Social Network Analysis of College Student Group Selection: Case Study Wasinee Noonpakdee jookjik@gmail.com Thammasat University Thailand Purpose: This research aims to analyse the social network of college student group selection to understand students' relationships and behaviour for effective teaching planning, group work assignments, and individual student management. Study design/methodology/approach: The samples are college students who study a master degree program in digital innovation and take course IO-001(business intelligence and data analytics) or IO-002 (cybersecurity). Data were collected using the questionnaire asking the college students for their 1st and 2nd choices of classmates who want to be in the same group. For data analysis, Gephi, open-source software, is used for social network analysis (SNA) and graph visualisation. Findings: According to the results, the most desirable students to work with and the least desirable students to work with are defined. The instructors can appoint the student who is the most desirable to work with into a position of influence to facilitate learning in the classroom to motivate inspiration during the study. For those students who the least desirable to work with, the instructors might make them gain more attention, including giving them some roles or let them have more identity. According to SNA, the least desirable students to work with have low IT background for both courses IO-001 and IO-002. Therefore, IT background might be one of the factors that influence student group selection. The instructors might provide additional courses to enhance IT skills for those students who have low IT backgrounds. Moreover, the instructors might assign the influent students to work with the least desirable students to make a better relationship. Originality/value: The results of this study can be used as a guideline in the formulation of teaching and learning courses, including setting the learning style to make each student more connected and to encourage each group of students to interact more with each other. The instructors can promote a learning environment to increase opportunities for innovation. Introduction Digital technology had increased and had thoroughly transformed numerous parts of our economic and social life. Different sectors had applied digital technology, including social networks, big data, blockchain, internet of thing (IoT), cloud computing and artificial intelligence (AI), to improve the organisation's performance and to gain a competitive advantage (Y. Chen, 2020; Greco, Percannella, Ritrovato, Tortorella, & Vento, 2020; Mirzaei Abbasabadi & Soleimani, 2021). Many countries anticipate learning skills that could help it correspond to the development of science and digital technology (Hamidi, Meshkat, Rezaee, & Jafari, 2011). With the application of digital technology, the education industry indicated extraordinary opportunities and challenges. Technology promotes the deep integration of education and teaching (L. Chen & Wang, 2019). Therefore, many courses and curricula are conducted to increase students' potential to enter the digital age confidently (Olszewski & Crompton, 2020). To support the role of digital technology and prepare the country for economic and social using digital technology and innovation as the driving force, a case study university offers a master degree program in digital innovation to create added value and international competitiveness and to drive all dimensions of economic goals related to finance production and services. The International Journal of Management, Knowledge and Learning | ISSN 2232-5697 Volume 10 (2021) 149–155 | https://www.doi.org/10.53615/2232-5697.10.149-155 program provides a lot of courses related to digital technologies. In many courses, group project assignment is conducted to help students develop team-work skills that are significant in the digital era. Group project brings social, psychological, and learning benefits to students (Bojanova, 2012). This paper aims to analyse the social network of college student group selection to understand students' relationships and behaviour for effective teaching planning, group work assignments, and individual student management. Social Network Analysis A social network comprises structures made up of actors and their relationships with each other (Can & Alatas, 2019). A social network component is a node that represents a person or thing within a network and the edge or relationship that describes the relationship between the nodes (Radha & Nithia, 2017). Social network analysis (SNA) analyses networked structure leveraging graph theory to explain relationships between individuals, organisations, groups, or whole societies (Hu, Liu, & Zhang, 2014). The analysis of such networks benefits society, and simultaneously it creates a better understanding of society (Radha & Nithia, 2017). Social networking has been used frequently in various areas such as social science, education and computer technology (Can & Alatas, 2019). Social networks influence a broad range of