Social Interaction and Academic Performance of Construction Management Students

Social interaction between students is a crucial but under-researched part of the education realm. Understanding how connections form in university classes and their effects on learning outcomes may provide extraordinary knowledge for researchers, educators, and policy-makers. This paper collected data from the questionnaire survey and then processed them with Gephi software to produce visualization and measurement. Initial results seem to indicate a significant correlation between students' connectedness and academic performance in one class. However, in another class, the results show a contrasting situation as there is no evidence that social network attributes impact learning performance. Taken together, these results would seem to suggest that the characteristics of the network should be judged on a case-by-case basis, and large-scale SNA analyses have been rarely reported. This present study provides a springboard for a new way to shed some light on classmates' interconnection. Using a similar approach to this article, it is believed that there is ample opportunity to study the association between classmate connectedness and career success. Research techniques and approaches around Social Network Analysis are expected to evolve further in the foreseeable future.


INTRODUCTION
Social interaction between students is an essential but undiscovered part of the higher education field. Few researchers have addressed whether the social connection has any association with his/her academic performance. Understanding how learning relationships form in college classes and the effects on learning outcomes can provide practical information for educators in unique ways. It could conceivably lead to a foundation for policies and programs for educational innovation. Although this domain is interesting, it suffers from the lack of formal methods for detecting relational patterns. Social Network Analysis (SNA) offers an essential set of tools to explore relationships. Generally, SNA is genuinely a powerful toolkit for research using relational data in a wide range of domains. Particularly in the education sector, SNA has so far proved suitable for a context containing various complex interactions between actors, both human and non-human. This present study introduces the basic concepts in SNA and corresponding data collection, data processing, and data analysis.
SNA is generating considerable interest in terms of cross-disciplinary research collaboration. This practical method has been used successfully for testing and creating models in a wide variety of research fields. Very recently, for instance, food sharing (Harvey et al., 2020), bullying (Rambaran et al., 2020), IJAE Page 378 criminology (Sabon et al., 2021), digital games (Khajeheian and Kolli, 2020). Network analysis covers two significant hypotheses. First, hypotheses seek to understand what affects the formation of relationships in a given set. Second, hypotheses consider the effect of the relationship structure on outcome-shaping, either individually or collectively. More and more research on social connectivity in higher education is being conducted, testing the correlation with multiple variables such as academic performance, alcohol use, or career choices. SNA is a widely-used approach to describe the structure of the links between certain actors and apply quantitative processes to compute metrics to evaluate network-level and node-level properties. However, the social interaction and academic performance of Vietnamese students are still poorly understood. This paper will provide general and detailed perspectives on students' connectedness in two tertiary classrooms in a higher education institution in Hanoi. At the same time, the correlation between network measures and academic achievement will be elucidated.

Network measures in school settings
School is a social setting where young people interact with each other in their unique ways and rules. Usually, students feel comfortable in an environment of peers who empathize with their ideas and concerns (Ortiz et al., 2004). Having claimed that, connections with friends play an essential part in each student's social network and his/her tendency to get better or worse in career and life (Cairns et al., 1988, Farmer andRodkin, 1996). In a seminal article, Grunspan et al. (2014) claim that SNA is the key to helping researchers uncover suspicions or implications for school environment interactions. In recent years, education researchers have put network measures concerning a variety of domains such as academic performance (Putnik et al., 2016, Saqr et al., 2018a, blog interaction Bonk, 2016, Jimoyiannis et al., 2013), international friendship (Rienties and Nolan, 2014), online learning (Saqr et al., 2018b, Ouyang andScharber, 2017). Indeed, the complex interactions between human and nonhuman actors in school settings require a sophisticated tool such as SNA.

Research in the context of construction management class
Construction management (CM), as dissimilar to other fields under the construction umbrella, construction management (CM) is a relatively recent branch of knowledge that embraces both the natural and social sciences (Dainty, 2008). That may be the reason why research on CM education has only flourished in recent decades. As a young field, it is not surprising that the research topics into CM teaching and learning are not diverse. The study of CM education mainly revolves around: curriculum improvement (Ahmed et al., 2014, Lee et al., 2011a), BIM integration (Ghosh et al., 2015, Shanbari et al., 2016, Wong et al., 2011, simulation (Jaeger and Adair, 2012, Glick et al., 2012, Lee et al., 2011b, Jaeger and Adair, 2010, sustainability integration (Lim et al., 2015, Wang, 2009. The CM education studies that refer to a classroom setting or student-involved interaction are even infrequent. Maghiar et al. (2015) examined cognitive-content, affective-personal, and behavioral learning in CM students, analyzing the relationship between course valuing scores, Approaches to Study Inventory (ASI), and course performance. In a similar vein, Farrow and Wetzel (2020) provided an indepth analysis of student experiences within the active learning environment, including a perceived impact on learning and engagement. Davis (2012) and Kim (2018) investigated the fit-for-purpose mobile devices and applications for the learning activity in CM classrooms to keep pace with industrial transformation. Likewise, by drawing on the technological aspect, (Kim et al., 2019) studied student perceptions of the learning experience and compared the two VR viewing options. What is known about CM education in classroom settings is primarily based on student learning experience and technical consideration. However, despite the enormous potential benefit of network study in education, very little is known about human and social relationships within CM classrooms. This lack of research leaves CM educators with insufficient empirical support for understanding their students and other stakeholders.

