1.1 Research background

Programming courses are important for learners who are information technology majors, but learning programming is not an easy task for either learners or instructors (Kordaki, 2010; Ozmen et al., 2014; Tan, Ting, & Ling, 2009). Learners have to understand the semantics of programming, and analyze and modify the structure debugging to acquire good programming skills. There are many factors that affect undergraduate learners' programming learning including the difficulty of understanding programming structures, method finding, coding and applying the programming concepts to solve problems (Miliszewska & Tan, 2007; Özmen & Altun, 2014). Besides, it was found that learners are often passive in terms of finding relevant and specific learning information during learning, and thus, their lack of practice, effort, motivation (Hawi, 2010; Özmen & Altun, 2014), and problem-solving skills (Derus & Ali, 2012) make them tend to follow instructors' instruction directly without digesting the content, and cause them to fail to accomplish the programming tasks. Research has revealed that learners who adopt appropriate programming learning strategies and methods would achieve significantly better learning achievements than those who learned with the conventional programming approach (Hsu & Hwang, 2017). Within the strategies for programming, it has been noticed that the process of debugging is a good way to promote learners' programming ability since programming problems can arise for various reasons, and require learners' problem-solving ability to test and debug the resulting codes (Brooks, 1999; Bergersen & Gustafsson, 2011). Thus, the processes of programming debugging create opportunities for learners to think, explore, and improve their programming ability (Lee & Wu, 1999).

Related studies have found that integrating technology into teaching and learning can have a positive influence on promoting learners' motivation, attitudes, and achievement (Bennett, 2001; Yang, Tzuo, Higgins, & Tan, 2012), and well-facilitated e-learning supports could enhance learners' motivation and efficacy in programming learning (Law, Lee, & Yu, 2010). The Internet-based environment provides learners with an important and open resource for programming learning, so that it need not be restricted to school (Gomes, Areias, Henriques, & Mendes, 2008). Of the various technology-supported learning methods, WebQuest is a student-centred learning method which encourages learners to learn from web-based environments, and instructors provide learners with a list of related information to prevent them from accessing inappropriate sites. Constructivism is the principal theory behind the WebQuest strategy, and the adoption of the WebQuest method provides learners with a good structure as learning scaffolding while they are programming by giving them opportunities to apply their past and present learning experiences, and to organize information for solving their learning problems (Seger & Verhoeven, 2009; Davidson-Shivers & Rasmussen, 2006). In the WebQuest processes, learners undergo a conceptual shift through developing knowledge from information (Aoki, 2004; Young & Wilson, 2002), They also change their role, becoming active learners during the learning activities and thinking reflectively with the provided and organized web-based information throughout the learning process (Kundu & Bain, 2006). The learning processes that learners experience while using WebQuest foster their abilities of analyzing and integrating information (Dodge, 2001), and are also effective for learners' logical thinking ability (Cigrik & Ergül, 2010).

1.2 Research purpose and questions

Programming courses are perceived as being difficult by learners, and the lack of effort and knowledge are factors that cause failure in programming learning (Hawi, 2010). Considering the above-mentioned critical programming learning issues, it is important to provide learners with well-designed and adequate programming learning instruction and debugging training to promote their positive experience of programming learning (Lee & Wu, 1999), while it is not an easy task for novice programmers because it require experience to be aware of the errors, and the extra support and assistance would be needed to help novices improve their program comprehension abilities and accumulating debugging experiences (Lee & Wu, 1999). Researchers have noticed that the WebQuest activities foster learners' self-regulated learning (Seger & Verhoeven, 2009; Davidson-Shivers & Rasmussen, 2006), and the web-based learning environment provides learners with rich information and opportunities for gaining knowledge. Moreover, instructors' filtered and organized web-based learning resources help learners to pay more attention to the targeted learning problems instead of wasting time collecting related information and checking the reliability and relevance of the large amounts of information they have found (Kuiper, Volman, & Terwel, 2005).