social processes by accessing finance, society, nature, physical and human capital, and information content relevant to them. In addition, social networks can influence development in policies, plans, strategies, projects and collaborations that builds a solid foundation, including design, implementation, and results (Can & Alatas, 2019). For the education sector, researchers have increasingly adopted social network analysis (SNAs) to understand better relationships among students in the entire learning community in conference rooms or networked learning settings such as blended learning, MOOCs, blogs, or online forums. Furthermore, it is found that SNA helped enhance understanding with students' participation in networked learning, as shown in a mathematical method of measuring students' participation through their positions in social work (Liu, Chen, & Diana Tai, 2017). The study in (Liu et al., 2017) showed how elementary school students teamed and collaborated with their peers to create multimedia stories and analysed their participation with SNA. In Moreno (1960), the dining table partner choices of the girls' school dormitory were studied. According to Karimi and Matous (2018), the research applies social network analysis to map students' social activities. Leveraging large-scale data on students' participation in social activities at a major public university demonstrated how to adapt social network analysis to visualise the student's overall social structure and explore a mixed and separate social group at the core and the scope of the student's extracurricular activities. SNAs have recently been used to study, understand and visualise relationships in social networks (Radha & Nithia, 2017). For instance, a social structure can be represented as a network made up of groups of nodes and a set of links that represent connections between them (Can & Alatas, 2019). To better understand social networks, the visualisation of large graphs has been applied to utilise human perception capabilities to find features in network structures and information (Bastian, Heymann, & Jacomy, 2009). Analysing social networks encourages people and organisations to make a tremendous opportunity to the advent of constant SNA in the existing state of development. Several SNA toolsets such as UCINET, Pajek, SocialAction, and NodeXL (Hansen et al., 2012), and various programming languages such as R programming language (Maddumage & Dhanushika, 2018). Noonpakdee | Social Network Analysis of College Student Group Selection: Case Study 150 Gephi is open-source software for graph and network analysis. The software provides modules that can import, filter, visualise, deploy and export all types of networks. It can handle a vast network built on the multi-task model and multi-core processors. Node design can be customised to be a texture, a panel or a photo. The algorithms can be run in real-time with a configurable layout. For example, the Force Atlas algorithm, a specific force-directed algorithm developed by the team, offers speed, gravity, repulsion, auto- stabilise, inertia or size-adjust with real-time settings (Bastian et al., 2009). Research Methodology The research process comprises three phases: literature review, data collection and analysis, and conclusion, as illustrated in Figure 1. Figure 1: Research Process Literature review comprises the study of social network analysis (SNA), application, and tools. For data collection, similar to the study of Moreno (1960), the questionnaire was distributed among college students in course IO-001 and IO-002 for their 1st and 2nd choices of classmates who wanted to be in the same group. Table 1: Attribute Description Attribute Type Description Student Nominal Student name Batch Nominal A group of students who enter the college in the same semester. Gender Binary Male or Female IT background Nominal Students' background in IT which is Low, Medium, and High. GPA Ordinal GPA range which is: 3.00-3.25 3.26-3.50 3.51-3.75 3.76-4.00 Table 1 illustrates some attributes of the collected data: student name, batch, gender, Information technology (IT) background, and GPA. For IT background, the data can be categorised as High, Medium, and Low: High = student has an education background in computer science, computer engineering, information technology (IT), or has work related to IT technical, computer, or programmer. Medium = Student has an education background in management information systems (MIS), Computer management, computer graphic, or has work related to IT management, digital marketing, or graphic design. Low = student has an education background in other fields such as business, management, journalism, or laws. Noonpakdee | Social Network Analysis of College Student Group Selection: Case Study 151 For data analysis, Gephi, open-source software for graph and network analysis (Bastian et al., 2009), is used for SNA and graph visualisation. Graphs are usually laid out with “Force-based” algorithms, which follow a simple principle: linked nodes attract each other, and non-linked nodes are pushed