Single or Combined? A Study on Programming to Promote Junior High School Students’ Computational Thinking Skills

The Definition of Computational Thinking

The definition of CT continues to expand and change over time, and many scholars, educational policies, and organizations have been working hard to define CT. CT can be traced back to Papert (1980), who first explained CT as the thinking ability that children train and cultivate when learning through the computers. Wing (2006) defined CT as the use of basic concepts of computer science to solve problems, design systems, and understand human behavior. Since then, CT has formally entered the people’s field of vision as an essential skill in the 21st century, triggering an upsurge in CT research. Voskoglou and Buckley (2012) pointed out that CT has become a new problem-solving process due to its application in the computer science field. Similarly, Kalelioğlu (2015) developed a framework for describing CT skills based on problem-solving. Also, some scholars focus on defining and describing the composition model of CT elements. For example, Brennan and Resnick (2012) proposed a three-dimensional CT framework that includes three dimensions: computational concept, computational practice, and computational perspective. Shute et al. (2017) developed a CT skills model with six main elements including decomposition, abstraction, algorithm design, debugging, iteration, and generalization, which can be used for CT evaluation. In this study, we defined CT as thinking skills transfer and projection in the process of solving ‘real’ and important problems (Román-González et al., 2017). We used the CT standards developed by Selby and Woollard (2013) to detail CT elements, which were refined into the thinking skills of abstraction, decomposition, algorithm, evaluation, and generalization, presented through the test questions of the Bebras International Computational Thinking Challenge.

The Teaching of Computational Thinking: Programming

In the international context, there are two main methods for cultivating students’ CT skills in programming classrooms: plugged-in and unplugged (Del Olmo-Muñoz et al., 2020; Sun et al., 2021b). Among them, plugged-in programming has been widely promoted and applied in K-12, and its effect has also been supported by sufficient evidence. Zhang and Nouri (2019) reviewed 55 empirical studies that provided evidence for the development of CT through Scratch programming. Pérez-Marín et al. (2020) verified a metaphor-based method using Scratch to teach basic programming concepts and found that elementary school students’ CT skills have been developed in the process. Chou (2020) observed that students’ CT skills have been significantly improved by integrating ScartchJr into the programming course, and most of their CT skills are retained one month after the end of the course. In addition, introducing CT without computers has also become one of the focuses, the so-called ‘unplugged programming’. Evidence has emerged, unplugged programming can also effectively develop students’ CT skills and help them transfer CT skills to other problem-solving situations (Hermans & Aivaloglou, 2017). Besides, the unplugged activities provide a broader framework than the plugged-in activities, focusing on a broader range of CT skills. For example, Looi et al. (2018) used unplugged activities to enable ninth graders to explore computational concepts and found that unplugged activities can be a scaffold to promote students’ CT skills. A quasi-experiment in elementary schools showed that students who participated in unplugged activities improved more decomposition, pattern recognition, abstraction, algorithm design, and other specific skills (Brackmann et al., 2017). Unplugged activities can be understood as a catalyst to support learning programming, but if we want to involve different types of students, we need the support of appropriate teaching methods and pay attention to the role of teaching situations (Bell, 2021).

Although the technical tool presented by plugged-in and unplugged programming are not consistent, in essence, both programming approaches point to the thinking connotation beyond code, especially CT skills. Of course, these two programming approaches are not an independent choice. Combining the two approaches in the programming classroom may achieve even better results. Saxena et al. (2020) designed a two-stage programming activity in early childhood education: the first stage is unplugged, and the second stage is plugged-in. It was found that the unplugged activities laid the foundation for the subsequent plugged-in activities. Similarly, Del Olmo-Muñoz et al. (2020) conducted a quasi-experimental study and found that a combined approach that combines unplugged and plugged-in activities (first unplugged and then plugged-in) is more effective in developing primary students’ CT skills than just teaching through plugged-in activities. As mentioned above, the existing research using combined programming approaches is mainly concentrated in kindergarten and elementary school, and there is still a lack of cases of combined programming approaches research in junior high school. In addition, the existing research on combined programming approaches is relatively simple, only the research on the combined programming approach of unplugged first with plugged-in after has been carried out, and there is a lack of research on the role of teaching sequence when the two single approaches are combined. How to better promote students’ CT skills in programming education still needs us to continue to explore.

Gender and Programming Experience in Programming on CT Education

Regarding the gender differences in programming and CT education, many studies have been conducted at present, but the results of the studies are not the same. Mouza et al. (2020) identified CT concepts before and after programming learning and examined potential gender differences and found that boys and girls have differences in CT, and boys have an advantage over girls. Del Olmo-Muñoz et al. (2020) also discussed the impact of unplugged activities on students’ CT development and found that boys develop faster than girls during this process. However, on the contrary, Noh and Lee (2020) found in a robotic experiment that the gender differences in students’ CT skills did not reach a statistically significant level. Similarly, Zhong et al. (2016) designed and implemented the Computational Thinking Three-Dimensional Integrated Assessment (TDIA) framework and found that in all CT tasks, the scores of boys and girls were not significantly different. Kalelioğlu (2015) found in the Code.org task that the reflective thinking skills towards problem-solving of females were slightly higher than that of males. These contradictory results have prompted scholars to analyze the causes of these gender differences more deeply, and strive to balance this gap by changing teaching methods. Angeli and Valanides (2020) found in the robotics classroom that students of different genders benefit from the two scaffolding techniques, but there is an interaction effect between gender and scaffolding techniques: boys benefited more from the individualistic, kinesthetic spatially oriented, and manipulative-based activities with cards, while girls benefit more from collaborative writing activities. The emergence of the Robotics Engineer Barbie™ is seen as a way to get girls interested in STEM, and toys may be used as a solution to gender equality in the field of computer science and as a way to create confident girls (Mauk et al., 2020). Just as Krommer (2006) suggested using gender-sensitive strategies in the classroom, it provided free space and activities for boys who cannot thrive in an environment that requires a lot of sitting and a lot of participation in language and speech and provided girls with activities that promote cooperation and emotional expression in language. It is also necessary to further study the gender difference, and ultimately it is necessary to choose more methods to encourage girls to choose as early as possible in the compulsory education stage, which may avoid the future gender gap.

Programming experience is also an influencing factor often considered in programming education. Helmlinger et al. (2020) investigated the neural processes underlying programming experience and found that individuals with rich programming experience may develop a form of computational thinking. Master et al. (2017) also pointed out that the lack of programming experience of young girls may be one of the main reasons for their less interest to participate in the fields of computer and engineering. Therefore, the influence of different individuals’ early programming experience should be taken into account when designing programming activities. How to make it easier for students with different programming experience to join in programming learning and how to design programming activities to better eliminate the influence of programming experience are topics that we should further discuss.

Research Objectives

In summary, the main purpose of this study is to evaluate the effects of different programming approaches on students’ CT skills and long-term retention when CT is introduced into junior high school. The four experimental groups respectively accept two single approaches activities (i.e., plugged-in and unplugged,) and two combined approaches that combine the two single approaches due to the teaching sequence (i.e., plugged-in-first and the unplugged-first). In the process, we also considered the role of students’ gender and previous programming experience in the effectiveness of the approaches in terms of CT skills. After the intervention, we conducted interviews with the teacher and conducted a questionnaire survey on 24 students, hoping to mutually confirm the quantitative results from the perspective of qualitative results. To achieve the study objectives, our research questions are as follows:

RQ 1. Can different programming approaches (single & combined) promote 7th graders’ CT skills? And which approaches will have stronger effectiveness?