Hence, in this study, the researcher aimed to design the modified WebQuest activity with debugging practice for college learners' programming learning and aimed to promote learners' programming motivation as well as their problem-solving skills by having them gather, summarize, synthesize, and evaluate the information to complete the tasks set by the instructors in the WebQuest activity. The research questions of the study are: (a) Would the WebQuest activity promote learners' programming learning performance? (b) What are learners' perceptions of learning programming with the WebQuest activity?

This paper could serve as a reference for programming teachers who wish to integrate the WebQuest strategy to encourage learners to engage in programming problem-solving and to provide useful insights and implications for learning technology researchers as a guide for designing WebQuest learning in a variety of higher education settings.

2.1 Programming learning

Programming learning requires learners' cognitive processes for understanding programming grammar, syntax, and problem-solving skills during the writing processes. Beginners need to build their programing skills by entering the code, building software and debugging and correcting their programs. Researchers have revealed that programing learning combines not only cognitive and mental processes, but also requires a process of problem-solving (Mow, 2008) that involves a series of higher order thinking skills such as association, comprehension, analysis, and generalization (Garofalo & Lester, 1985). It has been noticed that without proper guidance, learners frequently select too simple or too complicated problems and, as a result, either become bored or lose enthusiasm or interest in programming learning. As for the instructors, it is challenging to design and provide a good learning strategy for learners to overcome the programming difficulties for a lecture with a large number of learners in the classroom, especially when there are novices who lack sufficient computer knowledge (Krpan, Mladenović, & Rosić, 2015; Woei, Othman, & Man, 2014; Robins, Rountree, & Rountree, 2003). A lack of time, teacher expertise, pedagogical choices, or lack of examples and practice in lectures are the factors which influence learners' understanding of programming (Waite, Curzon, Marsh, & Sentance, 2020). During practice for learning programming, learners may face many difficulties (Lahtinen, Ala-Mutka, & Järvinen, 2005), especially in programming debugging. These difficulties, which have been found to be related to programming knowledge, programming skills, and learners' interest and self-efficacy, will affect their success or failure (Ozmen et al., 2014). Thus, a proper guidance strategy as support would be helpful for learners to acquire better learning achievement and self-efficacy, and to enable them to apply their knowledge in new situations when learning programming (Hogan & Pressley, 1997; Wang & Hwang, 2017). Besides, issues which need to be further explored include how to give learners good guidance while providing them with opportunities to think and develop their own programming skills so as not to decrease their enthusiasm and interest in programming learning, and to promote positive experience and self-regulated programming learning as well (Law, Lee, & Yu, 2010; Uysal, 2014).

2.2 Related studies of programming learning

Practice makes learning perfect, and programming practice plays an important role in developing higher order thinking and problem-solving ability for programming learning (Fessakis, Gouli & Mavrodi, 2013; Bergersen & Gustafsson, 2011). Studies have been conducted to explore how integrated teaching strategies, methods, or e-learning tools can assist programming learning and practice. For example, researchers have adopted the problem-based learning strategy (Chang, Chung, & Chang, 2020), a structured-resource issue-quest approach (Hsu & Hwang, 2017), design thinking disposition (Tsai & Wang, 2020), or game-based learning (Jantan & Aljunid, 2012) to support programming training, and have argued that integrating various teaching and learning strategies benefited learners' learning performance, and promoted their satisfaction, enjoyment and motivation in computer programming. Venigalla and Chimalakonda (2020) adopted learning strategies with debugging practice to train students' programming ability, and found that students' confidence in learning programming was promoted after they participated in the guided debugging practices in the programming course (Chiu & Huang, 2015).

