Tendencies towards Computational Thinking: A Content Analysis Study

In this research, we analyzed the content of a practice-based research published in SSCI, ESCI and ERIC indexed journals related to Computational Thinking (CT) between 2019 and 2021. For this purpose


Introduction
With the use of computer in many fields of daily life, the knowledge and skill of using it effectively in order to develop solutions to the problems encountered has gained importance.In the 21st century, where problems are digitized and transferred to the computer environment, it is necessary to have some high-level thinking skills.Problem solving, analytical thinking, critical thinking, algorithmic thinking, and computational thinking (CT) are among these important skills (International Society for Technology in Education [ISTE], 2015).CT skill is the knowledge, skill and competence required for using computers and other information processing tools to solve problems encountered in daily life (Özden, 2015).Seymour Papert (1996) referred to the current understanding of CT by emphasizing the use of the computer as a tool to assist thinking.There has been more discussion on CT since Wing (2006) showed that CT was among important basic skills such as reading, writing and arithmetic operations in everyday life, not just for computer science.
Although research on CT has increased in recent years, there is no consensus on the definition, scope, development and evaluation of CT (Weintrop et al., 2016).Román-González, Pérez-González and Jiménez-Fernández (2017) categorized CT-related definitions into general definitions (Aho, 2012;Wing, 2006), operational definitions (CSTA & ISTE 2011), and curriculum-related definitions (Brennan & Resnick, 2012).Tang, Yin, Lin, Hadad and Zhai (2020) classified CT as definitions related to computer science (Brennan & Resnick, 2012;Weintrop et al., 2016) and problem-solving skills (CSTA & ISTE 2011;Selby & Woollard, 2013).Within these definitions, different sub-skills revealing the development of CT are also included.Researchers have stated that skills such as algorithmic thinking, abstraction, decomposition, generalization, and debugging are a part of CT (Grover & Pea, 2013).Hsu, Chang and Hung (2018) addressed CT within the framework of 19 different subskills, also including abstraction.Stating that CT is the ability to design planned, systematic and reproducible methods to solve problems, the researchers associated CT with high-level thinking skills such as decomposition, pattern recognition, abstraction and algorithmic thinking skills (Rowe, Asbell-Clarke, Baker, Gasca, Bardar, & Scruggs, 2018;CSTA, 2017;Wing, 2006).As can be seen, different sub-skills used within the framework of different definitions can also expand the scope of CT.
With the increasing understanding of the importance of CT after the research, while policy makers and researchers debate on how to integrate CT into their curriculum (Bocconi et al., 2016;So, Jong, & Liu, 2020), educators, on the other hand, have started to make applications in the courses for the development of CT skills of students in different age groups.Various attempts are being made to integrate CT and computer science into primary, secondary and high school curricula in many countries such as the USA, UK and Australia (Passey, 2017).The fact that CT is seen as a basic skill used in daily life (Wing, 2006) has also accelerated research on CT development of students at different levels from pre-school to higher education.However, since CT is a new concept, it has been stated that teachers have some lack of knowledge about what exactly this skill is, how it is developed, how it is evaluated and how it can be integrated into the courses (Hsu et al., 2018;Mannila et al., 2014;Yadav, Gretter, Good, & McLean, 2017).Recent research has focused on teacher education (Baroutsis, White, Ferdinands, Lambert, & Goldsmith, 2019;Umutlu, 2021) and CT development of younger students (del Olmo-Muñoz, Cózar-Gutiérrez, & González-Calero, 2020, Wang, Choi, Benson, Eggleston, & Weber, 2021).With increasing research, the trends of CT are also discussed.As CT is seen as a basic skill and its importance has been revealed by researchers, it has also increased the efforts to include this concept in the curriculum of students at different levels.We anticipate that it will also support efforts to reveal common points in research, include CT in curricula and apply it to classrooms.For this reason, examining the research in the literature, especially in practice-based and high-indexed journals, will guide policy makers, practitioners, and educators.There has been previous systematic review or content analysis research on CT.Haseski and İliç (2019) conducted a content analysis study to determine the data collection tools used to measure CT.Kalelioğlu (2018) systematically examined the research on CT and revealed the trends of them.Tang et al. (2020) conducted content analysis research to reveal the international publication trends and research typology related to CT.Since CT is a new and rapidly developing concept, it is important to examine the research conducted at more frequent intervals.In addition, it is necessary to examine the relevant research in a wider framework and to reveal the common points of the practices adopted in them.This research evaluates the development approaches adopted, the learning tools used, the sub-skills addressed, the research groups applied, the measurement tools used, and the results obtained regarding CT.In this context, we attempted to answer the following research questions: • RQ1: What are the development approaches adopted regarding CT?
• RQ2: What are the learning tools used in the development of CT?
• RQ3: How are the research groups in which CT-related applications are made?
• RQ4: What are the sub-skills covered in CT?
• RQ5: What are the preferred measurement tools for the evaluation of CT?
• RQ6: What are the prominent results pertaining to the research on CT?

