Computational Thinking (Algorithms) Through Unplugged Programming Activities: Exploring Upper Primary Students’ Learning Experiences

In the recent year, Computational Thinking (CT) has gained much attention in educational research and practice. CT skills can be taught via computing activities that involve different types of programming tasks or via Unplugged Programming Activities (UPA) that do not involve the use of digital devices to represent and deliver programming concepts. UPA is an appropriate teaching approach for schools that do not have sufficient technological infrastructure. Studies have shown the effectiveness of UPA in developing CT skills and is comparable to the technology driven learning method. The aim of this study is to explore the experience of primary school students on their learning of algorithms, which is one of the CT skills, through the UPA method. A total of 31 students from a rural primary school were exposed to the learning about the algorithm concept (an aspect of CT skills) via UPA learning materials. From the responses gathered through interviewing nine of these participants, four main themes (Good Learning Quality, Much Knowledge, Easy and Useful) related to their learning experiences have been derived. These positive themes provide evidence on the appropriateness of employing UPA for teaching the algorithm aspect of CT, particularly for schools located in areas with limited access to adequate technological infrastructure. This study may serve as a reference in establishing a comprehensive UPA module for teaching algorithms aspect of CT skills.


Introduction
The term Computational Thinking (CT) was first introduced in 2006 (Selby, 2013). CT skills refer to a collection of mental tools that enables an individual to solve problems more effectively by imitating a computer scientist's way of thinking (Wing, 2006). It is an approach for solving problems, designing systems and comprehending human behaviour that draws on the fundamental concepts of programming (Wing, 2006). Over the past decade, CT has gained much attention in educational research and practice (Wright, Rich, & Leatham, 2012;Shute, Sun & Asbell-Clarke, 2017). Many researchers view CT skills as essential in computer coding or programming (Wing, 2010;Wright et al., 2012;Shute et al., 2017). However, according to National Research Council (2012), CT skills are relevant not only to programmers but should also be acquired by everyone.
Today, CT is considered as one of the skills that must be taught, learnt and mastered by the young generation (Weintrop et al., 2014). Mastering CT skills will provide this young generation with greater opportunities to build up their competitiveness in the fast-paced digital economy (Bocconi et al., 2016;Weintrop et al., 2014). CT skills have been integrated into the education systems across the world and Malaysia is the first country in South East Asia region that incorporates CT skills into its national curriculum (Abas, 2016).
CT skills are often taught via computing activities that involve different types of programming tasks (Brackmann, et. al., 2017). However, another approach to teach CT skills is via Unplugged Programming Activities (UPA). UPA is defined as activities that do not involve the use of digital devices; instead logic games, cards, strings or physical movements are used to develop the understanding on programming concepts (Kalelioglu, Gülbahar & Kukul, 2016). UPA is an appropriate teaching approach for schools that do not have sufficient technological infrastructure such as electricity, Internet, computers, mobile devices, and/or other electronic devices (Amrita, Muir & Rao, 2016).
The limited access to technological devices and facilities such as computers, tablets, educational robotics and other devices is a global issue (Sadatul, 2017). Countries in Africa, Asia and even many parts of European continent are still left behind with regards to the implementation of technology-based teaching and learning approach (UNESCO Institute for Statistics, 2014). A similar situation is found in Malaysia where the use of technology in education, particularly in some rural areas, is still in its infancy. Fifty percent (50%) of teachers use their own computers for teaching and learning purposes in five areas of Selangor and 35% of schools do not have well-equipped computer laboratories and servers (Samuel & Zaiton, 2007) while in a recent report, about 70% of schools in Sarawak recorded minimal use of information and communication technology (ICT) for teaching and learning and about 6%, mostly in the rural and interior, still require much assistance and intervention for its adoption (Aubrey, 2019). Hence, UPA can be a great alternative to teach CT skills in such schools. The aim of this study is to explore the experience of primary school students on their learning of algorithms, which is one of the CT skills, through the UPA method.

