The flipped classroom is a learning model which reversed or flipped classroom activities and homework (Jensen et al., 2018). In traditional teaching, instructors teach knowledge in the classroom, and students do their homework after class. In the flipped model, instructors have students learning or doing homework first, followed by discussing during class time and putting some ideas into practice (Kong, 2014; Sever, 2014; Çevikba, and Argün, 2017; Yavuz and Ozdemir, 2019). The flipped classroom provides some advantages. Class time management has become more effective, saving learners time in participating and collaborative activities (Baker, 2000; Cole and Kritzer, 2009; Fulton, 2012; Milman, 2012; Halili and Zainuddin, 2015). The interaction between learners and teachers has improved (Lage et al., 2000; Bergmann and Sams, 2012; Roehl et al., 2013; Arnold-Garza, 2014). Moreover, it helps teachers to immediately monitor learner performance and tutor learners with difficulties (Lage et al., 2000; Fulton, 2012; Millard, 2012).

Zhang et al. (2016) believe that flipped learning is particularly useful for programming courses. Many works proved their effectiveness. Mok's study (Mok, 2014) evaluated the effects of flipped classrooms in programming courses based on the perspectives of students. The study concluded that this method is effective for learning programming, the participation rate of students in the learning process is higher with higher satisfaction of students. Souza and Rodrigues (2015) compared the experimental research of flipped teaching and traditional teaching in C programming courses. According to the research results, compared with traditional teaching methods, the flipped classroom significantly improves the programming self-efficacy and academic success rate of students. Chiang (2017) analyzed the impact of flipped learning on the problem-solving ability of students in Java programming learning. The experiment concluded that the problem-solving strategies of students are more effective than their previous problem-solving attitudes. Özyurt and Özyurt (2018) proposed an Adapted flipped classroom approach (AFCA), which analyzes the effects of the use of AFCA on the programming success, attitudes, and self-efficacy of the students in learning programming. The study results show that employing AFCA to teach programming yields positive effects in terms of the programming success and self-efficacy of students. There was no significant change in their attitude scores following the implementation.

Although flipped classrooms have brought some positive effects on learning, some studies have raised some challenges. For example, Milman (2012) mentioned that if learners play computer games while watching instructional videos, learners will be distracted by them and the learning process will be hindered. In addition, Moffett and Mill (2014) believed that if learners are unwilling to experience online learning, the learning effect will not improve. Moore and Chung (2015) mentioned that every learner has a different learning style, and flipped learning may not be able to respond to the needs and preferences of learners. Jawawi et al. (2015) pointed out that learners may have limited access to the tools or resources needed for online learning, which hinders flipped learning (Kissi et al., 2017).

To ensure that flipped learning or flipped classrooms can achieve better results, some studies have begun to adopt hybrid flipped teaching methods. For example, Miras et al. (2021) designed a hybrid teaching method of flipped classroom and peer instruction in an introductory course of programming. Comparing with those obtained in the previous offerings using traditional teaching, the results showed that the new methodology was used, the dropout rate and fail share decreased significantly and the academic results have improved. This study adopted a hybrid strategy. To solve the inadequate consideration of individual differences in flipped learning, adaptive teaching was combined in the classroom to reinforce it.

Adaptive instruction means creating a learning environment and finding instructional approaches and techniques that conform to meet individual needs of students (Inan and Grant, 2008). It can be adaptive teaching or adaptive learning (Matei and Gogu, 2017). In the teaching process, in accordance with the abilities, status, interests, and needs of learners, appropriate responses and adjustments are made to improve the learning effect and achieve the expected teaching goals. Parsons et al. (2018) identified that early literature focused on the decision-making of teachers. Between 1995 and 2004, literature often referred to scaffolding and teacher reflection. Until 2008, adaptive teaching started being mentioned in literature. There are some major issues usually discussed in terms of adaptive teaching, such as how to evaluate adaptive teaching, how to judge the degree of success of adaptive teaching for individual students, what are the relations between adaptive teaching and professional competence of teachers, and which methodological approaches are useful for investigating outcomes of adaptive teaching (Hardy et al., 2019). In addition, smart devices and intelligent technologies were applied to effectively promote the development of personalized learning and adaptive learning. Peng et al. (2019) introduced a teaching method enabled by a smart learning environment and deeply analyzed personalized learning and adaptive learning. Four aspects, which are learner profiles, competency-based progression, personal learning, and flexible learning environments, constructed personalized adaptive learning. Peng et al. explored a form of learning profiles model and a generative path recommendation pattern of personal learning.

There is almost no literature discussing the implementation of adaptive teaching in the classroom of programming teaching. It may be because it is difficult to implement adaptive teaching alone in a limited time class. When discussing “adaptive” and “computer programming learning” in the literature, most of them talk about an adaptive web-based system or an adaptive e-learning platform for learning programming (Chatzopoulou and Economides, 2010; Troussas et al., 2021). Although we can find relevant papers by searching with “flipped learning” and “adaptability” as keywords, they also talk about the use of adaptive platforms to assist students in studying off class (Fang et al., 2019; Kaw et al., 2019; Clark and Kaw, 2020; Louhab et al., 2020; Mojtahedi et al., 2020; Hsieh et al., 2021; Liu et al., 2021).