apart. Furthermore, the “Adjust by Sizes” option is applied to avoid node overlapping, depending on the size of each node (Gephi, 2011). Finally, the conclusion is presented to summarise the results. Results The attributes of the collected data are shown in Table 2 and Table 3. Table 2 illustrated students' detail of course IO-001, and Table 3 presented students' detail of course IO-002. IO- 001 course is related to business intelligence and data analytics. For IO-002, the course is about cybersecurity, Data Protection, and Cryptography. Table 2: Students’ Details of Course IO-001 Student Batch Gender IT-Background GPA A X Female Low 3.00-3.25 B X Female Medium 3.51-3.75 C X Male High 3.51-3.75 D X Female High 3.51-3.75 E X Female Low 3.26-3.50 F X Female Medium 3.26-3.50 G X Male High 3.26-3.50 H X Male Medium 3.00-3.25 I X Male High 3.00-3.25 As shown in Table 2, Students who enrolled in IO-001 are in batch X. The genders are 4 males and 5 females. 4 students have high IT background, 2 students who have medium IT background, and 2 students have low IT background. For GPA, there are 3 students whose GPA is 3.00-3.25, 3 students whose GPA is 3.26-3.50, and 3 students whose GPA is 3.51-3.75. From Table 3, 18 students enrolled in IO-002, 6 students from batch Y, and 12 students from batch Z. The genders are 7 males and 11 females. 2 students have high IT backgrounds, 4 students who have medium IT background, and 12 students who have low IT background. For GPA, there are 4 students whose GPA is 3.26-3.50, 8 students whose GPA is 3.51-3.75, and 6 students whose GPA is 3.76-4.00. Table 3: Students’ Details of Course IO-002 Student Batch Gender IT-Background GPA A Z Female Low 3.51-3.75 B Z Male Low 3.26-3.50 C Y Male Low 3.51-3.75 D Y Female Low 3.51-3.75 E Z Male Low 3.76-4.00 F Y Male Medium 3.26-3.50 G Z Male Medium 3.51-3.75 H Z Male Low 3.26-3.50 I Z Female Low 3.76-4.00 J Z Female High 3.76-4.00 K Z Male Low 3.76-4.00 L Z Male Medium 3.51-3.75 M Y Male Medium 3.51-3.75 N Z Female Low 3.26-3.50 O Z Male Low 3.51-3.75 P Y Female Low 3.76-4.00 Q Z Female Low 3.76-4.00 R Y Male High 3.51-3.75 Noonpakdee | Social Network Analysis of College Student Group Selection: Case Study 152 For the analysis using SNA, the directed graph of students who enrolled in course IO-001 is illustrated in Figure 2, and the graph of students who enrolled in course IO-002 is displayed in Figure 3. From Figure 2, student D is the most desirable to work with among all the students, followed by student F. On the other hand, Student A is the least desirable to work with. Regarding attributes, student D has high IT background, and GPA is 3.51-3.75, and student F has medium IT background and GPA is 3.26-3.50. On the other hand, student A has low IT background, and GPA is 3.00-3.25. Figure 2: Student Network of course IO-001 From Figure 3, student C is the most desirable to work with among all the students, followed by students A, Q, D, and O. On the other hand, Student H is the least desirable to work with. Regarding attributes, student C has low IT background and GPA is 3.51-3.75, while student H has low IT background and GPA is 3.26-3.50. Students I, J and L are in the same group as they selected each other's, and this group is isolated from different groups of students. The group is connected by student E, who selected student L as the second choice. Students I, J and L are in the same batch (batch Z), but they had different IT backgrounds. Figure 3: Student Network of course IO-002 Noonpakdee | Social Network Analysis of College Student Group Selection: Case Study 153 From SNA results, the instructors can appoint the student who is the most desirable to work with into a position of influence to facilitate learning in the classroom to motivate inspiration during the study and be a centre for alumni association after graduation. For those students who the least desirable to work with, the instructors might make them gain more attention, including giving them some roles or let them have more identity. In addition, instructors can ask questions about the cause and observe behaviours to find answers. For example, students whose peers do not want to work with might have some problems, have an introverted personality, or do not help their friends when working in a group. According to SNA, of course, IO-001 and IO-002, the least desirable students to work with have low IT backgrounds for both courses. Therefore, IT background might be one of the factors that influence student group selection. Thus, the instructors might provide additional courses to enhance IT skills for students with low IT backgrounds. Moreover, the instructors might assign the influent students to work with the least desirable students to make a better relationship. Furthermore, if any student has a problem in case of emergency, instructors can ask for initial information from students in a similar relationship. The SNA can be used as a guideline in the formulation of teaching and learning courses, including setting the learning style to make each student more connected and encouraging each group