Computational Thinking (CT) Skills Six Distinctive Aspects of CT Skills
Computational thinking is one of the crucial skills for successful problem solving in an innovation driven and complex society (Kale et al., 2018). According to Shute et al. (2017), CT skills consists of six aspects; (a) Abstraction -identify the important elements of a problem or situation (b) Decomposition -breaking a problem into smaller parts.
(c) Algorithms -identifying steps in solving a problem.
(d) Evaluation -finding the most efficient solution to a problem. (e) Generalization or pattern recognition -applying a previous solution or approach to solve a problem. (f) Automation -using the information-processing agent as the solution to a problem.

Effects of CT Skills on Learning
The goal of learning CT is to improve knowledge on some fundamental elements of Computer Science and support problem solving based on computational concepts (Wing, 2006). The integration of CT skills in teaching and learning has brought positive impacts to students' academic achievement as well as their learning motivation (Mironova, Amitan & Vilipõld, 2012;Weintrop et al., 2014;Gardeli & Vosinakis, 2017). Mironova et al. (2012) reported that students, especially those from the weak group, showed improvement in their examination results for several subjects after learning CT skills. Their performance, particularly in Science and Mathematics subjects, has improved and thus, reducing the gap between them and the advanced group. Possessing CT skills also helps a learner to deepen the learning of certain contents, particularly for STEM subjects (Repenning, Webb, & Ioannidou, 2010). Zaman et al. (2019) has also reported how group interaction benefits the development of computational thinking skill among rural children and Belanger, Christenson and Lopac (2018) examined the effects of teaching CT on the confidence and problem solving ability among students of different age levels in suburban and rural schools.
Developing CT Skills Through UPA Brackmann et al. (2017) reported a significant increase in CT skills among students after being exposed to UPA in their teaching and learning sessions. The use of UPA in developing CT skills is as effective as using technology-driven systems or gadgets, thus turning it into a good alternative for introducing programming to students (Brackmann et al., 2017;Faber, Wierdsma, Doornbos, Van Der Ven, & De Vette, 2017). The use of UPA in developing CT skills has also shown positive effects on students' motivation (Faber et al., 2017;Gardeli & Vosinakis, 2017).

Algorithm
Algorithm is one the distinctive aspects of CT (Shute et al., 2017;Wing, 2006). Erickson (2019) defines algorithm as "an explicit, precise, unambiguous, mechanically-executable sequence of elementary instructions, usually intended to accomplish a specific purpose". Various problems can be solved using algorithms (Sharma & Khurana, 2013). For example: How to plan a tour that involves visiting several towns in the cheapest possible order? How to share information as well as keep secrets? Algorithms are useful for solving routine questions or exercises in which algorithms can be derived based on previous experience (Bodner, 1987).
Increasing scholarly attention has been given to the importance of algorithms in our daily contexts (Musiani, 2013). Algorithms are seen as powerful entities to control, govern and shape numerous contexts moving beyond its conventional definition as encoded procedures to transform input data to a desired output (Lee, 2018;Musiani, 2013). As algorithms have been introduced to Malaysia education syllabus through the subject "Teknologi Maklumat dan Komunikasi (TMK) -Information Technology and Communcation", it raises the question on how primary school students perceive the learning of the algorithm concept? Studies about algorithms as well as other CT skills have been carried out in many western countries, particularly in European countries. However, such studies are rather minimal in the Asia region. In Malaysia, studies on CT skills mostly focus on activities that involve the use of technological devices (Anna et al., 2017;Anna & Sabariah, 2014).

Research Design
The study employed a mixed method where both quantitative and qualitative methods were used to gather data for the purpose of this study. A pre-test and post-test experimental design was used to assess 31 students' perceived learning of the algorithm skill of CT. Nine out of 31 participants were interviewed after exposing them to the concept of algorithm via the UPA approach.