Learning effectiveness is an index used to judge the learning achievement of learners. After learners participate in learning, their learning status can be measured by evaluating changes or differences in the performance of these indicators. Based on the results of indicators, learners could improve learning methods and teachers could improve teaching methods (Guay et al., 2008). Kirkpatrick and Kirkpatrick (2006) proposed a four-level training evaluations model, which includes four items: reaction, learning, behavior, and results. The reaction aspect refers to the degree to which learners like the course. The learning aspect refers to whether learners gain knowledge accumulation or mental growth. The behavior aspect refers to whether learners have learned to change their behavior. Result aspect refers to the degree to which learners can apply the content of learning. The learning effects discussed in this research include the perceptions of students about the class (reaction aspect), academic performance (learning aspect), and learning motivation (learning aspect).

Students' perception of the course is related to their satisfaction with the course. Student satisfaction can be defined as the attitude of students on the course learning experience, teacher's teaching quality, and teaching materials (Gao et al., 2021). Many pieces of literature use the opinions of students as the basis for measuring the effectiveness of the course implementation (Mok, 2014; Choe et al., 2019; Gao et al., 2021). Academic performance is the most direct result of what students learn. Additionally, it is one of the most used indicators to measure learning outcomes (Hwang et al., 2013; Cheng et al., 2019; Choe et al., 2019; Miras et al., 2021). The success of learning depends on the enthusiasm of the learner and motivation drives learners to achieve learning goals. Therefore, the learning motivation of learners is probably the important key to affecting learning effectiveness (Filgona et al., 2020). Hadre et al. (2007) believe that motivation is one of the most powerful factors that determine the success or failure of students in school. Stimulating the enthusiasm of students for learning in school and motivating students to succeed in school is an important issue, and it is also one of the biggest challenges facing education (Filgona et al., 2020). There are many studies in the literature on how to enhance the learning motivation of students (Jensen et al., 2018; Awidi and Paynter, 2019; Fang et al., 2019; Hardy et al., 2019; Ling et al., 2021; Miras et al., 2021), which shows the degree of importance it is paid to.

Our flipped classroom study involved 52 sophomores of the Information Management Department in a Java programming course. Participating students have taken a programming course (C language) in the freshman year, so they have the basic concepts of programming. To understand the impact of the design of this experimental course on the learning performance of students, the grades of the previous programming course were used as the pre-test data and the semester grades of this experimental course were used as the post-test data of “learning performance.”

The first step of the teaching process was drawing up the topics and goals of the weekly courses. The online teaching material would be uploaded for students to study in advance. The beginning of the 3-h classes is a simple test for the preview result of students. The main purpose is to urge students to read the teaching material before the classes. After the test, the teacher will reinforce the important content. Some classroom activities will be arranged afterward, which were different before and after the midterm exam. Before the midterm exams, students mainly strengthen their programming foundation and problem-solving skills. Therefore, the activities were mainly homework exercises or small competitions. After the midterm exam, the students were divided into two groups based on their midterm exam results. Students with a score of 70 or more are classified into the “Achievement challenge group,” and the others are classified into the “ability improvement group.” The learning content before the midterm exam focuses on understanding Java and rarely involves more complicated logic problems. Based on previous teaching experience, students with mid-term grades above 70 will be able to deal with complex problems. Students with a mid-term score of 70 or less still need to strengthen their basic programming skills to pass the exam at the end of the term. Therefore, 70 points are used as the grouping limit. Then, the course adopted the concept of adaptive teaching and performed different classroom activities. Students in the “Achievement challenge group” challenged more difficult learning tasks in groups. At the same time, students in the “Ability improvement group” practiced more basic topics and were guided by teachers. Before the end of the course, the “Achievement challenge group” shared the results of the challenge and the difficulties encountered in the process. After discussing with the students, the teacher made a review and summary of this week's course and previewed the course of next week.

The research tools in this study included a pair of pre- and post-grades, a pair of pre- and post-test questionnaires for measuring the learning motivation of students, and a post-test questionnaire for perceptions of students on the course.

The data of pre-test of learning performance of students were collected from the semester grade of programming course (C language) they had taken in the freshman year. The semester grade of this experimental course was the post-test data of learning performance. As the two control variables (course time, teaching materials, measuring methods, and teachers) were not the same, in addition to the original grade for verification, the pre- and post-test grades were normalized by the Min-Max. Min-max normalization is one of the most common ways to normalize data. The minimum value and the maximum value of grades get transformed into a 0 and a 1, respectively. Therefore, every grade would be transformed into a decimal between 0 and 1. To make it easier to compare with the original grade, the normalized value was multiplied by 100. The formula is (x − min)/(max − min) ^* 100. Then, a paired t-test was performed to detect the impact of this course on the academic performance of students.

