exercise higher-order thinking and higher-order capabilities and promote deep learning among learners. As the future development trend of education and learning concepts, deep learning is an important goal pursued by DBL.

According to the learning model and characteristics of DBL, and combined with the actual situation of engineering design education, this paper constructs a DBL teaching model that promotes deep learning from the perspectives of teaching stages, teacher and student activities and deep learning. As shown in Fig. 1, it is divided into three stages in total. These include the stage of creating a situation, the stage of designing scheme, and the stage of evaluating and reflection.

The activities of the teacher are arranged as follows. In the first step, the teacher carefully arranges the teaching and learning tasks according to the teaching objectives and the actual situation. The learning task needs to be challenging and stimulate the students' willingness to participate and investigate. In the second step, the teacher creates a teaching situation based on the design of the task so that the content can be taught. It includes teaching the necessary new knowledge and skills. In the third step, the teacher groups students and provides the necessary resources and technical tools for the task.

The students' activities are arranged as follows. In the first step, students memorize and understand the new knowledge explained by the teacher and actively recall their previous knowledge base. In the learning process, the old and new knowledge systems are constantly coupled and deepened. In the second step, students work in small groups to identify what problems and constraints exist through information gathering and analysis tasks. Students serve a common practical project by integrating and sharing information and resources from different areas of expertise. In the third step, students collate and analyze the issues from their research, including core and secondary issues, and share them in class using a PowerPoint presentation.

At this stage, students are fully motivated to learn. Through the memorization and understanding of theoretical knowledge in the research process, their ability to integrate and analyze information resources is exercised. It facilitates the formation of a systematic framework for compiling knowledge. At the same time, students learn to analyze problems critically and prioritize them.

The teacher's activities are arranged as follows. In the first step, the teacher assists students in interpreting the problem and finding practical problem points. In the second step, the teacher communicates the students' initial design scheme and determines the general design objectives and directions. In the third step, the teacher manages and organizes the students' instruction and asks them to plan their time.

The students' activities are arranged as follows. In the first step, students brainstorm through discussions between groups. Each person designs a considerable number of schemes. In the second step, students communicate within the group based on their schemes and select the appropriate ones. In the third step, students revise and optimize their chosen schemes. Later, a presentation is made and shared in class.

At this stage, students deepen their impressions through the application and analysis of their knowledge. At the same time, students learn to communicate with each other and solve problems together. Students have a broader design thinking, a richer design language and a more comprehensive approach to design.

The activities for teachers are arranged as follows. In the first step, the teacher designs evaluation criteria of the work based on the course content. In the second step, the teacher marks the students' work according to the evaluation criteria and gives suggestions for revision.

The students' activities are arranged as follows. In the first step, students will display and present their designs. In the second step, students self-evaluate their work and evaluate each other in groups according to the evaluation criteria. In the third step, students make timely revisions and reflections based on the feedback and comments from the assessment.

At this stage, students optimize and iterate on the design scheme based on actual situations and summaries of comments. They learn to reflect, acquire learning methods and skills, and persevere for their learning goals.

The teaching of the course took place over two years, including 2019 and 2021 (the online course conducted in 2020 due to COVID-19 did not allow for a situational teaching session). A total of 49 postgraduate students participated in the course in the class of 2019. The class was taught using traditional design instruction and set as a control class. A total of 56 postgraduate students participated in the course in 2021. The class was taught using DBL teaching and set as an experimental class. They all came from different design disciplines and had a good knowledge base. The experimental class included 16 product design students, 14 environmental design students, 21 visual communication design students and 5 public art design students. Before the course started, 56 students were divided into 7 groups. Each group consisted of 8 students, 1 group leader and 7 group members. There were at least 3 different majors in the group to fully ensure that the groups were well staffed. The same grouping was used for the control class. At the end of the course the experimental and control classes used the same marking criteria to mark the students' design work.

At the end of the course, each student scored based on teacher and student evaluations and student performance in class. The specific evaluation criteria were listed in Table 2. The results were divided into four grades based on the four dimensions of integrity, rationality, innovativeness and aesthetic, grade A (100-90), grade B (89-80), grade C (79-70) and grade D (69-0). The total score was 100, with 50 points for integrity, 20 points for rationality, 20 points for innovativeness and 10 points for aesthetic. Integrity means that the work was very complete, with rich content and good professional division of labour. Rationality represented a work that was realistic and highly feasible to operate. Innovativeness represented work that solved practical problems and created good value. Aesthetic represented work that reflected good design skills and aesthetic ability. The specific calculation of each student's grade is as follows: Y = x1 × 60% + x2 × 20% + x3 × 10% + x4 × 10%. Y represented the total student achievement score, x1 represented the teacher assessment score, x2 represented the group mutual assessment score, x3 represented the individual self-assessment score, x4 represented the classroom performance score and attendance score. Students and teachers graded on the basis of uniform evaluation criteria. Ultimately, the student's course grade was based on the percentage of marks given by teachers and students.