Learning Material
The learning material which was designed to develop the concept of algorithms consisted of two parts. The first part is the grid for 'Steps for Making Fruit Salad' which was adapted from Computer Science Without a Computer (n.d.). In this part, as shown in Figure 1, students were required to draw arrows on the grid to indicate the steps. Students' performance on accomplishing the task will depict their ability to make planning of steps in solving a problem.
The second part requires students to list out all the steps for making fruit salad. This part, as shown in Figure 2, was adapted from "Teknologi Maklumat dan Komunikasi" Year 6 textbook for Malaysian primary education as well as the study by Brackmann et al. (2017). Students' performance on this task will depict their ability to give step-by-step instruction in solving a problem.

Instruments
Two instruments were developed for the purpose of this study: • Students took the pre-test before experiencing the learning material and took the post-test after going through the learning material. The test consisted of two section: Section A and Section B. Section A comprised five multiple-choice questions while Section B comprised three structured questions. The test was adapted from the Year 6 "Teknologi Maklumat dan Komunikasi (TMK)" textbook. • Structured interview questions adapted from Chen and Keong (2017) were used to further explore students' perceived learning. Students with the highest scores and who showed more than 20 marks increment in their post-test scores compared with their pre-test scores were selected for the interview sessions.

Participants
Thirty-one students from one primary public school in Simunjan, Sarawak took part in this study. All participants were students of upper primary classes with their age ranging from 10 to 12 years old. From the thirty-one students who took part in pre-test and post-test, nine were chosen for the structured interview.
The selection of the participants for interview sessions were based on their performance in the pre-test and post-test. Four participants (P2, P5, P8, P9) scored 100% mark in their post-test. Three participants (P3, P4, P6) showed increment of more than 20% in their post-test compared to their pre-test. Two participants (P1, P7) showed decrement of more than 15% in their post-test compared to their pre-test. Thus, the sample comprised students with perfect score, students who showed improvement in algorithm learning as well as students who showed deterioration in algorithm learning.

Data Collection Procedure
The data collection was conducted after obtaining approval from the Ministry of Education, State Education Department and principal of the school involved in this study. The researcher briefed all 31 students and their parents on the purpose of the study. They were also informed about the audiotaping during the interview sessions. Parents of all participants had given their consent by signing the consent form witnessed by the school principal. All interview sessions were conducted after school and Malay language was used during the sessions as most students are better versed in this language.
The pre-test was administered to all the 31 students. They were exposed to the learning materials for about 40 minutes after school hours in batches. The post-test was administered right after each batch completed the learning materials.

Data Analysis
Data gathered from the pre-test and post-test were analysed using SPSS software. Paired ttest was carried out to examine whether there is a significance difference between the pre-test and post-test.
The interview data gathered was qualitatively analysed to identify the emerging themes. The interview transcript was read through and participants' significant responses toward the interview questions on perceived learning were coded. The appropriate themes were then derived from the compiled coded responses. Themes were classified into favourable and unfavourable responses. Through prompts and follow-up questions, descriptions and justifications of participants' favourable and unfavourable responses were gathered. Figure 3 shows the steps taken to derive the themes.

Findings and Discussion
As shown in Table 1, there is no significant effect of the learning material in increasing the participants' algorithm skills (p>.05, p=.259) although the mean for post-test score (66.9%) is slightly higher than the mean for pre-test score (64.3%) as shown in Table 2. This result is probably due to the frequency of the exposure to the learning materials where all the participants had only been exposed one time to these materials. Higher frequency of exposure may facilitate the participants' learning (Effiong & Igiri, 2015) and this may potentially assist participants in producing better score for the post-test. A total of nine participants were selected for structured interview. Table 3 shows the matrix of the selected participants' demographic background.  What do think of your learning about giving step-by-step instruction when doing the activity?". The result shows that a total of eight participants (89%) expressed that they had good learning quality while one participant (11%) expressed that she had poor learning quality throughout the 'planning of steps' activity (Question 1(i)).

