1. Introduction
As entering the era of the Fourth Industrial Revolution, software education has taken an important position. Students are required to understand software coding to prepare for the future, and they have to improve the level of computationalthinking abilities through software education to be a member of software society. Computational thinking is consideredessential as a basic skill taught in schools along with 3Rs which are reading, writing, and arithmetic. Therefore, it is obvious that improving the level of computational thinking abilities is important.
Many universities in Korea offer software coding classes that include computational thinking in curriculums. Some of them have been designated as mandatory courses for every student. However, it is not easy to take coding classes for students majored in humanities or social sciences. Since the students are having difficulties in understanding the coursematerials, they think negatively on either software or computational thinking. Such dissatisfaction on courseslowers the effectiveness of the education as well as theachievement of the courses[1]. Negative attitude towards of tware courses affects the students to establish negative perceptions for software. It is very important for students who are future workers to have positive perceptions about the software because the bad perceptions can make a badimpact when working in the future.
We need an study to let students have positive cognition on software education. If students have the positive cognition on software education, it would improve the students’ levelof computational thinking abilities and it can make students have open minds toward software. In the near future, when the students work as a member of software centered society, open minds toward software and effect of the education can make students have a confidence on dealing with software.
In this study, three factors for positive cognition of software education were used as independent variables for regression analysis, and the improvement of computational thinking was used as a dependent variable for the research model. The three independent variables for positive cognition on software education were ‘SWImprvMe’, ‘NeedSW’, and ‘ SWMkBetter & rsquo;. After confirming that dependent variable was dependent on independent variables, we confirmed that moderator variables have a moderating effect on therelationship between independent variables and a dependent variable. The survey was conducted for three years to findout the moderator variables. We collected the questionnaireans wer from all 928 students who took the software coding classes. As a result of the questionnaire, three moderatorvariables could be applied to the research model. The threemoderator variables were ‘SmllGrp’, ‘LabRelated’, and ‘MajorRel’.
2. Research Background
2.1 Improvement of CT
A survey was conducted for three years and the 928 students who took software coding classes wereparticipated for the questionnaire. The result of questionnaire showed that computational thinking (CT) scores were higher when they had a better positive cognition on software.
(Table 1) The Result of the Survey
The composition of the survey participants was as shown in Table 1. 311 students were participated at the first year, 220 students at the second year, and 397 students at the third year. Hence, the total participants were 928 students.
Figure 1 indicates that not only the cognition on (Figure 1) Better Cognition, better CT software was improved every year, but the scores of the computational thinking were also getting higher.
The first question related positive cognition on s of twareeducation was ‘Do you need software education even thoughyou are majored in humanities or social sciences?’ At the first year, only 180 students out of 311 that is 57.9% of therespondents answered positively. The second year wasimproved up to 69.8% of the respondents as positive ans wer, and the third year hit 79.8% as positive response.
The second question was ‘Do we think learning s of tware that enhances the thinking abilities can improve yourself?’ In this question, the mentioned thinking abilities are meant to be comprehensive and generally refer to the computationalthinking that is currently receiving intensive attention. At the first year, 211 students that is 69.8% showed positiveresponse. At the second year, 80.9% of the respondents ans wered positively; and 83.9% of the respondents ans wered positively at the third year.
The third question was ‘Do you think learning s of twarecan make better software centered society?’ At the first year, 233 students that is 74.9% showed positive response. At thesecond year, 85% of the respondents answered positively; and 95.0% of the respondents answered positively at the third year.
The last question was ‘Are you certain that your level of computational thinking abilities is improved in many aspects?’ At the first year, 224 students that is 72.0% showed positive response. At the second year, 85.9% of therespondents answered positively; and 97.0% of therespondents answered positively at the third year.
2.2 Assessment for Computational Thinking
In Section 2.1, the question of whether the level of computational thinking abilities improved is needed to beverified. Hence, there were the questions to test computational thinking abilities in the survey. Thequestionnaire items about computational thinking werecomposed with 11 different areas[2,3,4,5,6,7,8,9].
Figure 2 shows the result of the radiation graph for computational thinking abilities. In Figure 2, the innermostradiograph shows the score of computational thinking abilities before taking no software coding class. The middleradiation graph in Figure 2 corresponds to the scores of computational thinking abilities of students who took one basic software coding class. The outermost radiation graph corresponds to the scores of computational thinking abilities of students who had taken two mandatory software coding classes.