The questionnaire of learning motivation was from the paper published by Hwang et al. (2013), which was modified according to the measurement method developed by Hwang and Chang (2011). There was a total of seven questions with a six-point rating scheme (6 = “strongly agree” and 1 = “strongly disagree”). The items in the questionnaire are listed below. The content of this questionnaire contains values (e.g., item 1, 3, and 7), expectations (e.g., item 2 and, 6), and emotions (e.g., item 4 and 5) (Eccles, 1983) to evaluate the mental thinking of students about learning Java. Cronbach's alpha value was 0.79. At the beginning of the course, the students took the pre-test of learning motivation questionnaire. At the end of the course, the students took the post-test of the learning motivation questionnaire. Compare the differences to understand the impact of curriculum activities on the learning motivation of students.

To understand the views of students on the design of this course, students took a questionnaire at the end of the course. We used the “questionnaire of students' perceptions of the online community-based flipped classroom” introduced by Lin and Hwang (2018). The questionnaire evaluates the students' perceptions of flipped learning by a five-point rating scheme (5 = “strongly agree” and 1 = “strongly disagree”) with 14 items. To be consistent with the rating scheme of the learning motivation questionnaire, a six-point rating scheme was used during the actual test. These 14 items were divided into four dimensions. Items 1–5 are for content dimension, which test whether the student thinks the content of the course is helpful. Items 6–8 are for communication dimension, which test whether the student thinks that the curriculum design is helpful for communication with teachers and peers. Items 9–11 are for performance dimension, which measure whether the student thinks that the course contributes to their personal performance. Items 12–14 are for interest dimension, which measure whether the student thinks the class is enjoyable. Cronbach's α values of the individual dimensions were 0.85, 0.79, 0.78, and 0.87, respectively, showing acceptable reliability in internal consistency. The items of each dimension are as follows:

In this study, the semester grade of the previous programming course taken by the students was the pre-test value of the “learning performance,” the semester grade of the experimental course was the post-test value. The original grades were normalized by the Min–Max. Both grades performed the paired sample t-test.

As shown in Table 1, the mean values and SDs of the original grades were 72.60 and 10.22 for the pre-test, and 77.93 and 13.93 for the post-test. The t-test result (t = 3.68, p < 0.05) shows that there was a significant difference between the two tests. The result t-test of normalized grade (t = 4.53, p < 0.05) yielded the same result. The average score increased from 50.27 to 64.69, and the SD decreased from 29.19 to 26.48, indicating that students generally performed better after the course, and the distribution of grades was more concentrated.

After the midterm exam, the participating students were divided into two groups according to their grades in the midterm exam. There are 30 students in the “Achievement challenge” group with the grade higher than or equal to 70. There are 22 students in the “Abilities Improvement” group with the grade lower than 70. After the midterm exam, adaptive instruction was performed. To better understand the two groups of the learning performances of students, the two groups, respectively, performed t-tests of their semester grades. The results are shown in Table 1.

In the “Achievement challenge” group, the means and SDs of the original grade were 77.17 and 9.71 for the pre-test, and 87.68 and 7.56 for the post-test. The means and SDs of the normalized grade were 63.66 and 27.74 for the pre-test, and 83.24 and 14.38 for the post-test. Both t-test results of before standardization (t = 5.83, p < 0.05) or after standardization (t = 4.17, p < 0.05) showed that students have made significant progress in academic performance after the course. In the “Ability improvement” group, the means and SDs of the original grade were 66.36 and 27.74 for the pre-test, and 64.62 and 8.40 for the post-test. Although the average grade of the original has dropped, it has not reached a significant level (t = −1.29, p > 0.05). However, the t-test result of normalized grade (t = 2.13, p < 0.05) showed significantly better learning performance after the course.

The above analysis shows that the design of this course in terms of adaptive teaching has a considerable degree of help for students with different levels. It has a significant improvement of the “achievement challenge” group. However, for the “capacity improvement” group, there was a relatively limited degree of improvement.

Table 2 shows the t-test results of the pre- and post-test ratings of the learning motivation questionnaire for all students and the two groups. The means of the questionnaire ratings have increased regardless of whether statistics were conducted in all or groups, which show that the learning motivation of students has improved. The t-test results showed significant difference between pre- and post-test ratings for all students and both groups, where t = 3.75 (p < 0.05) and t = 3.70 (p < 0.05), and t = 1.74 (p < 0.05), respectively. The results showed that the introduction of flipped learning and adaptive teaching in the course has a positive and significant effect on the improvement of the learning motivation of students.

As it was a new experience for the students to learn under the flipped learning combined with the adaptive instruction, it is interesting to know the views of the students about the class. There are 14 items in the questionnaire of perceptions of students about the course, which are divided into four dimensions: content, communication, performances, and interest. This study adopted the six-point rating scheme. The questionnaire was administered at the end of the course. The average rating of four dimensions of perceptions of students about the course is shown in Figure 1. Overall, students' perception of curriculum design is positive. Among the four dimensions, the average scores are 4.75, 4.78, 4.72, and 4.93, respectively. Students were most satisfied with the “interest” aspects of the course. The average ratings for each aspect of the achievement challenge group are all greater than the ability improvement group. The dimension with the least difference is content, and the dimension with the greatest difference is performance. This result is reasonable because the teaching materials provided in the course were the same. Meanwhile, the achievement challenge group has more discussion and problem-solving training.