This study was based on the “deep learning: opportunities and outcomes (student questionnaire)” developed by the deep learning research project SDL, on Biggs' learning process scale (SPQ) and the deep learning questionnaire developed by Li et al (2018). The Deep Learning Process Questionnaire (Appendix 1) and the Deep Learning Ability Questionnaire (Appendix 2) were designed. Both questionnaires were scored on a 5-point Likert scale, ranging from “strongly disagree”, “disagree”, “not necessarily”, “agree” to “strongly agree” on a scale of 1, 2, 3, 4 and 5, respectively.

In order to ensure the reliability and validity of the “deep learning process questionnaire” and “deep learning ability questionnaire”, two questionnaires were distributed to 78 students before formal teaching. In this study, the questionnaire was analyzed for reliability and validity using IBM SPSS 25.0, as shown in Tables 3 and 4. The deep learning process questionnaire consisted of three dimensions: learning motivation (A1–A5), learning input (A6–A11) and learning strategy (A12–A16). There were five questions in the “learning motivation” dimension, six questions in the “learning input” dimension and five questions in the “learning strategy” dimension. The overall reliability coefficient of the Deep Learning Process questionnaire was 0.869, indicating that the overall reliability of the questionnaire was very high, and the KMO measure was 0.866, meaning that the validity of the questionnaire was good. The deep learning ability questionnaire consisted of three first-level dimensions: cognitive, interpersonal and personal. It also consisted of six second-level dimensions: critical thinking (B1–B3), problem solving (B4–B6), teamwork (B7–B9), effective communication (B10–B12), learning to learn (B13–B15) and learning to persevere (B16–B17). There were six questions in the “cognitive” dimension, six in the “interpersonal” dimension and five in the “personal” dimension. The overall reliability coefficient of the deep learning ability questionnaire was 0.888, and the reliability coefficients of the other primary and secondary dimensions were all above 0.7. The KMO value of 0.843 indicated that the questionnaire had good validity.

As shown in Table 5, the experimental classes differed significantly (p < 0.05) in the dimensions of learning motivation, learning input and learning strategy. In the pre-test data, the mean values for the three dimensions of students' deep learning status were 3.46, 3.67 and 3.41. After the implementation of DBL instruction, all three were improved, with mean values of 4.03, 3.97 and 3.74 for the post-test data. Learning motivation increased relatively the most before and after the implementation of DBL. The mean score of 4.03 was relatively high. This indicated that students' learning motivation increased and they were more inclined to take the initiative in learning during the teaching and learning process.

The results in Table 6 shown that the p values for the post-test were all less than 0.05, indicating that there were significant differences between the experimental group's six deep learning abilities before and after the experiment. All abilities improved compared to the previous ones. Among them were critical thinking ability (T=−3.173, p = 0.002 < 0.05), problem solving ability (T=−3.830, p = 0.000 < 0.05), and learning to learn ability (T=−3.923, p = 0.000 < 0.05). This indicated that students' improvement in the three dimensions of critical thinking, problem solving and learning to learn was particularly evident. In the interpersonal area, there was a significant difference between the pre-test and post-test results for effective communication ability (p = 0.0026 < 0.05). It indicated an increase in effective communication ability. However, the mean values of 3.22 and 3.60 for the pre-test and post-tests were relatively low. It indicated that there was a need to further improve teaching strategy in the study of effective communication.

The results of the control and experimental classes were graded separately after the lesson according to the same grading scale. As Table 7 shown, the experimental class performed better than the control class after the optimization of the teaching reform. Their difference reached a significant level (t = 3.561, df = 103, P< 0.05).

As shown in Figure 4, the percentage of students in the control class scoring in the four grades of A, B, C and D were 16.3%, 46.9%, 32.7% and 4.1% respectively. The experimental class scored 25.0%, 55.4%, 19.6% and 0.0% in the four grades A, B, C and D respectively. After the implemented DBL teaching model, the experimental class scored significantly better at A and B levels, significantly lower at C level and had zero D level scores. It indicated that the reformed curriculum had led to an increase in the proportion of students achieving outstanding results. The situation of students' design thinking and design skills has improved. It indicated that the DBL teaching model was effective in improving the quality of teaching and learning.