Method
In this research, we examined the practice-based research (experimental, case) published on CT between 2019 and 2021 and analyzed the content of the research.We did not include survey research, scale development and systematic review research.Content analysis is an effective method for summarizing, classifying, comparing, and expressing the findings obtained in research numerically (frequency, percentage, mean, and alike) (Chen, Monion, & Morrison, 2007).

Sample
We searched Science Direct, Google Scholar and Web of Science article indexes in order to reach academic publications on CT.We used the keyword "Computational Thinking" in these indexes.By filtering according to the years 2019 and 2021, we accessed the article contents published in full text.In accordance with the purpose of the research, we conducted content analysis of a total of 97 articles scanned in SSCI, ESCI and ERIC indexes.

Data collection tool
We examined the topics covered in the content analyzes related to CT (Haseski & İlı̇c, 2019;Kalelioğlu, 2018;Tang, Chou, & Tsai, 2020).We determined the main elements of the publication classification form according to the development approaches, sub-skills and measurement tools that emerged in these studies.We examined the research that made content analysis using the publication classification form (Çiltaş et al., 2012; İslamoğlu et al., 2015).We created the publication classification form using the Excel spreadsheet application (Appendix-A).The form was reviewed by another researcher conducting research in the field of computer and instructional technologies, and necessary arrangements were made.In the first part of the form, descriptive information such as article title, article year and article index is included.In the other sections, there are topics related to development approaches, learning tools, sub-skills, research groups, measurement tools and results.

Data Analysis
According to the sampling criteria, the researcher first made the first classification of 97 articles with the publication classification form.Then, 10 publications randomly selected from the bulk of research were sent to another male researcher.The classification consistency of two researchers over 10 articles was compared.As a result of the evaluation, we determined the reliability of the classifications as 85%, according to the Huberman and Miles (2002) safety level formula (reliability=consensus/(consensus+disagreement)).We calculated the consistency of the headings in the classification form separately, and in general, the consistency of all categories was over 75%.The fact that this rate is over 70% indicates that the coding is consistent, and the research is reliable (Huberman & Miles, 2002).By transferring the data obtained from the classification form to the SPSS statistical program, we revealed the frequency values and percentiles of the codes.We transferred the results of the research to a word processing program and carried out the coding on the Nvivo software.The first researcher coded.Then the second researcher made the coding.A common code list was created by comparing the resulting codes.Both researchers recoded according to this code list.Cohen's Kappa reliability coefficient of consistency between coding was calculated as 0.80.The resulting value indicates that the agreement between encoders is at a good level (Landis & Koch, 1977).

Results
In this research, we examined 16 (16%) practice-based research conducted in 2019, 31 (32%) conducted in 2020, and 50 (52%) conducted in 2021.Most of the reviewed articles (76%) were published in SSCI indexed journals.Other articles (24%) were published in ESCI and ERIC indexed journals.The publication of articles mostly in SSCI indexed journals indicates that serious research has been carried out on CT and that the importance given to CT is gradually increasing.