Besides, there are studies which have adopted e-learning tools or environments to support programming. For example, Mendes, Ivanov, and Marcelino (2005) used a web-based system to support instructors' monitoring of learners' work, and to propose and test the programming for learners. The system provided learners with feedback, self-practice, and challenges to facilitate their learning, and it also helped instructors to identify learners' learning problems. Areias and Mendes (2007) adopted a dialogue-based educational tool, ProGuide, that motivated and guided learners through a dialogue and put questions to support low-achievement learners in programming problem-solving activities. Law et al. (2010) developed the Programming Assignment aSsessment System (PASS) to give learners infrastructure and facilitation for learning computer programming, and they found that a well-facilitated e-learning setting enhanced learners' programming learning motivation and self-efficacy. Hsiao, Sosnovsky, and Brusilovsky (2010) and Jovanovic and Jovanovic (2015) used an adaptive e-learning system to create an individualized programming learning scenario through providing learners with personalized learning materials. The findings indicated that the adaptive navigation learning system helped learners promote participation and success rate in their programming performance (Hsiao et al., 2010). Muraina, Adegboye, Adegoke, and Olojido (2019) employed multiple modes, including video, game, practical and online classroom modes for programming learning. They indicated that employing multimodality for instruction was helpful for promoting students' programming learning performance. Weintrop and Wilensky (2020) explored the learning effects of adopting a block-based or text-based environment for programming learning, and revealed that no differences in programming practices or attitude were found between the two conditions.

Within the research of giving learners programming learning supports, scholars have suggested that Internet-based online environments are an important resource that provide students with more opportunities, and that the practice does not need to be limited to schools (Gomes et al., 2008). The Internet-based environment with a wide range of resources is good for learners' reference, while an environment full of unfiltered information might prevent students, especially novices, from accessing appropriate data related to the targeted learning content. Within the various e-learning support instruments and studies, the adoption of WebQuest, which inspires learners to read and study the essential and filtered thematic resources on websites, and then conduct tasks through reflecting on their own metacognitive processes, caught the researcher's eye.

2.3 The theories underpinning WebQuest and related studies

According to Dodge and March in 1995, WebQuest is a design activity that asks learners to search for information to complete a task in a way that challenges their intellectual and academic ability (Vidoni & Maddux, 2002). WebQuest is also an Internet and learner-centred learning method based on constructivism, which facilitates scaffolding learning. Constructivism is also an approach that encourages students to construct their own knowledge based on their experiences (Elliott et al., 2000). It also emphasizes that learners construct and develop knowledge through active participation instead of receiving information passively (Savery & Duffy, 1995; Tobias & Duffy, 2009).

Meanwhile, the scaffolding is an instructional technique that is used to help learners achieve better levels of learning acquisition that they would not be able to achieve without assistance. In the WebQuest environment, the learners are provided with links to organize and filter websites. The learners then undergo a conceptual shift through developing knowledge from the information and developing higher levels of learning through inquiry. Thus, WebQuest is also a scaffolded learning structure that helps students to access target web resources more easily through specifying and organizing the related links (March, 2004). The guides in the WebQuest activity then assist students in moving to higher levels (Aoki, 2004; Young & Wilson, 2002), which facilitates their construction of knowledge (Dodge, 1998; Lara & Reparaz, 2007), and their problem-solving ability (Sunal & Haas, 2008).

Several studies have been conducted on the integration of WebQuest, and have found that learners who learned with a WebQuest learning curriculum had better learning performance and better learning satisfaction (Yang, 2014). For example, Norlidah, Saedah, Mohd, and Alija (2013) conducted a review study to analyze the trends of WebQuest research which indicated that WebQuest is helpful in improving learners reading comprehension performance, and for college-level activities. Allan and Street (2007) integrated the WebQuest method into higher education for teaching training and found that college learners agreed that it was a good way to promote higher order thinking. It has also been suggested that the method be integrated into higher education classes (Şahin & Baturay, 2016). Ebadi and Rahimi (2018) adopted a WebQuest-based classroom to teach academic writing skills. Renau and Pesudo (2016) adopted the WebQuest strategy to enhance elementary school learners' motivation to learn English and to improve their digital competence. Segers, Mienke, and Ludo (2010) integrated WebQuest into an elementary school class for language learning, and found significant learning gains after the course, especially for reading comprehension. Chang, Chen, and Hsu (2011) created a learner-centred, collaborative and self-presentation learning environment, and explored the learning effects under three scenarios, traditional instruction, traditional instruction with WebQuest, and outdoor WebQuest instruction, for environmental education. It was found that the adoption of WebQuest helped learners to acquire more knowledge and experience, and they were better able to express their opinions and perspectives. Besides, the learners engaged more in the use of higher order thinking skills during the WebQuest activity than in traditional instruction.