RQ 2. Which programming approaches can make students acquire the maximum CT skills retention after two months of intervention?

RQ 3. Are there significant differences in the effectiveness of the approaches in terms of CT skills related to students’ previous programming experience?

RQ 4. Are there significant gender differences in the effectiveness of the approaches in terms of CT skills?

Method

Design

Based on the main objective of this study and considering that random sampling of samples cannot be achieved under actual educational conditions, we proposed a 5-month quasi-experimental design, including four experimental groups and one control group. The overall experiment can be divided into two parts: two-stage teaching sessions and four test links. Before the beginning of the first phase of the teaching session, students’ CT skills were tested to assess their initial CT level. In the first phase of the intervention, a total of four 40-minute unplugged activities and four 40-minute plugged-in activities were developed in four weeks. After the first phase, students took a CT mid-test to assess the impact of the first phase activities on students’ CT skills. Once the mid-test was completed, the second phase of programming teaching will begin. Similar to the first phase, we also developed four 40-minute unplugged activities and four 40-minute plugged-in activities in 4 weeks.

In the experimental intervention of a total of 8 lessons, of the four experimental groups, two of them used single programming approaches: the plugged-in group engaged in 8 plugged-in activities, and the unplugged group engaged in 8 unplugged activities. The other two groups used combined programming approaches: the plugged-in-first group engaged in 4 plugged-in activities in the first phase and engaged in 4 unplugged activities in the second phase, and the unplugged-first group engaged in 4 unplugged activities in the first phase and engaged in 4 plugged-in activities in the second phase. Our experiments were carried out in the school’s after-school practice class, and students usually do some homework exercises and discussions in this class. So, in the blank control group, the students only accepted some computer operation exercises in the teaching plan according to the school’s arrangement, such as Word and Excel operations, without involving any programming activities. After the second phase, the CT post-test was conducted to reassess the students’ CT development. To supplement the quantitative data results from the perspective of qualitative data, we also conducted interviews with the teacher and conducted a questionnaire survey of some students. In addition, two months after the intervention, we test students’ CT skills again to assess their CT retention. Figure 1 describes the development and implementation process of this study.

OPEN IN VIEWER

Figure 1.

Research Concept Map.

Participants

The research sample included 158 7th-graders from a public junior high school in Beijing, divided into five groups. Considering that it is impossible to complete random grouping in the natural teaching situation, the grouping is carried out based on the original teaching class, that is, each group is a separate teaching class, and the backgrounds of these classes are similar. Among them, there are 33 students (17 boys and 16 girls) in the plugged-in group, 33 students (17 boys, 16 girls) in the unplugged group, 29 students (13 boys and 16 girls) in the plugged-in-first group, 32 students (17 boys and 15 girls) in the unplugged-first group, and 31 students (15 boys, 16 girls) in the control group. Students from the same group gather in a classroom to participate in activities, but students from different groups are not gathered together. Besides, we collected students’ age, gender, and previous programming experience. Table 1 summarizes the information of samples in different groups. The average age of the sample students is 12.84 years, and in the overall sample, only 32.9% of the students have previous programming experience, while most of the students (67.1%) have not participated in programming activities.

OPEN IN VIEWER

Table 1.

The Information of Samples in Different Groups.

Group	N	Gender		Programming Experience
		Male	Female	Have	No
Plugged-in Group	33	17 (51.5%)	16 (48.5%)	10 (30.3%)	23 (69.7%)
Unplugged Group	33	17 (51.5%)	16 (48.5%)	12 (36.4%)	21 (63.6%)
Unplugged-first Group	29	13 (44.8%)	16 (55.2%)	8 (27.6%)	21 (72.4%)
Plugged-in-first Group	32	17 (53.1%)	15 (46.9%)	9 (28.1%)	23 (71.9%)
Control Group	31	15 (48.4%)	16 (51.6%)	13 (41.9%)	18 (58.1%)
Total	158	79 (50.0%)	79 (50.0%)	52 (32.9%)	106 (67.1%)

Participants selected for the questionnaire surveys included 24 students (12 boys and 12 girls) who participated in the study during this period. We adopt a voluntary registration principle, and all students participate in the survey voluntarily. We selected 6 students in each experimental group, 3 boys and 3 girls. In the process of selecting students, we try to diversify students’ programming experience, that is, students of different genders with different programming experiences. In the end, 24 representative participants were selected to represent different programming approaches, previous programming experience, and gender. All questionnaire surveys were conducted at the end of the second phase of intervention. The list of students under investigation and their representative characteristics are shown in Table 2.

OPEN IN VIEWER

Table 2.

The List of Students Under Investigation and Their Characteristics.

Group	Students	Gender	Programming Experience	Group	Students	Gender	Programming Experience
Plugged-in Group	Student A	Male	Have	Unplugged-first Group	Student M	Female	No
	Student B	Male	Have		Student N	Male	No
	Student C	Female	No		Student O	Female	Have
	Student D	Female	Have		Student P	Male	Have
	Student E	Female	No		Student Q	Male	Have
	Student F	Male	No		Student R	Female	No
Unplugged Group	Student G	Female	Have	Plugged-in-first Group	Student S	Male	Have
	Student H	Male	No		Student T	Female	No
	Student I	Female	Have		Student U	Female	Have
	Student J	Male	Have		Student V	Male	No
	Student K	Female	No		Student W	Female	No
	Student L	Male	No		Student X	Male	Have

Instruments

Computational Thinking Tests

Without measurement, CT will be difficult to successfully incorporate into K-12 classrooms. Kalelioğlu (2015) advocated for more discussion on assessing students’ grasp and application of CT skills in real life. Many scholars have built a series of tools for evaluating students’ CT skills based on the definition and framework of CT (Korkmaz et al., 2017; Román-González et al., 2017; Sáez-López et al., 2016). In this study, the CT measurement instrument is constructed from a series of task questions extracted from the ‘Bebras Computational Thinking Challenge’ to assess the extent to which students can transfer their CT skills to different types of problems and contexts (Román-González et al., 2017). Existing research has proven that Bebras is an effective tool for evaluating CT skills (Del Olmo-Muñoz et al., 2020; Tang et al., 2020). The Bebras Challenge is designed to help understand the CT skills of children aged 8–18. According to different ages, they are divided into five age groups, namely Castors (age 8–10), Benjamins (age 10–12), Cadets (age 12–14), Juniors (age 14–16), and Seniors (age 16–18). Dagienė and Futschek (2012) determined the criteria for a good competition task. The tests for each group of tests are divided into three difficulty levels: A, B, and C, and the difficulty of each difficulty level is the same. Based on the age characteristics of the research sample, this research selected and designed four sets of CT test questions from the 2017–2020 Bebras Challenge Cadets (12–14 years old). Each set of test questions contains 12 questions (some questions must be completed with the help of a computer, and paper assessment cannot be carried out so they are excluded), and finally contains 6 A-level questions, 3 B-level questions, and 3 C-level questions, combined the scoring rules of the contest and different points are assigned to the questions according to the actual difficulty of the questions, with a full score of 120 points. A full list of test items can be seen in Appendix A (online only). And the reliability and validity test results showed that the Cronbach’s alpha coefficients of the four sets of CT test questions are respectively 0.912, 0.907, 0.923, and 0.917, indicating that the tools used to test students’ CT skills in this study have good reliability and validity.