(Figure 2) Improvement of Level of CT
2.3 Descriptive Questionnaire
The 397 students who participated the third year of questionnaire survey were conducted a descriptivequestionnaire on what they wanted for software coding classesto improve their computational thinking abilities. We analyzed the students' answers and found the main keyword.
The most highly mentioned keyword was ‘Small Class. ’ The students wanted a small group class instead of alarge-scale class. If there are too many students, students haddifficult time to pay attention on the class.
The second most mentioned keyword was ‘Related Lab. ’It seemed that students had hard times if the theoretical classand the practical lab were not related. Even though sometheoretical classes were not available to design a properrelated practical lab, students wanted to see the connection between the theoretical class and the practical lab.
The third frequently mentioned keyword was ‘ Major Related. ’ Instead of learning the famous problem, students wanted problems which were related to their major for problem-based learning. Since the students took a class because it is mandatory for the graduation, they wanted useful problems for lab assignment rather than common problems.
Other keyword were ‘More Example,’ ‘Class Time,’ and ‘ Textbook. ’ We only selected the top 3 from the keyword to apply as the moderator variables to improve the computational thinking abilities.
(Figure 3) Keyboards for Better Class
3. Literature Review
3.1 Software Literacy
A researcher studied on providing a smart learningenvironment to improve the efficiency of education[10]. Toprepare for the education in the era of the Fourth Industrial Revolution, applying high technology in the educational fieldis efficient than the traditional learning process to improvelearning effect. To get advantage of using high technology in the classroom, students need to understand the concept of software because many class materials supplied as s of twareproducts. For better education and learning, it is obvious thatevery one involved in the education should be aware of software literacy.
There is a growing need for software education such as computational thinking education for convergence with otherscholarships in universities. It is due to the advent of the 4th Industrial Revolution[11]. Software literacy is an imperative to understanding everyday 21st century contexts where the focus is on software. Up to high school, students can only be a consumer of software, but when students enter anuniversity, students must be a consumer and also a producer. To produce a software, students must learn the basicconcepts of software and practice software coding.
3.2 Computational Thinking
Information technologies are the base of the worldinfrastructure. Nowadays, the effort has been oriented mainly to convert students into users of computer tools. This has gone from being necessary to being insufficient, because the use of software applications means to manage a digitallanguage that is obsolete in a time that is not proportional, in effort, to the time that has been invested in acquiring theseskills. That is, instead of teaching students only the syntaxof a changing language, they should be instructed in the rules that allow them to know how the digital language is constructed. Thus, computational thinking emerges as aparadigm of work, and the programming is established as the tool to solve problems[12].
Improving the computational thinking abilities is essential for everyone not just for understanding digital language. Computational thinking will be a fundamental skill, which can be used by everyone in future to strengthen and enhance learning[13]. It is a thought process that involves formulating problems so that solutions can be represented as computational steps and algorithms. In other words, computational thinking is problem solving ability, not simply software understanding.
3.3 Cognitive-Based Education
Existing programming education often leads to learner' scognitive burden. The traditional programming education is based on the programming method of teaching or project-based method in which students are required to findsolutions depending on their ways of thinking and someguidance by teachers. The programming ability can be improved by various factors. The emotional intelligence is a concept that can complement existing cognitive-basededucation [14]. When a learner participates a programming class with positive cognitive learning, the programming ability can be empowered.
The primary missions of educational institutions, fromelementary to graduate and professional schools. are to impartknowledge and to teach cognitive skills. One of the most important cognitive skills is no doubt problem-solving ability. Problem solving is of course predominantly involved in basic courses such as mathematics and science and in professional areas such as medicine. engineering, andarchitecture. But problem solving pervades almost allareas of instruction; reading and writing have important problem-solving components, for example. Even such arudimentary process as retrieving information stored in long-term memory can be viewed as a problem-solving activity [15]. Hence, learning for problem solving is bond with cognitive skills.
4. Method
4.1 Research Model
After analyzing the three-year survey, how students think about learning software affects improvement of computational thinking. The results of the analysis proved when the students had a positive cognition on purpose for the software coding class, the level of computationalthinking abilities is improved more than others who had a negative cognition. Therefore, we adopted the improvement of computational thinking as a dependent variable, and applied factors of cognition on s of twareeducation as independent variables. Figure 4 presents ourresearch model, and it has independent variables part, dependent variable, along with moderator variables.
The independent variables for cognition on s of twareeducation are ‘NeedSW,’ ‘SWImprvMe,’ and ‘ SWMkBetter & rsquo; those are explained as the research background in Section 2.1.
For statistical analysis, IBM SPSS package is used. Through the statistical analysis, we verified that the proposed research model was correctly suggested.