of students to interact more with each other. In addition, the instructors can promote a learning environment to increase opportunities for innovation. Conclusion This paper presents a social network analysis of college student group selection. The research is conducted by a questionnaire survey of 9 students in course IO-001 (Business intelligence and data analytics) and 18 students in course IO-002 (Cybersecurity). Data is analysed using Gephi, which is open-source software for graph and network analysis. According to the results, the most desirable students to work with and the least desirable students to work with are defined for both courses IO-001 and IO-002. The results can help instructors understand students' relationships and behaviour to provide effective teaching courses by establishing the learning method to make more connections and persuade each student with more interactions. Moreover, the instructors can apply the results to promote an innovative learning environment of the course. Future research considers this study regarding different attributes, courses, or activities, or other networks such as organisational networks. Furthermore, some network modelling, for example, exponential random graph models (ERGMs), or some properties of actors in the network can be studied to analyse complex social network structures. References Bastian, M., Heymann, S., & Jacomy, M. (2009). Gephi: An Open Source Software for Exploring and Manipulating Networks. ICWSM, 8, 361-362. Bojanova, I. (2012, 3-6 Oct. 2012). Work in progress: Examination of student experiences in immersive vs traditional group projects. Paper presented at the 2012 Frontiers in Education Conference Proceedings. Can, U., & Alatas, B. (2019). A new direction in social network analysis: Online social network analysis problems and applications. Physica A: Statistical Mechanics and its Applications, 535, 122372. doi:https://doi.org/10.1016/j.physa.2019.122372 Chen, L., & Wang, J. (2019, 23-25 Aug. 2019). Talking About the Application of Information Technology in College Aerobics Teaching. Paper presented at the 2019 10th International Conference on Information Technology in Medicine and Education (ITME). Chen, Y. (2020). IoT, cloud, big data and AI in interdisciplinary domains. Simulation Modelling Practice and Theory, 102, 102070. doi:https://doi.org/10.1016/j.simpat.2020.102070 Gephi. (2011). Gephi Tutorial Layouts. Retrieved from https://gephi.org/tutorials/gephi-tutorial-layouts.pdf Noonpakdee | Social Network Analysis of College Student Group Selection: Case Study 154 Greco, L., Percannella, G., Ritrovato, P., Tortorella, F., & Vento, M. (2020). Trends in IoT based solutions for health care: Moving AI to the edge. Pattern Recognition Letters, 135, 346-353. doi:https://doi.org/10.1016/j.patrec.2020.05.016 Hamidi, F., Meshkat, M., Rezaee, M., & Jafari, M. (2011). Information technology in education. Procedia Computer Science, 3, 369-373. doi:https://doi.org/10.1016/j.procs.2010.12.062 Hansen, D. L., Rotman, D., Bonsignore, E., Milic-Frayling, N., Rodrigues, E. M., Smith, M., & Shneiderman, B. (2012, 14-16 Dec. 2012). Do You Know the Way to SNA?: A Process Model for Analysing and Visualising Social Media Network Data. Paper presented at the 2012 International Conference on Social Informatics. Hu, J., Liu, M., & Zhang, J. (2014, 17-20 Aug. 2014). A semantic model for academic social network analysis. Paper presented at the 2014 IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining (ASONAM 2014). Karimi, F., & Matous, P. (2018). Mapping diversity and inclusion in student societies: A social network perspective. Computers in Human Behavior, 88, 184-194. doi:https://doi.org/10.1016/j.chb.2018.07.001 Liu, C.-C., Chen, Y.-C., & Diana Tai, S.-J. (2017). A social network analysis on elementary student engagement in the networked creation community. Computers & Education, 115, 114-125. doi:https://doi.org/10.1016/j.compedu.2017.08.002 Maddumage, C., & Dhanushika, M. P. (2018, 5-7 Dec. 2018). R programming for Social Network Analysis - A Review. Paper presented at the 2018 3rd International Conference on Information Technology Research (ICITR). Mirzaei Abbasabadi, H., & Soleimani, M. (2021). Examining the effects of digital technology expansion on Unemployment: A cross-sectional investigation. Technology in Society, 64, 101495. doi:https://doi.org/10.1016/j.techsoc.2020.101495 Moreno, J. L. (1960). The dining table partner choices of the girls' school dormitory. The Sociometry Reader. Glencoe (II.), The Free Press, 35. Olszewski, B., & Crompton, H. (2020). Educational technology conditions to support the development of digital age skills. Computers & Education, 150, 103849. doi:https://doi.org/10.1016/j.compedu.2020.103849 Radha, D., & Nithia, P. T. (2017, 21-23 Dec. 2017). A Case Study on Social Network Analysis: Thesaurus Book. Paper presented at the 2017 2nd International Conference on Computational Systems and Information Technology for Sustainable Solution (CSITSS). Noonpakdee | Social Network Analysis of College Student Group Selection: Case Study 155