Combined with the definition of CT, it is not difficult to see that the CT skills represented in the Bebras test questions directly correspond to the CT skills described by Selby and Woollard (2013), which has also been adopted by Computing at School in the United Kingdom (Csizmadia et al., 2015), which includes abstraction, decomposition, algorithmic, evaluation, and generalization. For example, in the task of ‘Making Burgers’ (see Figure 2), students first abstract the problem and understand that the core of solving the problem lies in the difference of the topping; secondly, the problem is decomposed, the different interlayers corresponded to the letters, and the corresponding method is found by formulating an algorithm, and finally make a generalization to find the answer and evaluate the answer.

OPEN IN VIEWER

Figure 2.

The Example of Bebras Task—Making Burgers.

The Questions of Questionnaire and Interview

To supplement the results of the quantitative data from the qualitative results, based on study objectives, we formulated relevant questions to interview the teacher and conduct questionnaire surveys for students to understand their intuitive feelings in programming activities. The teacher’s interview questions are all open, mainly around the teacher’s teaching experience and the observed changes in the performance and mood of the students in different groups in activities, especially the performance of the students in the second phase of the two combined programming approaches groups. In addition, we have also formulated some questions to understand the performance problems of students of different genders and some insights from the teacher on cultivating students’ CT skills. The questions in the student questionnaire also take the form of open-ended questions, mainly around students’ emotional orientation after participating in programming activities, students’ cognition of the development of CT skills after participating in activities, and students’ understandings of the help of these activities in future programming learning. Questions were asked about the design of programming activities and the evaluation of difficulty, as well as students’ ideas about programming-related work in the future. At the end of the questionnaire, we solicited suggestions and opinions from students. The expression of the survey question in the questionnaire takes into account the students’ cognition level so that the question can be concise and clear. The teacher’s interview questions and students’ questionnaire survey questions are listed in Appendix B (Online only).

Code.org

Code.org (https://code.org/) was launched in 2013 and provided students with the opportunity to learn computer science through drag and drop programming. The website is available in 63 languages and its structure is similar to a game, which helps to arouse students’ attention to computer science. This course is suitable for students from kindergarten to eighth grade and above. The course content includes both online and offline formats. Common programming concepts involve activities such as sorting, loops, conditions, functions, variables, decomposition, patterns, abstractions, and algorithms (Kalelioğlu, 2015). This course uses a combined approach to teach programming, which means that students can learn from online, self-directed activities and teacher-led activities that do not use computers at all. As an open-source programming platform, Code.org is widely used in the field of K-12 programming education. For example, Arfé et al. (2020) found that programming exercises through Code.org can not only significantly improve children’s problem-solving ability but also improve their planning and control ability. Similarly, Kalelioğlu (2015) demonstrated the influence of using Code.org on elementary school students’ thinking and problem-solving ability. Therefore, this study developed programming activities based on some activities on Code.org (Chinese version), including plugged-in activities and unplugged activities. To plan the activities of different approaches of programming and the sequence, taking into account the grades of the sample students, we start from Code.org course E (8–12 years old, https://studio.code.org/s/coursee-2017), course F (9–13 years old, https://studio.code.org/s/coursef-2017) and the offline course section (https://code.org/curriculum/unplugged) selected and adapted different activities. An example of an plugged-in activity in Code.org is presented in Figure 3.

Data Analysis

The data sources in this study mainly include the results of students’ CT tests and the data of interviews with the teacher and representative students after the intervention. In this work, we focus on checking the students’ CT development based on the results of CT tests at different stages. We conducted four CT tests (pre-, mid-, post-, and postpone-test) to check students’ CT skills at different phases to record the changes. When dealing with quantitative data, we used descriptive statistical analysis to characterize students’ CT skills. Paired samples t-test was used for intra-group comparison, which can reflect the development of students’ CT skills. One-way ANOVA was used for comparison between groups to reflect the differences in students’ CT skills in different groups, and to visualize the development of students’ CT skills in different groups through the establishment of growth models. When analyzing the influence of students’ gender and previous programming experience on the CT skills development, firstly, one-way ANOVA were used to conduct a preliminary analysis, and secondly, the difference of change in students’ CT skills was analyzed by establishing the interaction terms of programming experience and gender. All analyses were performed using SPSS 26.00.

After the experiment, we interviewed the teacher and conducted a questionnaire survey on some representative students to obtain qualitative data to answer the research questions from all aspects and supplement the quantitative data. We interviewed and surveyed the teacher and 24 students to understand the teacher’s intuitive observations and feelings during the teaching process and students’ experience of participating in programming activities. Next, we used NVivo to conduct a qualitative analysis of the interview and survey data. We first transcribed the teacher’s interview recording and combine the notes to form the interview text, and then convert the student questionnaire into text. Second, all entries were entered into the NVivo and encoded the text, nodes were also created based on the topic of the interview, mainly including the key ideas of the teacher and students around the interview question.

Procedure

As mentioned in the design part, the implementation of the experimental course includes two teaching phases, each with four 40-minute courses. In each teaching phase, 4 sessions of plugged-in activities and 4 sessions of unplugged activities were developed respectively. A total of 16 programming courses in the two teaching phases enable students to develop their CT skills. When designing programming activities, we based on some coding and computer science knowledge on Code.org. The Code.org platform organizes programming activities in different forms, including both plugged-in and unplugged programming activities. Because of the characteristics, these activities are good choices to support students in laying a programming foundation. We have selected 8 plugged-in programming activities and 8 unplugged programming activities on the Code.org platform according to the theme orientation, these activities include the corresponding programming concept topics, such as sorting, loops, events, conditions, functions, decomposition, algorithms, and debugging. Because the platform provides a good combination and alternative activities, it can be used to establish a peer-to-peer relationship between plugged-in and unplugged activities (Del Olmo-Muñoz et al., 2020). The arrangement of 16 programming activities in the two phases is shown in Figure 4.

OPEN IN VIEWER

Figure 3.

Puzzle 2 of the Sequence in Labyrinth Activity.

The courses of the four experimental groups and the control group were all completed by the same teacher to eliminate differences caused by different teachers. This teacher is a computer science teacher in the sample school who has 6 years of teaching experience. He is mainly responsible for teaching computer science courses for 7th-grade students, so he understands the basic learning situation of the sample students. And he has programming experience, and once served as a tutor to guide students to participate in programming competitions. We have prepared a teaching guide for each activity, and the teacher only needs to follow the guide to carry out the activity. The teaching guide was mainly composed of activity information (activity name, teaching object, and teaching duration), teaching content analysis, teaching goals, teaching resources, teaching strategy, and teaching process, etc. After determining the content of the basic activities, we used the ‘5E’ teaching model to guide the activities (Bybee, 1997). Specifically, the activities were carried out according to the process of engage-explore-explain-elaborate-evaluate. During this process, the ‘4P’ learning theory runs through the entire programming activity process, starting from the project, starting from passion, peers cooperate, playing with games (Resnick, 2017). Regarding the resources required for programming activities, students need to use computers for plugged-in activities, while some materials (such as cards, brushes, paper cups, etc.) are used for unplugged activities. Figure 5 shows the design of the programming activities teaching mode, and Table 3 shows the description and one activity example. All 16 activity designs can be found in Appendix C (Online only). Students’ performance in some activities can be found in Figure 6.