(Figure 4) Moderator Model
4.2 Multiple Regression Analysis
The multiple regression is applied to check therelationship between several independent variables and adependent variable. Three independent variables and onedependent variable were applied for the multiple regression analysis for the research model as Figure 5.
(Figure 5) Multiple Regression
As Figure 5, all requested independent variables areentered for the dependent variable ‘CTImprv.’
4.3 Moderator Model Analysis
In statistics and regression analysis, moderation occurs when the relationship between two variables depends on a third variable. The third variable is referred to as the moderator variable. For the suggested research model, we applied 3 variables as moderator variables as shown in Figure 6.
The rationale for the selection of moderator variables is described in Section 2.3. The moderator variables are & lsquo; Smll Grp, ’ ‘LabRelated,’ and ‘MajorRel.’
(Figure 6) Moderator Variables
5. Result
The results of the statistical analysis using SPSS are as follows.
(Figure 7) Model Summary for R2
The correlation between the independent variable and the dependent variable is highly correlated as 0.455 as Figure 7. The value of adjusted R square is 20.7%, and it shows the total explanatory power of dependent variable of independent variables. The value of Durbin-Watson is 2.040. Since the numerical value of Durbin-Watson is close to 2 and not close to 0 or 4, there is no correlation between the residuals; hence, the regression model is interpreted as appropriate.
(Figure 8) F-test for Research Model
The linear regression’s F-test has the null hypothesis that there is no linear relationship among the variables. With F= 34.131 and 396 degrees of freedom the test is highlysignificant, thus we can assume that there is a linearrelationship between the variables in our model as shown in Figure 8.
In Figure 9, the linear regression’s t-value for the influencing relationship of ‘NeedSW’ toward ‘CTImprv’ is 2.943 which satisfies the condition over ±1.96, and significance probability is .003 that satisfies p < 0.05. Therefore, ‘CTImprv’ is dependent on ‘NeedSW’ in ourmodel. The t-value of ‘SWMkBetter’ toward ‘CTImprv’ is 2.449 which satisfies the condition over ±1.96, and significance probability is .015. Therefore, ‘CTImprv’ is alsodependent on ‘SWMkBetter’ in the research model.
(Figure 9) t-value for Research Model
(Figure 10) Histogram & Plot
(Figure 11) Moderator Model Analysis
The t-value of ‘SWImprvMe’ toward ‘CTImprv’ is 2.319which satisfies the condition over ±1.96, and significanceprobability is .021 that satisfies the condition p < 0.05. Therefore, ‘CTImprv’ is also dependent on the last moderatorvariable ‘SWImprvMe’ in the proposed research model.
The last thing we need to check is the homosced asticity and normality of residuals. The histogram indicates that theresiduals approximate a normal distribution. The plot shows that in our linear regression analysis there is no tendency in the error terms as shown in Figure 10.
The result of moderator model analysis is shown in Figure 11. Since the significance F change is .000, which is less than 0.05, it can be judged that there is a moderating effecton our model. It shows that ‘Smllgrp’, ‘LabRelated’ and ‘ Major Rel & rsquo; are moderator variables for ‘CTImprv’ which is the dependent variable on the research model.
6. Discussion
This study started with the hypothesis that the positive cognition on software education improves the effectiveness of getting computational thinking abilities. For three years, the questionnaires were conducted, and the total of 928 students who took software coding classes answered the questionnaires. The result of the questionnaires were analyzed to verify the hypothesis.
As the software coding classes are repeatedly opened for the mandatory courses, students' perceptions about the classes have changed positively and gradually agreed to thenecessity of software education because of emphasis on the importance of software in various media. As the result, it derive the positive cognition on software education. Having positive attitude toward the class makes the students accomplish the better educational effect. The multiple regression analysis was processed to prove the hypothesis, and it has proven that the improving computational thinking abilities is dependent on the positive cognition on s of twareeducation.
Furthermore, to find out additional factors that canimprove computational thinking abilities even more, we proposed a moderator model in the research model. In order to confirm the moderator variables, 397 students were asked to write a descriptive questionnaire. Then three moderatorvariables with the most frequent responses were selected. Thru statistical analysis, the selected 3 moderator variables were determined that they have moderating effects on the proposed research model.
This study concluded that positive cognition on s of twareeducation can make the better improvement of computationalthinking within proper moderator variables. Nevertheless, it should be noted that there are variables that can not be used in general situation, since this study is limited to specificcircumstances of our institution.