In addition to the above-mentioned learning topics, it was considered that the adoption of WebQuest instruction would be helpful for programming learning as well. For example, Hsu and Hwang (2017) found that the web issue-quest approach benefited novices' (tenth graders) programming learning concepts, and they stated that instructors can adopt the structured resource-based issue quest to improve students' learning efficiency. Bui, Kim, Ho, Hồ, and Pham (2018) developed a WebQuest model with a mind-map tool to assist college students in learning programming language for fostering critical thinking. Adekunke (2020) revealed that teaching computer science with WebQuest processes made learners perform better than those who learned with the conventional lecture method. In sum, WebQuest fosters learners' abilities of analyzing and integrating information, and promotes their logical thinking ability (Dodge, 2001). It also helps them acquire knowledge and develops their critical thinking skills (Norlidah et al., 2013). Since programming learning requires mathematical, logical and technical abilities, and since programming means asking students to use the above-mentioned ability to solve the problems, these procedures are actually similar to the WebQuest processes, including what thought processes are needed to solve the tasks and what next steps have to be taken once the problems have been solved (Mărunțelu & Popescu, 2020). As related studies have revealed that WebQuest holds considerable promise for effective learning, the researcher aimed to integrate this potential learning support into developing a pedagogical model for programming learning.

2.4 Design strategies of WebQuest for programming learning

A WebQuest activity consists of a series of web pages and has five essential parts: Introduction, Task, Process, Resources, Evaluation and Conclusion. First, a topic and background knowledge are introduced, and particular tasks to be done are given. Then, the sources of information are listed to support and give learners individual steps and guidance for accomplishing the tasks. The instructors have to search for and organize suitable Internet resource such as video clips, relevant websites, educational software, and multi-media resources. The rubric of task evaluation is also provided. Last, a conclusion that summarizes the learning process is given. When designing WebQuest activities, the Internet environment provides learners with rich information for gaining knowledge in the resources part. It has been suggested that learners be provided with an organized webpage as an interface between them and the Internet so that they can focus on solving the tasks instead of searching for and checking irrelevant information (Kuiper et al., 2005; Segers & Verhoeven, 2009). One study also suggested including an additional Share and Compare stage in the WebQuest activity after learners have finished their work so as to encourage them to experience solutions, make comparisons with peers, and reflect upon their learning (Young & Wilson, 2002). Moreover, numerous studies have revealed that incorporating WebQuest and learning strategies into instruction such as project-based learning instruction (Chen, 2019; Laborda, 2009), the problem-solving method (Aydin, 2016), the 5E-learning model (Şahin & Baturay, 2016), cooperative learning (Kurtulus¸ & Kılıc¸, 2009), discovery and the scaffolded learning method (Şahin & Baturay, 2016) promoted experiential and meaningful learning.

When constructing the guidance for programming learning, problematizing scaffolds would be effective for promoting their active participation and the development of self-regulation strategies (Feng & Chen, 2014) because solving practical programming problems facilitates learners' self-regulation and inspires them to care about the issues. The process of debugging is a way for learners to experience a way of solving practical programming problems which involves exploration, observation and reflection (Heikkilä & Mannila, 2018). Meanwhile, the debugging activity is considered as an important central feature of programming because learners improve their programming comprehension ability during the repeated trial-and-error debugging problem-solving processes (Lee & Wu, 1999). While, Clark (2009) found that when problem solving, learners without relevant concepts in their knowledge background cannot easily find solutions, and might end up blindly searching and learn nothing. Learners, especially beginners, could also have problems checking the quality and reliability of the information they have found (Kuiper et al., 2005), which might increase their cognitive load.

In this respect, the debugging practice could be integrated into the WebQuest activities to create opportunities for learners to think, to explore, and to develop their problem-solving skills. At the same time, the WebQuest activity would be good scaffolding for learners to experience problem-solving since it gives tasks to learners and asks them to search for and organize data to find the answers. The WebQuest lets instructors organize the possible useful information on the Internet while giving learners a certain degree of freedom to choose and read the content according to their needs, this can help learners pay more attention to the targeted learning problems instead of spending too much time collecting and checking the reliability and relevance of the large amounts of information they have found (Kuiper et al., 2005). Meanwhile, it also leaves opportunities for learners to think and to explore from the filtered learning information, and hence they would be able to address the tasks and attain the knowledge during the problem-solving process with appropriate guidance (Clark, 2009).