OPEN IN VIEWER

Figure 4.

Schematic Diagram of the Arrangement of Intervention Activities.

OPEN IN VIEWER

Table 3.

Description and Examples of Programming Activity Teaching Mode.

Content	Information	Description	Example
Activity information	Activity name	Provide the name of the programming activity	Sequence in Labyrinth
	Teaching object	Introduce the sample situation	7th-graders
	Teaching duration	Provide the time required to complete the activity	40 minutes
Teaching content analysis	Activity form	Indicate the type of programming activity used in the activity, plugged-in or unplugged	Plugged-in activity
	Programming concept	Define the programming concepts involved in the activity	In this activity, the study and practice of the concept of ‘sorting’ are mainly carried out
	Computational thinking	Discuss the CT involved in different links in the activity (abstraction, decomposition, algorithmic, evaluation, and generalization)	Abstraction - abstract the maze problem
			Decomposition - disassemble the problem
			Algorithmic - to find and correctly find the answer to the maze exit
			Evaluation-debug and evaluate the code results from time to time
			Generalization- understanding of the sorting concept
Teaching goals	Cognition	Promote students’ understanding of programming concepts	Enable students to learn and master the related content of the sorting during the process of completing the labyrinth activity.Be able to sort the commands as sequential steps in the program and run the program in the correct order.
	Emotion	Improve students’ cooperation and social skills in the process of cooperating to complete tasks	Practice communicating ideas through codes and symbols, and cultivate students' sense of cooperation.
	Action skill	Develop students’ CT skills and the ability to use programming concepts to solve problems	During the learning process, students will be able to develop their CT skills and be able to transfer the skills they have learned to other problem-solving situations.
Teaching resources	Plugged-in	Computer, Code.org platform	Multimedia classroomInternetCode.org platform
	Unplugged	Cards, brushes, paper cups, and other materials
Teaching strategy	‘5E’ teaching model	The ‘5E’ teaching mode is a constructivist teaching model, including 5 teaching links: engage, explore, explain, elaborate, and evaluate.	This activity guide is designed according to the ‘5E’ teaching model, instructing the teacher to carry out activities following the steps of Engage, Explore, Explain, Elaborate, and Evaluate.
	‘4P’ learning theory	The ‘4P’ learning theory is included in the activity, students complete the learning of programming and the improvement of their thinking in the spiral of Project, Passion, Peer, and Play.	Project-start from the maze activity, which is the core activity of programming learning
			Passion- the ‘Angry Birds game role makes students start from enthusiasm and interest
			Peer- students cooperate during the activity, share and construct each other’s works
			Play- programming teaching activities through interesting gamification teaching methods
Teaching process	Engage (5 min)	The teacher provides meaningful learning activities to attract students’ interest in learning.	With ‘Angry Birds’ as the starting point to attract students’ attention and stimulate interest in learning.
	Explore (20 min)	In the inquiry link, students need to conduct inquiry activities for specific content.	Students should complete course 2 (https://studio.code.org/s/coursee-2017/stage/2/puzzle/1) this series of programming activities with the ‘Angry Birds’ game role. They should develop sorting algorithms to transform the bird’s move from one side of the maze to the pig’s position on the other side (As shown in Figure 3).
	Explain (10 min)	After the investigation is completed, students should use their own words to explain the results. Then, the teacher gives a scientific explanation, terminology, or concept.	After the students complete the task, the teacher explains the concept of ‘sorting’. Students can experience the connotation of sorting by combining the programming practice in the inquiry activity and the teacher’s explanation.
	Elaborate (3 min)	The teacher guides students to think and discuss to strengthen their understanding of new concepts.	The teacher asks new questions, for example, what are the steps of cooking? Make students associate the ‘sorting’ with the reality of life.
	Evaluate (2min)	The teacher needs to observe how students apply new concepts and methods to solve problems.	Regarding the questions raised, the teacher evaluates the students by organizing group discussions and recording.

OPEN IN VIEWER

Figure 5.

Schematic Diagram of Teaching Mode.

OPEN IN VIEWER

Figure 6.

Student’s Performance in the Activities.

OPEN IN VIEWER

Figure 7.

Difference Analysis Results of Four CT Tests in Each Group.

OPEN IN VIEWER

Figure 8.

The Growth Model of Students’ CT Skills in Each Group.

Results

The data of this study come from the results of four tests of students’ CT skills, as well as interviews with the teacher and questionnaire surveys of students. Therefore, the study results are also carried out from two aspects: the quantitative analysis of CT skills and the description of qualitative results.

The Quantitative Analysis Results of CT Skills

Considering the research goals, we grouped and displayed the results obtained in each test (Pre-, Mid-, Post, and Postpone-test) of different groups. First, the results related to CT skills are introduced. Then, consider the interaction effects of student’s programming experience and gender in the development of CT skills. Regarding CT skills, Table 4 shows the average scores of four CT tests for different groups. The standard deviation of each measurement was also included in parentheses. It can be seen that, compared with the control group, students’ CT skills in the four experimental groups showed an increasing trend throughout the intervention process. And it can be initially observed that after the second teaching phase of the unplugged-first group, obvious changes in the development of students’ CT skills have occurred, which is a greater improvement compared to the first phase.

OPEN IN VIEWER

Table 4.

Computational Thinking Tests Average Results.

Group	N	Pre-test	Mid-test	Post-test	Postpone-test
Plugged-in Group	33	58.48 (8.70)	73.33 (13.84)	84.24 (17.15)	89.09 (15.69)
Unplugged Group	33	62.12 (17.10)	73.64 (19.81)	81.82 (17.58)	85.15 (14.17)
Unplugged-first Group	29	57.59 (17.46)	68.28 (16.49)	89.31 (12.23)	95.52 (11.83)
Plugged-in-first Group	32	61.25 (13.85)	73.75 (19.80)	83.13 (22.21)	86.88 (16.93)
Control Group	31	68.71 (20.61)	72.26 (22.76)	75.48 (21.73)	78.06 (15.58)
Total	158	61.65 (16.23)	72.34 (18.65)	82.71 (18.87)	86.84 (15.81)

The paired samples t-test was used to analyze the results of four CT tests in different groups to more accurately assess the impact of different programming teaching approaches on students’ CT skills. In addition, to find out which programming approaches can improve students’ CT skills to a greater extent, we used the one-way ANOVA to compare the average differences of CT post-test and pre-test to observe the effects of different programming approaches. The results are shown in Table 5. And Figure 7 visually presents the results of the intra-group difference analysis of the four CT skills in each group.

OPEN IN VIEWER

Table 5.

The Results of Paired Samples t-Test and One-Way ANOVA of CT Skills in Each Group.

Group	N		Diff mid-pre	Diff post-mid	Diff post-pre	Diff postpone-post
Plugged-in Group	33	M (SD)	14.85 (9.06)	10.91 (7.65)	25.76 (12.99)	4.85 (8.34)
		t	9.419***	8.192***	11.384***	3.34*
		p	.000	.000	.000	.020
Unplugged Group	33	M (SD)	11.52 (11.21)	8.18 (12.11)	19.70 (14.68)	3.33 (11.09)
		t	5.899***	3.882***	7.707***	2.727*
		p	.000	.000	.000	.049
Unplugged-first Group	29	M (SD)	10.69 (12.80)	21.03 (11.76)	31.72 (15.37)	6.21 (10.83)
		t	4.498***	9.636***	11.116***	3.087**
		p	.000	.000	.000	.005
Plugged-in-first Group	32	M (SD)	12.50 (13.68)	10.38 (11.72)	21.08 (17.51)	4.11 (9.91)
		t	5.170***	11.129***	15.133***	5.217*
		p	.000	.000	.000	.023
Control Group	31	M (SD)	3.55 (15.61)	3.23 (10.13)	6.77 (17.96)	2.58 (10.32)
		t	1.266	1.773	2.100	1.393
		p	.215	.086	.054	.174
One-way ANOVA		F	3.581**	11.777***	10.593***	0.610
		p	.008	.000	.000	.656

Note. ***=p < .001, **=p < .01, *=p < .05.