References
- O. Han and J. Kim, "Examining the relationship between educational effectiveness and computational thinking in smart learning environment", Journal of Internet Computing and Services(JICS), Vol. 19, No. 2, pp. 57-67, 2018. https://doi.org/10.7472/jksii.2018.19.2.57
- J. Wing, "Computational thinking", Communications of the ACM, Vol. 49, No. 3, pp. 33-35, 2006. https://doi.org/10.1145/1118178.1118215
- S. Jacob and M. Warschauer, "Computational thinking and literacy", Journal of Computer Science Integration, Vol. 1, No. 1, 2018. https://doi.org/10.26716/jcsi.2018.01.1.1
- A. Juskeviciene and V, Dagiene, "Computational Thinking Relationship with Digital Competence", Informatics in Education, Vol. 17, No. 2, pp. 265-284, 2018. https://doi.org/10.15388/infedu.2018.14
- S. Grover and R. Rea, "Computational Thinking in K-12 : A Review of the State of the Field", Educational Researcher, Vol. 42, No. 1, pp. 38-43, 2013. https://doi.org/10.3102/0013189X12463051
- Google for Education. Exploring computational thinking. Retrieved from https://edu.google.com/resources/programs/exploring-computational-thinking/#!home
- D. Barr, J. Harrison, and L. Conery, "Computational Thinking: A Digital Age", Learning and Leading with Technology, ISTE(International Society for Technology in Education), pp. 20-23, 2011. https://files.eric.ed.gov/fulltext/EJ918910.pdf
- D. Weintrop, E. Beheshti, M. Horn, K. Orton, K. Jona, L. Trouille, and U. Wilensky, "Defining computational thinking for mathematics and science classrooms", Journal of Science Education and Technology, Vol. 25, No. 1, pp. 127-147, 2016. https://doi.org/10.1007/s10956-015-9581-5
- K. Boom, M. Bower, A. Arguel, J. Siemon, and A. Scholkmann, "Relationship between computational thinking and a measure of intelligence as a general problem-solving ability", ITiCSE 2018 Proceedings of the 23rd Annual ACM Conference on Innovation and Technology in Computer Science Education, pp. 206-211, 2018. http://delivery.acm.org/10.1145/3200000/3197104/iticse18main-id73-p.pdf?ip=115.145.32.105&id=3197104&acc=ACTIVE%20SERVICE&key=0EC22F8658578FE1%2EB50D9BE1468BDDBD%2E4D4702B0C3E38B35%2E4D4702B0C3E38B35&__acm__=1557318565_92eb52c0b66db82eff07b70015b2bfad
- O. Han and J. Kim, "A Study on the Technology Utilization for Smart Education in the 4th Industrial Revolution Era", Journal of Internet Computing and Services(JICS), Vol. 19, No. 4, pp. 71-82, 2018. https://doi.org/10.7472/jksii.2018.19.4.71
- W. S. Kim, "A Study on the Recognition of Freshman on Computational Thinking as Essential Course", Culture and Convergence, Vol. 39, No. 6, pp. 141-170, 2017. http://www.dbpia.co.kr/Journal/PDFViewNew?id=NODE07294178&prevPathCode= https://doi.org/10.33645/cnc.2017.12.39.6.141
- F. J. Garcia-Penalvo, "Computational thinking", IEEE Revista Iberoamericana de Tecnologias del Aprendizaje (IEEE RITA), Vol. 13, No. 1, pp. 17-19, 2018. http://dx.doi.org/10.1109/RITA.2018.2809939
- R. Mohanty and S. Bala Dasm, "A Proposed What-Why-How (WWH) Learning Model for Students and Strengthening Learning Skills Through Computational Thinking," In Progress in Intelligent Computing Techniques: Theory, Practice, and Applications, pp. 135-141. Springer, Singapore, 2018. https://link.springer.com/chapter/10.1007/978-981-10-33 76-6_15
- Y. Bae and W. Jun, "A Study on Design and Implementation of a Programming Teaching Model Using Emotional Intelligence", Journal of Internet Computing and Services(JICS), Vol. 19, No. 6, pp. 125-132, 2018. https://doi.org/10.7472/jksii.2018.19.6.125
- N. Frederiksen, "Implications of cognitive theory for instruction in problem solving," ETS Research Report Series, 1983(1), pp. 363-407, 1983. https://doi.org/10.1002/j.2330-8516.1983.tb00019.x
Cited by
- The Study on Direction of the Software Education - focused on the freshman students of the College of Social Sciences - vol.21, pp.4, 2019, https://doi.org/10.7472/jksii.2020.21.4.69