Approaches Adopted for the Development of CT
We report the different approaches used to enable the development of CT (Figure 1).

Learning Tools Used for the Development of CT
In the research, different coding interfaces and learning tools were preferred in teaching programming and robotics (Table 1).Block-based coding tools have been widely used in programming and robotics education.While Scratch is used for programming education in almost all age groups, it is used more in primary and secondary school student groups.In robotics education, mBot, Mindstorms and Arduino robotic kits, which can be coded with a block-based coding interface, are generally used.In text-based programming education, Arduino robotic kits are mostly preferred.Bee-Bot is partially used in small age groups such as pre-school and primary school.

Research Groups Focused on CT Research
The CT development approaches shown in Figure 1 were applied to different age groups at different levels of education (Figure 2).

Sub-skills Addressed in CT Research
In research using different development approaches, the sub-skills were differentiated (Table 2).

Preferred Measurement Tools for Evaluation of CT
The measurement tools used in the research are shown in Figure 3.

Prominent Results in CT Research
Content analyzes of 97 research results conducted based on the application were made.The variables having a positive effect on the development of CT skills were coded and their frequency values were extracted (Figure 4).N=7) are more than studies showing that it is not effective (N=3).After the applications, it was revealed that boys (N=5) showed more improvement than girls (N=2).