Enjoy
Five participants (P1, P5, P6, P7, P8) commented that they enjoyed the 'planning of steps' activity mainly due to the attractive presentation of learning materials that captured their attention and aroused their learning motivation.

Fun
Three participants (P2, P3, P4) stated that the activity was fun because they were learning with their peers. Table 5 shows the feedback from each participant on the 'planning of steps' activity. For question 1(ii), eight participants (89%) expressed that they had good learning quality throughout the 'giving step-by-step instruction' activity compared to only one participant (11%) who expressed that she had poor learning quality.

Fun
Four participants (P2, P3, P5, P6) stated that the 'giving step-by-step instruction' activity was fun and they really liked the activity. They commented the activity was easy to be accomplished.

Learn new knowledge
Four participants (P1, P4, P8, P9) commented that they learned new knowledge while doing the activity. P1 and P4 stated that it was their first time doing such an activity. Table 6 shows the feedback from each participant on the 'giving step-by-step instruction' activity. Eight participants (89%) stated that they gained much knowledge from the 'planning of steps' activity while one participant (11%) expressed that she gained little knowledge (Question 2(i)).

Problem solving: Finding the best path
Five participants (P1, P4, P5, P8, P9) commented that they had learnt about finding the best path. From the participants' comments, it can be concluded that they managed to learn about algorithm skill as finding the best path is one of the fundamental algorithm skills (Sharma & Khurana, 2013).
Problem solving: More than one way to solve a problem Two participants (P3, P6) stated that they had learnt that there is more than one way to solve a problem. Table 7 shows the feedback from each participant on the knowledge that they gained from the 'planning of steps' activity. Finding the best path to make fruit salad. Mencari jalan terbaik untuk buat salad buah.

P2
Ways to make fruit salad.

Cara-cara bagaimana nak buat salad buah. P3
There is more than one way to solve problem. Ada lebih dari satu cara untuk selesaikan masalah.

P4
Find the best path on how to make fruit salad. Cari jalan terbaik untuk buat salad buah.

P5
Find the best path on how to make fruit salad.

P8
Finding the best path to produce fruit salad.

Mencari laluan terbaik untuk menghasilkan salad buah. P9
Finding the best path to make fruit salad.

Mencari laluan terbaik untuk membuat salad buah.
Similar to question 2(i), eight participants stated that they gained much knowledge from the 'giving step-by-step instruction' activity compared to only one participant who expressed that she gained little knowledge although she also mentioned 'I don't know', which can be interpreted as not gaining any knowledge. Most of the participants took longer time before they could answer these two questions.

Correct and accurate steps
Seven participants (P1, P2, P3, P4, P5, P6, P8) commented that they had learned the correct and accurate steps to make fruit salad by doing the 'giving step-by-step instruction' activity.

Systematic
Participant 9 (P9) was the only participant who mentioned that he had learned about systematic way to solve a problem, which is capability needed by computational skills (Kalelioglu et al., 2016). Table 8 shows the feedback from each participant on the knowledge that they gained from the 'giving step-by-step instruction' activity. The correct and accurate steps to make fruit salad. Langkah-langkah yang betul dan tepat untuk buat salad buah.

P2
The correct and accurate steps to make fruit salad. I also learn how to take care of cleanliness when making fruit salad. Langkah-langkah yang betul dan tepat untuk membuat salad buah. Saya juga belajar bagaimana menjaga kebersihan apabila membuat salad buah.

P3
The correct and accurate steps to make fruit salad.

P4
The correct and accurate steps to make fruit salad. Langkah yang betul dan tepat untuk buat salad buah.

P5
The correct and accurate steps to make fruit salad.

P6
The correct and accurate steps to make fruit salad.