Specifically, the average difference between the plugged-in group CT pre- and mid-test (M = 14.85, SD = 9.06, t = 9.419, p < .001), and the average difference between pre- and post-test (M = 25.76, SD = 12.99, t = 11.384, p < .001) reached a significant level, indicating that the CT skills of students participating in plugged-in programming activities has been greatly improved throughout the research process. The average difference between CT pre- and mid-test in the unplugged group (M = 11.52, SD = 11.21, t = 5.899, p < .001), and the average difference between pre- and post-test (M = 19.70, SD = 14.68, t = 7.707, p < .001) also reached a significant level, indicating that unplugged programming activities can also effectively improve students’ CT skills. When combining the two single programming approaches, we considered the influence of the teaching sequence on the teaching effeteness. In the unplugged-first group, we considered performing unplugged activities first and then plugged-in activities. There was a significant difference between CT pre- and mid-test (M = 10.69, SD = 12.80, t = 4.498, p < .001) indicated that the 4 unplugged activities in the first phase improved students’ CT skills. The significant difference between mid- and post-test (M = 21.03, SD = 11.76, t = 9.636, p < .001) indicated that the 4 plugged-in activities in the second phase also improved their CT skills. Summarizing the two phases, there was a significant difference between the students’ CT pre- and post-test (M = 31.72, SD = 15.37, t = 11.116, p < .001), indicating that the combination of unplugged-first can effectively improve students’ CT skills. In the plugged-in-first group, we considered performing plugged-in activities first and then unplugged activities. There was a significant difference between CT pre-and mid-test (M = 12.50, SD = 13.68, t = 5.170, p < .001) indicated that the 4 plugged-in activities in the first phase improved the students’ CT skills. The significant difference between CT mid- and post-test (M = 10.38, SD = 11.72, t = 11.129, p < .001) showed that the 4 unplugged activities in the second phase also improved their CT skills. In summary, there was a significant difference between the students’ CT pre- and post-test (M = 21.08, SD = 17.51, t = 15.133, p < .001), indicating that plugged-in-first programming approach can effectively improve CT skills. For the control group that did not participate in any programming activities, there was no significant difference between the students’ CT pre-, mid-, and post-test.

Two months after the end of the intervention, the students’ CT skills were postponed, and the difference between the CT post- and postpone-test was calculated to observe the retention of students’ CT skills. The results of the one-way ANOVA from Table 5 showed that there was no significant difference in the difference between the delayed test and the CT post-test of different groups (F = 0.610, p < .05). Among them, students’ CT skills in the four experimental groups were significantly retained after the intervention, the unplugged-first group have more prominent results (M = 6.21, SD = 10.83), followed by the plugged-in group (M = 4.85, SD = 8.34), the plugged-in-first group (M = 4.11, SD = 9.91), and the unplugged group (M = 3.33, SD = 11.09).

As shown in Table 5, the one-way ANOVA results showed that there were significant differences between different programming approaches (F = 10.593, p < .001), and the unplugged-first group improved students’ CT skills the most (M = 31.72, SD = 15.37), followed by the plugged-in group (M = 25.76, SD = 12.99). We further increased the model to compare the differences in CT training between different groups. It can be seen from Figure 8 that compared with the control group, the two-stage programming activities of the four experimental groups have significantly improved the students’ CT skills, which once again shows that the four programming approaches can effectively train students’ CT skills. When comparing the experimental group using single programming approaches with the experimental group using combined programming approaches, it can be seen from Figure 8 that the unplugged-first group is the most effective for cultivating students’ CT skills. Therefore, it can be said that the teaching form based on unplugged activities followed by plugged-in activities was the most beneficial for junior high schoolers in terms of developing CT skills.

When considering the role of students’ gender and programming experience in the process of different programming approaches, we first adopt one-way ANOVA to analyze whether students’ gender and programming experience affect the improvement of their CT skills according to the group as the benchmark. The results are shown in Table 6.

OPEN IN VIEWER

Table 6.

Analysis of Differences of CT Skills in Programming Experience and Gender.

Groups			N	Pre-test			Diff Post-Pre
				M	SD	F	M	SD	F
Plugged-in Group	Programming Experience	Have	10	65.00	5.27	10.401**	39.00	5.68	26.989***
		No	23	55.65	8.44		20.00	10.87
	Gender	Male	17	59.24	9.51	2.628*	27.06	15.72	0.344
		Female	16	57.75	8.06		24.38	9.64
Unplugged Group	Programming Experience	Have	12	74.17	9.00	12.821**	15.83	9.00	1.319
		No	21	55.24	16.92		21.90	16.92
	Gender	Male	17	64.38	16.96	2.532*	18.24	15.90	0.340
		Female	16	60.00	17.50		21.25	13.60
Unplugged-first Group	Programming Experience	Have	8	67.50	10.35	3.938*	31.25	9.91	0.010
		No	21	53.81	18.30		31.90	19.21
	Gender	Male	13	61.25	17.51	2.607*	35.38	16.64	1.353
		Female	16	53.08	17.08		28.75	14.08
Plugged-in-first Group	Programming Experience	Have	9	66.67	7.07	1.974	41.11	6.01	31.208***
		No	23	41.11	14.35		14.35	13.76
	Gender	Male	17	62.76	12.37	2.049*	24.71	17.36	0.991
		Female	15	59.67	15.80		18.67	16.85
Control Group	Programming Experience	Have	13	79.23	12.56	6.998*	11.54	18.19	1.607
		No	18	61.11	21.20		3.33	17.49
	Gender	Male	15	72.00	13.73	2.734*	8.67	18.07	2.392*
		Female	16	65.63	15.55		5.00	18.26
Total	Programming Experience	Have	52	71.35	10.67	33.388***	25.96	16.60	6.238**
		No	106	56.89	16.41		18.68	17.52
	Gender	Male	79	67.15	15.02	1.153*	22.53	18.37	0.249
		Female	79	56.14	17.44		19.62	16.59

Note. ***=p < .001, **=p < .01, *=p < .05.

Before the intervention, there was a significant gender difference in students’ CT skills (F = 1.153, p < .05), and boys had a higher CT skill (M = 67.15, SD = 15.02) than girls (M = 56.14, SD = 17.44). But in the four experimental groups, this gender difference was not reflected in the students’ CT skills in programming activities. Students of different genders did not show significant gender differences in the CT improvement in different programming approaches. Whether in the plugged-in group (F = 0.344, p > .05), the unplugged group (F = 0.340, p > .05), the unplugged-first group (F = 1.353, p > .05), and the plugged-in-first group (F = 0.991, p > .05), the improvement of CT in the process of participation by students of different genders was homogeneous. There are also significant differences in the programming experience of students’ pre-test CT skills (F = 33.388, p < .001), but the differences in programming experience are not consistent in different programming approaches. In the plugged-in group, it can be seen that after 8 plugged-in activities in two-stage, students’ previous programming experience had an impact on their CT development (F = 26.989, p < .001). And the combined programming approach of plugged-in-first did not weaken the influence of students’ previous programming experience on their CT development (F = 31.208, p < .001). However, in the unplugged group, after 8 unplugged activities, the CT development of students with programming experience was not significantly higher than that of students without programming experience (F = 1.319, p > .05). Similarly, students’ previous programming experience did not play a role in the unplugged-first group (F = 0.010, p > .05), that is, regardless of whether students have programming experience, they can get the same opportunity to develop CT skills in the unplugged-first programming activities.