Discussion
It has been observed that the articles published between 2019 and 2021 are concentrated in SSCI indexed journals.Educators, policy makers, and practitioners are increasingly recognizing the importance of CT (Saxena, Lo, Hew, & Wong, 2020), and both efforts to include computer science-related fields (programming, robotics) in curricula (Bocconi et al., 2016) and the increase in STEM applications in schools (Rich, Yadav, & Larimore, 2020) provide clues that the importance given to CT will increase.
Research on CT has generally included programming, STEM, robotics, development courses and CS Unplugged activities.Weinberg (2013) stated that CT could be developed through applications such as unplugged activities, block-based programming, game design, educational robotics and STEM.In recent years, researchers have generally preferred these approaches for the development of CT.Programming is a popular practice used to teach CT concepts, practices, and perspectives (Brennan & Resnick, 2012;Lye & Koh, 2014;Román-González et al., 2017).In addition, robotics plays an important role in the development of students' problem solving, creative thinking (Karim, Lemaignan, & Mondada, 2015), learning programming (Numanoğlu & Keser, 2017), STEM and CT skills (Becker, Cummins, Davis, Freeman, Hall, & Ananthanarayanan, 2017).In some courses, CT concepts are integrated into STEM applications (Weintrop et al., 2016).CS Unplugged activities are preferred for beginners and younger age groups to gain programming logic and develop CT skills (Bakala, Gerosa, Hourcade, & Tejera, 2021;del Olmo-Muñoz et al., 2020).Apart from these studies, there are also CT development courses for pre-service and in-service teachers in the articles.Since CT is a new concept, it is stated that teachers have some difficulties in teaching and integrating this concept into their courses (Hsu et al., 2018;Mouza, Pan, Yang, & Pollock, 2017;Yadav et al., 2017).For this reason, in recent years, there has been an increasing number of courses organized on what the scope of CT is and how to teach it to pre-service and in-service teachers (Umutlu, 2021).
The inclusion of different definitions of CT in the literature causes different views on the scope of CT (Weintrop et al., 2016).CSTA and ISTE (2011) emphasized the 9 sub-skills of CT; data collection, data analysis, data representation, problem decomposition, abstraction, algorithms and procedures, automation, parallelization, and simulation.Rose, Habgood and Jay (2017) stated that components such as abstraction, algorithms, data, problem decomposition, parallelism, debugging, testing and control structure were the most commonly used sub-skills.This research has shown that sub-skills such as algorithmic thinking, decomposition, abstraction, and debugging are generally considered in common.Although Kalelioglu, Gulbahar, and Kukul (2016) expressed the full disclosure of the definition and scope of CT as a difficult goal, researchers and educators expressed abstraction, algorithmic thinking, decomposition and debugging as common skills reflecting CT (Barr & Stephenson, 2011;Grover & Pea, 2013;Lee et al., 2011;Rich et al., 2020;Yadav, Larimore, Rich, & Schwarz, 2019).Selby and Woollard (2013) reviewed CT research conducted between 2006 and 2013 to contribute to the debate about the definition of CT.They stated that the terms abstraction, decomposition, algorithmic thinking, evaluation and generalization are skills that are frequently used in definitions.Angeli et al. (2016) stated that for the CT development of primary school students, the skills of stripping, generalizing, decomposition, algorithmic thinking and debugging should be included in the curriculum.The common CT subskills demonstrated by the results of this research are similar to the CT subskills mentioned earlier in the literature.In addition to CT basic skills, the concepts, practices, and approaches model introduced by Brennan and Resnick (2012) were used in block-based programming activities.While Kong (2016) stated that this framework covered CT comprehensively, Nouri, Zhang, Mannila, and Norén (2020) stated that this framework provided a theoretical basis for the block-based visual programming language.In programming and robotics education, tools with a block-based coding interface are generally preferred.Scratch has become a software frequently used by all age groups.However, it is widely preferred in primary and secondary school levels.Scratch, which can work on mobile devices with different operating systems apart from computers, is used in 150 different countries and more than 60 languages around the world (Scratch, 2021).The Scratch program was used for CT development in previous years and positive results were obtained after the applications (Brennan & Resnick, 2012;Moreno-León et al., 2015).In robotic programming applications, mBot, Lego Mindstorms, Arduino and Bee-Bot robotic kits, which could be coded as block-based, were used.Arduino, which can also be coded as text-based, is preferred because it is an open-source software and easy for use, is open to access to different projects and many different electronic parts, especially different sensors (Arduino, 2021).Lego Mindstorms' NXT and EV3 series are among the robotics kits that are widely used in different education levels (Oluk & Korkmaz, 2018;Üçgül, 2018).Since mBot Scratch can be used with a coding interface, it can be used easily in primary and secondary school age levels, while Bee-Bot is mostly used in preschool student groups because it has an icon-based visual coding interface without the need for reading and writing knowledge.
With the increasing importance of CT, research has focused more on primary and secondary school students.While the research conducted by organizing courses for teachers to introduce CT is increasing, the research conducted for pre-school students is quite limited.Buitrago Flórez, Casallas, Hernández, Reyes, Restrepo and Danies (2017) state that CT should be taught to students at an earlier age in order to ensure their cognitive development.There are studies supporting that students can acquire CT skills at primary school level (Hsu et al., 2018;Lye & Koh 2014;Shute, Sun, & Asbell-Clarke, 2017).Djurdjevic-Pahl, Pahl, Fronza and El Ioini (2017) stated that CT could be taught at an early age starting from the first year of primary school.Due to the fact that programming learning and teaching activities are given more place at the university level, the research carried out in previous years mostly focused on the undergraduate level (Kalelioğlu, 2018).With the increase in programming activities carried out with primary and secondary school students in recent years, it is thought that CT research will be frequently included in these levels (Nouri et al., 2020).Although some studies emphasize the importance of CT development in early childhood (del Olmo-Muno et al., 2020;Djurdjevic-Pahl et al., 2016;Espino & González, 2016), the results show that studies on preschool are rather limited compared to others.As the importance of CT becomes better understood over time, it is anticipated that future research will tend towards these age groups.
In recent years, studies revealing the effect of the gender factor on the development of CT have been increasing.It is stated that the CT skills of boys develop more than girls (Esteve-Mon et al., 2020;Polat et al., 2021;Mouza et al., 2020).There are also studies showing that gender is not effective in the development of CT (Alsancak, 2020;del Olmo-Muñoz et al., 2020).When students start secondary school, gender differences and inequalities in prior knowledge and experience with computers become more evident (Ardito, Czerkawski, & Scollins, 2020;Witherspoon et al., 2017).The fact that male students have a slightly higher interest in computer science and technical subjects than female students may cause this situation (Polat et al., 2021).
There are also different views on how to evaluate CT.In this study, it was revealed that CT was evaluated mostly through scales and tests consisting of multiple choice and open-ended questions.It is seen that interview forms have been frequently used to support the data obtained with these data collection tools.These forms are also used to reveal the knowledge of pre-service and in-service teachers about CT (Li, 2021;Nouri et al., 2020).In addition to interview forms, observation forms are also used to monitor and evaluate students' activity processes (Herro et al., 2021;Mouza et al., 2020).Researchers state that there is uncertainty about how to evaluate CT (Lockwood & Mooney, 2018;Hsu et al., 2018).In her research, Yeni (2018) stated that students benefited from writing code, tests, observation, code block interpretation, ordering code blocks and open-ended questions to determine their level of understanding of programming terms depending on CT.While tests consisting of multiple choice and open-ended questions were used frequently in previous research (Atmatzidou & Demetriadis, 2016;Chen et al., 2017;Grover et al., 2014), scales for different student levels were also developed by researchers (Korkmaz et al., 2017;Kukul & Karataş, 2019;Yağcı, 2019;Kılıç et al., 2021).Especially in block-based programming environments such as Scratch, software such as Dr Scratch and Scrape are also used to evaluate the code sets quickly created by the students and to provide quick feedback to the students (Moreno-León et al., 2015;Wolz, 2011).Kalelioğlu (2018) also stated in his content analysis study that CT scales and tests were the most used data collection tools.While it is seen in the literature that tests and scales are mostly used in the evaluation of CT, it can be said that similar evaluation methods are also preferred in new studies.