2.5 Summary

Programming learning and teaching is challenging both for instructors and learners (Gomes & Mendes, 2007; Milne & Rowe, 2002; Robins et al., 2003). The debugging training is one of the important instruments for programming practice because learners' problem-solving ability to test and debug the resulting codes (Brooks, 1999; Bergersen & Gustafsson, 2011; Lee & Wu, 1999). Meanwhile, providing learners with suitable learning supports could help them achieve their learning goals. WebQuest is an Internet and learner-centred learning method whereby learners have to complete learning tasks with instructor-organized information that is possibly useful, while still having a certain degree of freedom to choose and read the content according to their needs and learning problems. The advantages of WebQuest for programming learning include motivating learners to interact with information-based Internet resources, and providing opportunities for learners to actively engage in learning. Hence, in this study, the WebQuest activities were designed and integrated into a programming course. The learners had to complete the programming tasks including debugging practice in the WebQuest activity, and be able to read, understand, recognize and trace existing errors in the programs. Besides, in order to encourage learners to experience the problem-solving process, the sequence of the WebQuest processes was modified such that the detailed WebQuest processes were not provided until after all of the activities. It was the aim of this study to use the WebQuest instructions as scaffolds to support learners' programming problem solving and to promote their learning motivation, positive experience and performance in programming as well.

3.1 The modified WebQuest processes

In order to encourage learners to conduct self-problem-solving, the third step, Process, was modified such that the instructor did not provide the learners with the detailed steps of the learning tasks at first (Figure 1a), instead, they were only given the outline as a reference. The theoretical foundation of the modified design is based on constructivism to implement the scaffolded learning. The learners had to read the steps and search for possible ways from the hints of the outline to solve the programming problems. They were also encouraged to use the outline information as keywords, and to further explore and digest other essential resources on the World Wide Web for completing the programming tasks. After the deadline of the WebQuest activities, the detailed steps of the tasks were then provided (Figure 1b). Figure 1 presents the print-screen of the WebQuest activity presented to the learners.

4.1 Research procedure

Before the formal experiment, all the participators had taken the pre-course for 16 hr to ensure that the learners at same starting point. The study duration was 18 weeks (3 hr/week). The E.G. underwent WebQuest teaching, while the C.G. received the information technology-integrated teaching. During the experience, the eight weekly challenge tasks, learning materials, and four WebQuest activities (only for the E.G.) were distributed to the learners. After the 18 weeks, both groups had to complete the programming tests to get their semester grades. In order to understand how the WebQuest activity facilitated programming learning, in the first week, the learners in the E.G. completed the programming-attitude and problem-solving experience pre-questionnaires. The post-questionnaire survey was administered separately to the experimental group learners after the experiment to collect learners' feedback of learning program through WebQuest activity. The research procedure of the whole study is presented as Figure 2-2.

4.2 Data collection

The study collected learners' programming test scores, semester grades, the numbers of learners completing the weekly challenge tasks, and the WebQuest activity scores to determine whether there were learning differences between the E.G. and C.G. The questionnaire results and learners' learning diaries were also analyzed for understand learners' feedback during the programming learning processes.

For the quantitative data, the programming test was conducted after the course was finished. All the learners had to take the programming test in order to measure their understanding of the topics in order to quantify the range of their knowledge changes. The test was prepared based on the curriculum objectives and teaching content that included three batteries of tests, each of which required the learners to complete the interface design and programming construction. The first battery test targeted the basic level and required the learners to use the objects event and the basic variable concept to create a collection game. The second battery test was at the intermediate level and required learners to use the variable concept to create an online calculator. The last battery test was at the advanced level and required the learners to use a variable to create flash animation and simulated object movement. The total score of the tests was 8 points, with 3, 2 and 3 points respectively for the three batteries of tests. The rubric of semester grades included learners' performance in the midterm and final term project presentation. The learners had to give a proposal plan of how they would use the learned programming skills to implement game-based learning materials as the midterm presentation, and then they constructed the materials and gave a demonstration in final term presentation.