Langkah-langkah yang betul dan bertepatan untuk buat salad buah. P7
(The participant didn't give her answer) P8 The correct and accurate steps to make fruit salad.
Theme 3 : Easy Two interview questions were related to the difficulty of the activity. The questions were: Question 3(i) : Do you find the' planning of steps' activity easy or difficult? Why? Question 3(ii) : Do you find the 'giving step-by-step instruction' activity easy or difficult? Why? Eight participants (P1, P2, P3, P4, P5, P6, P8, P9) mentioned that the 'planning of steps' activity was easy.

Easy to understand
Four participants (P1, P3, P4, P6) commented that the activity was easy to understand. From the comments received, it is inferred that the activity is rather simple for upper primary students. A more diverse set of algorithm learning materials with different levels of difficulties will be appropriate to facilitate further learning. According to Faber et al. (2017), students became much engaged and motivated with much difficult unplugged activities.

Sufficient time
Four participants (P2, P5, P8, P9) commented they were able to complete the activity within the given time frame. Table 9 shows the participants' responses on the difficulty level of the 'planning of steps' activity. Easy. I can finish the activity within the given time.

Easy to understand
Six participants (P1, P2, P3, P5, P8, P9) commented that the activity was simple to follow. Participants with age range between 10 to 12 may be given more difficult and more challenging activity.

Helpful figures
Two participants (P4, P6) commented that pictures in the learning materials were helpful. Figures such as pictures, graphs and charts are able to facilitate learning (Shabiralyani, Hasan, Hamad & Iqbal, 2015). Table 10 shows the participants' responses on the difficulty level of the 'giving step-by-step instruction' activity.

Applying the knowledge in daily routine
Seven participants (P1, P2, P3, P4, P5, P6, P8) stated that they could apply the knowledge gained from the 'planning of steps' activity in their daily routines including create a new menu. This indicates that the participants are capable to construct algorithms based on their prior experience (Bodner, 1987).

Sequencing skill
Participant P9 commented that he had learned sequencing skill from the 'planning of steps' activity. Table 11 shows the participants' responses on the usefulness of the 'planning of steps' activity. The same eight participants (89%) felt that the 'giving step-by-step instruction' activity was useful.
Help in memorising Seven participants (P1, P2, P3, P4, P5, P6, P9) commented that the 'giving step-by-step instruction' activity help them in memorising the steps for making fruit salad. This is diverting from the purpose of algorithm skill itself which is to guide the solving of problems. The participants might be too attracted with the making of fruit salad instead of linking it with the 'step-by-step' approach, which is an algorithm skill. Table 12 shows the participants' responses on the usefulness of the 'giving step-by-step instruction' activity. Useful. It helps me in memorising steps for making fruit salad.

Unfavourable Response
Participant 7 (P7) gave unfavourable responses during the interview session. To the researchers' opinion, there is possibility that she felt unmotivated because she was unable to do well with the learning materials and this led to her unfavourable responses for all interview questions. P7 was one of the two students who showed deterioration in algorithm learning as her post-test score was lower than her pre-test score.

Conclusion
This study shows that UPA is an appropriate alternative approach to teach CT skills especially for schools located in areas with limited access to adequate technological infrastructure. From the responses gathered through interviews, four main themes (Good Learning Quality, Much Knowledge, Easy and Useful) had emerged and these positive themes provide evidence on the appropriateness of employing UPA for teaching the algorithm aspect of CT. Although there is an insignificant difference between the pre-test and post-test that measures the learning of the algorithms skill, the average post-test scores are higher than the pre-test scores. Besides adding to the literature on the method of teaching CT skills (Kong & Abelson, 2019), the Ministry of Education, State Education Department, District Education Office, schools and teachers may use this study as a reference in establishing a comprehensive UPA module for teaching algorithms aspect of CT skills. Such module can be a good reference for teachers across the nation. The findings will add Future studies may consider reinforce the learning of the algorithms aspect by allowing students to experience with more learning materials and replicate the study in suburban and urban schools to examine the generalisability of the findings.