In addition, to further analyze the interaction between gender and programming experience on students’ CT development, we established models in four experimental groups to explore the impact of the interaction between gender and programming experience on the growth of students’ CT skills. It can be seen from Table 7 that gender did not have a significant impact on the development of students’ CT skills (F = 2.270, p < .05), but programming experience affects CT skills development (F = 6.006, p > .05). The interaction term between programming experience and gender also did not significantly affect the growth of students’ CT skills (F = 2.736, p > .05).

OPEN IN VIEWER

Table 7.

The Interaction of Programming Experience and Gender on CT Development.

	Mean square	F	p
Revised model	982.179	3.349*	.021
Intercept	68883.713	234.845**	.000
Programming experience	1761.593	6.006*	.015
Gender	665.936	2.270	.134
Gender*Programming experience	802.502	2.736	.100
Error	293.315
Revised total Sum of squares	48117.089
R2	0.061

Note. ***=p < .001, **=p < .01, *=p < .05.

Figure 9 shows the interaction effect between programming experience and gender, and we have superimposed a reference line (black solid line) to represent the average CT growth of the students in the control group. It can be seen that students’ CT development in the four experimental groups has been better than in the control group. On average, the growth of CT skills of students with programming experience was higher than that of students without programming experience, while the CT skills development of boys was higher than that of girls. Specifically, in the group with programming experience, the development of boys’ CT skills was higher than that of girls in the programming activities, while in the group without programming experience, the growth of boys was also slightly higher than that of girls, but compared to those with previous programming experience, this gap has narrowed significantly. It can be inferred that in this study, the designed programming activities better balance the gender difference of students who do not have previous programming experience.

The Qualitative Analysis Results of Interview and Surveys

Teacher

The results of teacher interviews explain the teacher’s intuitive experience in teaching. Two types of code lists are formed around the teacher’s key ideas: 1) The student’s performance observed in different programming approaches group; 2) Differences in students’ performance of different genders.

Regarding student performance in class, the teacher said that compared with unplugged activities, students prefer to participate in plugged-in activities that are easy to operate on the computers and are more interesting and intuitive. For junior high schoolers, unplugged activities may be a bit similar to the form of classroom exercises, lacking some richness and interactivity, so long-term participation in unplugged activities may reduce the enthusiasm of students. As the teacher said,

Students are more efficient in completing plugged-in activities. When a student completes it, they will be very excited to say ‘Teacher, I win, I win’, and the expression on their face will be very excited. In unplugged activities, the expressions of the students will be relatively calm.

For example, in the unplugged-first group, in the first phase of unplugged programming, the overall classroom atmosphere is relatively quiet, and students almost complete tasks individually; while in the second phase of plugged-in programming, students will have an emotional turn. If they encounter a problem, they will immediately say to the classmates next to them, ‘How did you do it?’ As if the interaction is a little bit more interactive, the students will have discussions, and the whole class is very active.

The teacher added. But compared with the plugged-in, unplugged activities also have some unique advantages. As the teacher mentioned,

In plugged-in activities, especially in the first session, due to the different abilities of students, some students have difficulties in integrating into the classroom. In the unplugged activity, almost all students can quickly participate in the activity.

When it comes to the differences in the performance of students of different genders in activities, the teacher believes that girls are more immersive in the classroom, while boys are more active in participation and complete tasks slightly faster. But this does not mean that girls’ programming ability is necessarily lower than that of boys. It’s just that girls’ introverted personality at the junior high school is caused by boys’ desire for expression. As the teacher said,

Girls are not inferior to boys in programming performance, and sometimes do better than boys, and have their ideas.

Students

Around the students’ viewpoints, a code list containing four categories was finally formed: 1) Emotional orientation after participating in programming activities. Students used some sexy vocabulary to describe their feelings after participating in programming activities. About 97% of the students expressed positive emotions after the activity. For example, a girl in the plugged-in group said,

I feel that programming is like a game. I look forward to doing it. I am happy and excited every time I do it.

2) The survey results also showed students’ cognition of CT development after participating in the activity. Specifically, the students discussed whether the programming activities they thought would improve their ability to solve computational problems. 93% of the participants believed that programming activities help improve their problem-solving skills and promote the development of logical thinking skills. As Student E in the unplugged group said,

The learning of programming activities not only improves my problem-solving ability and logical thinking but also improves my spatial imagination.

Similarly, Student K said, “I have made some progress in solving problems, and I now feel a lot easier to solve math problems than before”. It can be seen that after students have participated in activities, students have begun to transfer CT skills to actual problem-solving situations. Student M in the unplugged-first group also said, “I have made some progress. I can solve the problems in life in the simplest way, and my logical thinking has been greatly improved.”

3) When it comes to helping for future programming learning, almost all students said that this programming activity will help to study future programming. The students said that this programming activity can enable them to master some basic concepts of programming, which, as basic knowledge, will be of great help to learning programming in the future.

“I have mastered and understood how to write code, how to arrange it, what conditions are needed, insert sound effects, and understand these contents will help future programming learning,” said Student D.

When discussing the role of combined programming approaches, the students in the unplugged-first group all indicated that the unplugged activities in the first phase are beneficial to the plugged-in activities in the second phase. “I think those are helpful. The first-phase unplugged activities can prepare me easier to complete the second-phase plugged-in activities,” said Student N who has never participated in programming activities.

In the plugged-in-first group, about 63% of the students thought that the first phase of plugged-in activities was more difficult than the second phase of unplugged activities. As student T said, “It was a bit difficult when they first started learning, but after more practice, it will be easier”; likewise, student W said, “Sometimes it is difficult in the first stage. It takes a long time to get it, and the teacher’s help is needed”. Both of these students were exposed to programming for the first time. It can be seen that for some students without programming experience, unplugged is a good way to transition to plugged-in programming.

4) Evaluation of the design and difficulty of programming activities. When discussing the difficulties faced by programming activities, most students (76%) said that the difficulty of the activities is moderate and it is relatively easy to operate. But it will find that 24% of students still respond to a certain degree of difficulty, and this part of the students do not have programming experience, and 71% of them are girls. As Student R said, “In the process of completing the task, it will be slightly difficult for me,” “but the activity is still very interesting” he added. And some students showed changes in their feelings of difficulty in the process, “It was a little difficult at the beginning, but it was easy to operate in the middle, and finally found it easy,” Student M explained.

5) The idea of engaging in programming-related work in the future. When discussing students’ ideas for future programming-related jobs, the majority of students (59%) indicated that they are willing to engage in programming-related jobs in the future. As Student H (unplugged group) explained, “Because artificial intelligence is now hot and is widely used in industrial and military life, covering all walks of life, most of the repetitive tasks can use programming, so his development space is huge. I want to try it.” Similarly, Student X added, “I think programming is very interesting and helpful, and computers will be very common in the future.”