The weekly challenge tasks were announced at the beginning of each course and the tasks were the extended levels that required learners to adopt advanced knowledge to solve the programming problems. The learners were encouraged to conduct the tasks and share their problem-solving process with the instructor and peers. The study recorded the numbers of learners who finished the tasks for further analysis. Besides, the WebQuest activities scores and the numbers of learners completing the WebQuest activities were also recorded. The rubrics of the WebQuest activities scores were: handing in the programming code and completing the activities perfectly for 2 points, handing in the programming code but with bugs for 1 point and not handing in the programming code but describing the problems met during coding for 0.5 points.

The questionnaires regarding the learners' programming experience were conducted before and after the experiment for the E.G. This study used a satisfaction-level questionnaire to understand learners' feelings after they had undergone the WebQuest activities for the programming course. The study referenced relevant studies to prepare the questionnaire, which comprised three scales with a total of 15 items: five items for the programming learning experience scale, five items for the steps taken before problem-solving scale, and five items for the problem-solving reflection. The scale is a 5-point Likert scale where strongly agree was 5 points, agree was 4 points, neither agree nor disagree was 3 points, disagree was 2 points, and strongly disagree was 1 point. The overall Cronbach's alpha values of the pre-questionnaire results of each scale were 0.82, 0.71 and 0.82, and those of the post-questionnaire results were 0.82, 0.85 and 0.88, thereby indicating that the two questionnaires possessed superior internal consistency. Moreover, the learners were encouraged to record their learning diaries after each course, and their diaries were also collected as qualitative data for further analysis.

5.1 Learning performance: programming scores and weekly challenge task

The study adopted the independent samples t test to compare the learners' programming performance in order to answer the first research questions. The data from the programming test scores and their semester grades were analyzed. It was found that both the average scores of the programming test and the semester grades of the E.G. were higher than the scores of the C.G., and the learners in the E.G. performed significantly better in the semester grades than did the learners in the C.G. (Table 1). The researcher then analyzed the detailed scores of the programming test and found that the scores of the first battery test of the E.G. were significantly higher that the scores of the C.G. (Table 2).

When analyzing the data collected from the weekly challenge tasks, the study calculated the number of learners who completed the challenge tasks of the two groups (Figure 3). According to Figure 3, it can be seen that the number of learners in the E.G. who completed the tasks was significantly more than the number in the C.G. The average numbers of learners who completed the challenge tasks each week were 2.85 for C.G. and 18.25 for E.G.

Further analysis was conducted through adopting the programming test scores, WebQuest activity scores and the numbers of learners completing the WebQuest activity in the E.G. for linear regression analysis. According to Tables 3a, b, it was found that there were significantly positive correlations between the scores of the programming test and the WebQuest activities scores (Table 3a, p = .000; Table 3b, t = 3.847, p = .000), indicating that the learners who accomplished the WebQuest activities more completely performed better in the final programming test.

5.2 Learning performance: programming scores and debug practice of the WebQuest activity

In order to understand whether the strategy of integrating debug practice of the WebQuest activity had learning influence on programming performance, the study further analyzed the WebQuest activity scores of the E.G. in more detail. The learners were separated into three categories including the one who finished the debug practice completely, partly, and none. and the results indicated that there was a significant effect of test scores on learners' level of debug practice completion [F (2, 47) = 15.285, p = .000] (Table 4a). Post hoc comparisons using the Bonferroni test indicated that the mean score for the one who finished the debug practice completely (M = 6.30, SD = 1.525) was significantly different than the learners who completed the debug practice partly (M = 4.45, SD = 2.385), and who did not complete the debug practice (M = 2.00, SD = 0.926). Moreover, it was also found that learners who finished the debug practice completely had significant higher scores of the second [F (2, 47) =12.716, p = .000] (Table 4b) and third [F (2, 45) =12.326, p = .000] (Table 4c) battery of tests than those who did not. The findings revealed that learners who followed the WebQuest activity to complete the debug practice performed better in the intermediate and advance level test of the programming test (Table 4).