To measure students' perceptions of DBL teaching, Table 8 showed the survey results based on the Likert scale. The combination of “strongly agree” and “agree” combined creates a positive answer of “agree”. Similarly, “disagree” and “strongly disagree” combined form a negative answer of “disagree”. “Not necessarily” was a neutral attitude. As shown in Table 8, 85.7% of the responses were positive in terms of “active and deep learning”. Only 3.6% were negative. It indicated that active thinking and deep learning could be effectively promoted by changing the curriculum. The second highest value of 82.7% of positive responses was “innovative and interesting teaching”. It had only 5.4% negative responses. Only 69.7% of the positive responses were “rich and efficient interaction”. There was also a neutral response of 19.6% and a negative response of 10.7%. It indicated that interaction between students from different majors was problematic and that there was a need for more logical ways to bring students closer together. The answers for “scientific and rational training”, “diverse and integrated knowledge” and “timely and rapid feedback” were all highly positive, at 73.2, 76.7 and 74.9% respectively. In summary, the students accepted the new teaching model. At the same time, the students were pleased with the content and effectiveness of the instruction. In terms of teaching organization, the communication and interaction within the different groups was not good enough, which needed further improvement later on.

One student from each of the experimental classes with an A, B and C grade was selected for an interview after the lesson (Appendix 3). The results of the interviews were presented below. An A-level student found the course activities meaningful. He would take the initiative to learn different professional knowledge and look at issues critically. He felt a little different when interacting with students from different majors but was able to deal with it on time. A B-level student found the course very practical as they learned to apply their knowledge to solve practical problems. The course content was challenging. However, through their learning and working with students from different majors. The student solved the problems well and gained a sense of achievement. A C-level student felt that the course required collaboration with students from different majors. We all had different perspectives on understanding the problems, which led to some barriers to collaboration. However, they took the initiative to learn and contribute to the group. It was clear from the above results that students with good grades took the initiative to construct their own body of knowledge. They could deal critically with problems and communicate efficiently with students from different majors. The low scorers were still confined to their expertise and did not take the initiative to learn different professional content. They were narrowly considered and mechanical in their approach to problems.

The results of the questionnaire indicated that the deep learning of the students had improved. In terms of deep learning status, students' learning motivation was higher. They were more engaged in their learning, and their learning strategy were strengthened. In terms of deep learning results, students' critical thinking ability, problem solving ability, and learning ability had improved significantly. At the same time, teamwork ability and learning perseverance had improved. For effective communication, the improvement was relatively low. It required further research later. Over, the new teaching model helped develop higher-order thinking and higher-order capabilities and promoted deep learning for students.

The survey results indicated that students' grades had improved. The percentage of exceptional students had risen. Student achievement evaluation criteria are an important aspect of determining the quality of teaching and learning. This paper construed specific evaluation criteria. Students and teachers marked the work according to the evaluation criteria. The results showed a significant improvement in student performance through the teaching reform. The percentage of students who excelled increased more significantly. It indicated that the new teaching method had improved the quality of teaching and learning and helped to foster better talents.

The survey results indicated that students were more satisfied with the DBL method. They felt that it was a novel way of teaching. The interactive atmosphere of the classroom was increased through practical projects. It broke with the traditional rigid indoctrination style of teaching. They felt that the DBL method helped to promote their active thinking and deep learning.

In contrast to previous research, this paper focused on the deep learning and teaching situation of students. Many researchers also discussed the application of DBL in specific teaching situations. They studied the teaching effectiveness of the DBL approach (Jiang et al., 2020). They focused on students' learning experiences (Ai et al., 2020). Some scholars studied the implementation of DBL for cultivating students' ability (Huang et al., 2020; Puente et al., 2011). Some scholars conducted empirical studies combining the DBL method with other teaching methods (Zhang et al., 2021). Zhang et al. (2022) studied that specific DBL activities affect students' emotional experiences.

However, the above studies were not specific about the deep investigation of students' thinking and ability. The application and implementation of teaching methods should focus on the long-term development of students and the development of deeper thinking and ability. This paper focused on the specific situation and teaching of DBL for promoting students' deep learning. The results indicated that the DBL method could effectively promote students' deep learning and improve the quality of teaching and learning.