Discussion

In this work, we tracked the changes in junior high school students’ CT skills in different programming approaches, as well as the influence of students’ gender and programming experience on their CT development. To determine the goal of this study, the following questions were put forward: the effects of different programming approaches on promoting 7th-graders’ CT skills, as well as which programming approaches have stronger effectiveness, and the students’ previous programming experience and gender in the process and the differences in the CT retention among students participating in different programming activities approaches. To this end, a quasi-experimental study was conducted on 158 7th-graders, which included four experimental groups and one control group. To achieve this comparison of programming approaches, instructions were designed for the four experimental groups, divided into two phases. The plugged-in group received plugged-in programming activity intervention in two phases, while the unplugged group received unplugged programming activity intervention. These two groups respectively represent two single programming approaches. The unplugged-first group received unplugged programming activity intervention in the first phase, and plugged-in programming activity intervention in the second phase, while the plugged-in-first group received plugged-in programming activity intervention in the first phase, and received unplugged programming activity intervention in the second phase. These two groups represent two combined programming approaches due to the teaching sequence. The data came from the results of four CT tests at different stages, interviews with the teacher, and questionnaires from students.

In response to the RQ1, the quantitative results showed that the initial level of the sample students’ CT skills was relatively low, which is expected, although, under the guidance of national policies, the introduction of CT into the classroom of basic education is strongly advocated (Ministry of Education of the People’s Republic of China, 2017). However, due to the limitations of hardware equipment and the wave of test-oriented education, there are still many problems in the implementation of the policy, and many students still lack guidance on programming and CT education (Sun et al., 2021a). Just as in this sample group, only 32.91% of students have been exposed to programming, and most students do not have previous programming experience. Therefore, the proposed teaching focuses on CT concepts suitable for the education level of students. Although the initial test scores were low, after two phases of programming intervention, students’ CT skills in the four experimental groups have made significant progress. Just as 93% of participants believe that programming activities help improve their problem-solving skills and promote the development of logical thinking skills. Among them, the plugged-in group using plugged-in programming activities achieved the same degree of improvement in CT mid- and post-test. This result once again proved the effectiveness of plugged-in programming activities based on Code.org in promoting CT skills. And just as the results of the questionnaire survey showed that most students who participated in Code.org plugged-in activities expressed positive emotions, and the programming activities in Code.org were lively and interesting enough to stimulate their interest in learning. Kale and Yuan (2021) analyzed the third-year curriculum plan on Code.org and determined the various components of CT skills, they found that this computer teaching activity significantly improved students’ CT skills and problem-solving abilities. Similarly, Kalelioğlu (2015) checked students’ performance on Code.org and found that students’ reflective thinking skills in problem-solving have been improved. It can be said that Code.org can be used as a tool to promote CT skills, with a relatively complete activity design, which can be easily used by educators.

In addition, the results showed that unplugged programming activities can also improve students’ CT skills. This confirmed the effectiveness of unplugged in junior high school. There was still a lack of relevant research on unplugged programming in junior high school. This result provided evidence for using unplugged programming to cultivate 7th-graders’ CT skills. Brackmann et al. (2017) found that students who participated in unplugged activities significantly improved their CT skills than those who did not participate in programming activities. This will provide methods and ways to carry out CT education in areas that do not have computer hardware conditions. However, in concrete terms, the unplugged group using unplugged programming activities has achieved a great improvement in the CT mid-test, but the magnitude of this improvement has declined in the CT post-test. This may indicate that long-term unplugged activities may not last long for the effectiveness of cultivating students’ CT skills. In other words, students may not be able to achieve continuous improvement in CT skills during long-term participation in unplugged activities. This may indicate that these long-term unplugged activities are too simple, somewhat repetitive, or even slightly tedious for seventh graders. Unplugged programming activities may put more emphasis on the internalization process of a kind of thinking, focusing on the understanding of CT concepts. Compared with the plugged-in programming approach, it lacks a certain image display, and students’ participation time is less active, so the long-term study may make students lose motivation. As the teacher observed in teaching, compared with plugged-in activities, the enthusiasm of students participating in unplugged activities and the communication and cooperation between peers are less. And the quantitative data showed that compared to the plugged-in group, the unplugged group had a slightly inferior impact on the students’ CT skills. This confirmed the finding of Tsaravaet al. (2018) on how the combined plugged-in and unplugged activities allow an integrated constructivist approach to transmitting their respective content. As Del Olmo-Muñoz et al. (2020) pointed out that students’ satisfaction with unplugged activities is lower than plugged-in activities. Therefore, in unplugged activities, context is important. Appropriate teaching methods should be adopted according to the values, experiences, and expectations of learners so that students can participate enthusiastically and sustainably (Bell, 2021).

Of course, this is not to deny the effectiveness of unplugged activities in promoting students’ CT skills, but it should be advocated that appropriate methods should be used to introduce and optimize the design of unplugged activities to maintain students’ interest and enthusiasm for learning. Just as in the unplugged-first group, the form of unplugged-first has been improved to a small extent in the first phase, but it has been greatly improved in the CT post-test. And compared with the other three experimental groups, this programming form has the most prominent impact on promoting students’ CT skills. Combining the results of the analysis of differences in students’ programming experience, it can be found that in the plugged-in group and the plugged-in-first group, students’ previous programming experience affects their CT development. This is consistent with the research findings that Chen et al. (2019) found that previous programming experience is beneficial to students’ attitudes and achievements in computer science at the university level. However, we found that in the unplugged group and the unplugged-first group, this effect was not significant. This may be because both the unplugged group and the unplugged-first group underwent unplugged programming activities in the first phase. It can be inferred that unplugged programming activities can weaken the influence of students’ previous programming experience on their CT development. Therefore, when CT is introduced in the 7th grade, to ensure the fairness of CT education, we advocate that it can be introduced through unplugged programming, but after students have a certain programming foundation, we still believe that plugged-in programming can continue to efficiently cultivate students’ CT skills. These combined programming approaches may be able to achieve the most effective CT education effect. This is consistent with the views of Del Olmo-Muñoz et al. (2020), who believed that in programming activities, it is more appropriate to conduct CT education through a combined method that combines unplugged activities with plugged-in activities rather than just plugged-in activities. Similarly, Saxena et al. (2020) designed a two-stage programming activity (first unplugged and then plugged-in), which effectively improved students’ ability to master pattern recognition and sequencing. The results of the CT postpone-test showed that this form also helped students to internalize CT into their cognitive structure to the greatest extent, to achieve longer-term retention of results. Switching between different programming approaches may enhance the long-term impact of programming (Chen et al., 2019). Just as in the survey with the unplugged-first group, all the students said that the unplugged learning in the first phase is helpful for the plugged-in learning in the second phase and even the programming learning in the future. This may be explained by the fact that the unplugged programming activities in the first phase enable students to be more immersed in tasks in a form that does not rely on computers. This engaging learning strengthens students’ understanding of programming and CT concepts, and then in the second phase, students can strengthen their understanding of these concepts through more lively and interesting plugged-in programming activities to achieve the sublimation of knowledge and skills, this kind of sublimation first lays the foundation and this kind of progress will continue over time.