5.3 Questionnaire results

The data from the questionnaires of the E.G. were analyzed to answer the second research question. The questionnaire items collected the learners' programming experience, steps taken before problem solving, and problem-solving reflection. According to the results of the descriptive statistics of Table 5, the average score of each scale of the post-questionnaires was higher than the scales of the pre-questionnaires, especially for the programming experience scale which improved from 3.44 to 4.08 points. The improvement of the data indicated that the learners had better learning programming experience from the aspect of finding ways, methods and resources for programming problem solving, and beliefs, stratification and confidence for solving programming problems. Besides, the paired sample t test was also conducted to further analyze the data of each question (Table 5), it was found that the averages of most question items of the post-questionnaires were higher than the averages of the pre-questionnaires, and some of the evaluations achieved significant difference (Q1, Q2, Q3, Q5, Q6, Q7, Q8 and Q10). For example, the average scores of six questions (Q1, Q2, Q3, Q6, Q7 and Q13) from the pre-questionnaires were less than 4.0 points, but they were over 4.0 after the experiment. The statistical results of the paired sample t test also achieved significant differences, indicating that the learners in the E.G. were more confident in solving the programming problems; moreover, they would think from various aspects to find methods for overcoming the programming problems. Besides, although the average scores of three questions (Q5, Q8 and Q10) of the post-questionnaires were less than 4.0 points, the scores of these items still achieved significant improvement after the experiment in the aspect of learners' confidence in facing programming problems, finding more than one solution, and thinking before doing the programming problem solving, according to the results of the paired sample t test. Meanwhile, according to the results of Table 5, it was noticed that a small part of the question items regarding self-reflection evaluation did not improve (Q14 and Q15). These findings are discussed with the qualitative data later.

5.4 Qualitative feedback

The feedback from the learners' learning diaries was analyzed and is organized into four categories in Table 6 to answer the second question. According to the learners' feedback, it was found that the learning difficulties of the WebQuest activities were learners' fear and lack of confidence in their ability to program. Besides, the topic of programming variable and the weekly challenge tasks were the hardest parts for the learners. Some stated that they spent a great deal of time finding the solutions from the resources, but still failed to solve the problem, and thus felt a little pressure during the activity (Table 6a).

It was satisfying to find that most learners tried to do the self-survey when they met programming problems. When they got the WebQuest missions, they first followed the hints of steps on the WebQuest learning sheets to complete the basic tasks and to comprehend the basic parts (Table 6b). The learners mentioned that the way of trying on their own first through surveying on the Internet and getting the answers the next week helped give them a deeper impression of what they had learned. Moreover, it was also noticed that some learners used the step-hints in the WebQuest learning sheets as keywords to find information on the Internet. The learners revealed that when they got further information on the Internet, they filtered it to pick out the required parts, and then used trial-and-error. The learners indicated that through these repeated processes, they completed the tasks on their own and felt a sense of accomplishment (Table 6c).

7.1 Limitations and future work

Two limitations that need to be taken into account are that first, due to considering the real classroom learning scenario and the instructor's routine instruction, a pre-activity programming test was not collected as the pre-test for statistical analysis. Second, only four WebQuest activities were studied, and they were designed for this study. The results may therefore not hold for all implementations of WebQuest. Besides, future directions are suggested for the subsequent research. First, the integration of a longer period of WebQuest activities is suggested. The short-term goal of WebQuest is to enhance knowledge acquisition; a long-term goal could be adopted to examine whether learners' knowledge is extended. Second, after-practice WebQuest activities could be designed to strengthen and reinforce learners' self-reflection and to extend and refine their knowledge. Third, it is suggested that the expert validity of the designed web resources be promoted through inviting more programming experts to review and evaluate the learning material designed in the WebQuest activity. Finally, a pre-activity programming learning sheet or examination could be conducted to understand learners' prior knowledge before the experiment to ensure that they are all at the same starting point before the instruction.