Conclusion
In recent years, serious practice-based research on CT has been carried out and the importance of CT is better understood.While discussions on the definition, scope, development, measurement, and evaluation of CT still continue, the results of this research show that there is no consensus on this point.While CT development is evaluated within the framework of many different sub-skills, abstraction, decomposition, algorithmic thinking and debugging skills are considered common in different development approaches.It is important to address and evaluate these sub-skills primarily in research.The use of tests consisting of open-ended or multiple-choice questions that relate the subjects to daily life or the use of scales with validity and reliability are seen as effective measurement tools in the evaluation of CT developments.Although disagreements are not seen as an obstacle to the progress of CT, teachers who work at different levels, starting from pre-school, have important responsibilities to better understand CT.At this point, it is important to increase the development courses given to teachers within the scope of programming, robotics, STEM and CS Unplugged in the coming years in order to accelerate the integration of CT into courses.We especially attach importance to the fact that teachers related to the field of computer science have content knowledge in the use of Scratch, Lego Mindstorms, Arduino, mBot or other different programming and robotic learning tools.Research on pre-school student groups is limited, and at this point, it is necessary to increase practice-based research.Since gender differences appear to be influential in the development of CT, teachers need to take these differences into account in their classroom practices.

Figure 1 .
Figure 1.Approaches used to development of CT

Figure 2 .
Figure 2. Research groups focused on CT research

Figure 3 .
Figure 3. Measurement tools used in the research

Figure 4 .
Figure 4. Approaches that positively affect CT development

Table 1 .
Coding interfaces and learning tools used in programming and robotics education

Table 2 .
Brennan & Resnick (2012)hin the framework of CT development approaches Algorithmic thinking, decomposition, abstracting and debugging skills are among the skills that were examined more.Concepts, practices, and approaches model introduced byBrennan & Resnick (2012)was used in programming activities.Concepts and Sequencing skills revealing the conceptual structure in programming and robotics applications, were examined.