As far as gender is concerned, the results showed that boys at the initial level before the intervention have a higher level. This finding requires further attention. This gender gap may be explained by the difference in the cognitive development structure of boys and girls. Neuroscience research showed that compared with boys, girls’ hippocampus with memory and language functions can develop faster and larger, which will affect their ordering, vocabulary, reading, and writing skills; on the other hand, boys have more to define the cerebral cortex for spatial relationships, so they learn better through exercise and visual experience (Bonomo, 2010; Sousa & Tomlinson, 2011). Therefore, it is necessary to adopt different teaching methods for different genders students. Gurian (2003) pointed out that girls tend to prefer collaborative activities that share ideas with others, while boys prefer activities that are kinesthetic, spatially oriented, and manipulative. Similarly, Angeli and Valanides (2020) found significant interaction effects between gender and scaffolding strategies, indicating that boys benefit more from individualism, kinesthesia, spatial orientation, and manipulation-based activities, while girls benefit from collaborative writing activities. The results of this study pointed out that an appropriate programming activity plan can provide students with a positive learning experience and weigh the differences between different genders. We can infer that despite possible background differences, students of different genders can develop CT skills under the guidance of appropriate teaching methods. In addition, some studies point out that the gender differences may also be caused by girls’ misunderstandings in the programming field and the lack of confidence and interest (Master et al., 2016; Stout et al., 2011). Beyer (2014) found that students have gender differences in computer efficacy, stereotypes, interests, values, interpersonal orientation, and personality variables, which may explain the underrepresentation of women in computer science. Therefore, it is necessary to use targeted programming activities to motivate girls to enhance their motivation. Mauk et al. (2020) pointed out that developing coding plans for girls (such as Robotics Engineer Barbie™) is effective in enhancing girls’ self-confidence. Similarly, Stoeger et al. (2013) found that online guidance has multiple advantages in promoting girls’ interest in STEM. These targeted activities are expected to arouse the interest of girls and are essential to narrowing the gender gap related to programming-related CT understanding and strengthening women’s status in the computer and programming field. As Mouza et al. (2020) pointed out that effective teaching methods make boys and girls have no statistically significant differences in the learning and use of computer science concepts.

Implications for Policy and Practice

This research enhances our understanding to different approaches of programming activities to promote students’ CT skills. In the past in junior high school, researchers mainly used a single plugged-in programming approach to cultivate students’ CT skills, while less research paid attention to the effectiveness of unplugged programming activities in junior high school classrooms and whether the combination of the two can improve CT training effect. This study proved through a quasi-experimental study that besides plugged-in programming activities, unplugged programming activities can also promote the junior high school students’ CT skills, and the combination of the two approaches, the effeteness was still significant. In particular, the combined approach of unplugged-first was particularly effective in promoting CT skills, and this form also effectively weakens the influence of students’ previous programming experience on their CT development. The results of this study have enriched the methodology for developing programming and CT education in junior high school. Through empirical research, it has provided practical evidence for cultivating students’ CT skills. The results of this study may provide some practical suggestions for educators engaged in CT education in junior high school.

First of all, the introduction of unplugged into the junior high school can be used as an effective means of cultivating students’ CT skills. For some areas lacking computer hardware equipment, using unplugged activities to train students’ CT skills is an effective method. However, unplugged activities that are too simple and the course duration is too long will also reduce students’ interest in programming, because many unplugged activities are more like subject-based ‘questioning’ approaches, and lack a certain degree of practicality. Therefore, in unplugged programming activities, we suggest that teachers should pay attention to the richness and diversity of activity design, combine design with physical objectives (such as robots) to acquire CT skills, and pay attention to mobilizing students’ initiative, enthusiasm, and a good sense of programming learning experience in unplugged activities. Fully use unplugged programming as an effective and easy-to-handle method, and flexibly adopt a ‘random entry’ teaching method to reduce students’ fear of programming and reduce their psychological burden. Secondly, given the reality that most 7th-graders do not have previous programming experience, we advocate that when programming education is introduced in junior high school, the transition can be made through unplugged activities first, and when students have a certain basic programming concept, they can continue through plugged-in activities, to avoid some students’ technical alienation when they are initially exposed to computer programming. Finally, we believe that reasonable activity design can effectively avoid gender differences in programming activities. Therefore, teachers can promote the students’ CT skills of different genders through reasonable design in teaching, to reduce the gender gap in the computer field in the future.

Limitations

This study also has some limitations. First, this study was conducted in a Chinese educational context, and the sample only included 7th-graders. Therefore, the study results need to be treated with caution when extending to the educational contexts and stages of education in other countries. Future research can continue to study different approaches to education. The impact of programming activities on promoting students’ CT skills in different countries and different grades. Second, this study only researched the effects of different programming approaches on students’ CT skills, without considering the measurement of other variables. In future research, we can measure other variables surrounding students in programming activities, such as students’ programming attitude, motivation, and other factors. Third, this study only considers the influence of students’ previous programming experience and gender on their CT development and does not research other influencing factors (such as family background, parent’s occupation, etc.). Future research can further examine the effects of different factors on students’ CT development. Finally, this study only tracks the evolution of students’ CT skills during programming activities but does not examine the acquisition of students’ specific programming concepts. Future research can use work content analysis to check students’ performance and the acquisition of programming concepts.

Conclusion

Since current and future society aspires for people who can face CT skills, it is inevitable to include CT in the K-12 classroom. A quasi-experimental study on programming to promote CT skills involving 158 7th-graders was carried out. Two single programming approaches (plugged-in and unplugged) and two combined programming approaches (plugged-in-first and unplugged-first) were carried out in four experimental groups, while the control group did not participate in any programming activities. Students’ CT skills were tracked in this process. Quantitative results showed that the four programming approaches can significantly improve students’ CT skills, but the effeteness of the combined programming approaches of unplugged-first was particularly effective. Also, the role played by students’ programming experience and gender was also explored. The analysis results, including interviews and questionnaire surveys, supplemented the quantitative results with qualitative data. This study aims to make its contribution to programming and CT education in the field of junior high school.OPEN IN VIEWER

Figure 9.

Interaction Effect Between Programming Experience and Gender in Predicting the CT Development.

Supplemental Material

sj-pdf-1-jec-10.1177_07356331211035182 - Supplemental material for Single or Combined? A Study on Programming to Promote Junior High School Students’ Computational Thinking Skills

Supplemental material, sj-pdf-1-jec-10.1177_07356331211035182 for Single or Combined? A Study on Programming to Promote Junior High School Students’ Computational Thinking Skills by Lihui Sun, Linlin Hu and Danhua Zhou in Journal of Educational Computing Research

Supplemental Material

sj-pdf-2-jec-10.1177_07356331211035182 - Supplemental material for Single or Combined? A Study on Programming to Promote Junior High School Students’ Computational Thinking Skills

Supplemental material, sj-pdf-2-jec-10.1177_07356331211035182 for Single or Combined? A Study on Programming to Promote Junior High School Students’ Computational Thinking Skills by Lihui Sun, Linlin Hu and Danhua Zhou in Journal of Educational Computing Research

Supplemental Material

sj-pdf-3-jec-10.1177_07356331211035182 - Supplemental material for Single or Combined? A Study on Programming to Promote Junior High School Students’ Computational Thinking Skills

Supplemental material, sj-pdf-3-jec-10.1177_07356331211035182 for Single or Combined? A Study on Programming to Promote Junior High School Students’ Computational Thinking Skills by Lihui Sun, Linlin Hu and Danhua Zhou in Journal of Educational Computing Research