IJAE Page 380
other, you have a high cluster coefficient and vice versa. Gephi is open-source software for graph and network analysis that is used in this study. Gephi uses the 3D rendering engine to render large networks in real-time and speeds up data discovery. Gephi's flexible structure and multitasking allow users to work with complex data sets and produce visible results with high application value (Bastian et al., 2009). In the construction field, in addition to literature studies (Hosseini et al., 2018;Luo et al., 2018), many studies have used this tool; for instance, Wehbe et al. (2016) map the linkages between resilience and performance of building safety; Hosseini et al. (2020) explore features of corruption risks in Iranian construction projects; Xiong et al. (2018) seek the opinions of leaders regarding safety issues for construction workers; Akgul et al. (2017) examine SNA on Turkish construction companies operating in international markets.

Network connection and academic performance data collection
This study is conducted at a large public university (30,000-34,999 students) in downtown Hanoi. The two classrooms of interest are undergraduate construction management majors. There were 42 students (Class P) and 33 students (Class V) enrolled and finished the questionnaire survey. However, this article only analyzed 41 and 33 students because one student did not have recent semester results (deferment). The male/female distributions are 27 (65.9%)/14 (34.1%) and 22 (67.7%)/11 (33.3%) respectively. It is worth noting that although the credit system organizes the instruction, students remain within the academic year framework at the university. Generally, in class A, the academic-yearbased basis is relatively stable unless a student of class A is enrolled in another class or a student of class B is enrolled in class A (due to re-study or personal preference).
Social networking data was collected using an online self-report survey developed by Qualtrics. The Vietnamese survey asks students to identify who they usually have information exchange in life and study subjects. A roster of all students enrolled in the course is included with the survey to assist students' reporting. An excerpt from the SNA survey is translated into English and illustrated in Table  1. The attached informed consent form made clear that the raw data would be handled by a member of the research team who was unattached to the course and that if a student did not sign or return the form, any data provided about them by others would not be used. It also asked them to agree to the researchers obtaining demographic and attainment information from university databases.

Table 1. An excerpt from the SNA survey that was given in the CM classroom
Please rate your closeness to all classmates. If you go to the line with your name, choose "Intimate friend" by default. You may need associations such as the following instructions: An intimate friend is someone whom you have everyday information exchange with A close friend is someone whom you weekly have at least 2-3 times of information exchange with A casual friend is someone whom you monthly have at least 2-3 times of information exchange with An acquaintance is someone whom you rarely have information exchange with Information exchange could be done face to face or online.

Intimate friend
Close friend Casual friend Acquaintance Name 1 Name 2 Students' GPA represents students' academic performance data before the study, expressed on the typical 10.0 scale. Complete grades data set were downloaded from the university's electronic record system. Data management and analysis are carried out using Gephi and SPSS Statistics.

RESULTS AND DISCUSSION
Class P's resulting directed weighted network consists of 41 nodes and 1168 ties, where the closeness of friendship defines tie weight: Intimate friend = 3; Close friend = 2; Casual friend = 1 and Acquaintance = 0. Overall, the average degree is 28.488, and the average weighted degree is 49.585. Put another way, 28.488 is the average number of edges per node in the graph. 49.585 indicates how many times the edge is passed between a pair of nodes. Meanwhile, the network diameter is 3, meaning the average graph distance between all pairs of nodes is 3. This figure will be one if every node is connected to every other node. The measure of how many ties between actors exist compared to how many ties between actors are possible -graph density -in this case, is 0.712. The visualization and measurement reveal that the network studied here is generally dense, well-tied without any isolates. At the same time, the web of Class V students embraces 33 nodes and 659 ties. The average degree is 19.97, and the average weighted degree is 33. To put in perspective, a student in Class P has more and stronger connections with fellow students.

Correlation between social connectedness and academic performance
For the Pearson r correlation, both variables should be normally distributed (normally distributed variables have a bell-shaped curve). After checking by drawing the histogram, the pairs of variables have a normal distribution with each other. Correlation analysis was performed using SPSS and the results obtained are shown in Table 2 and Table 3  Academic achievement, as demonstrated in this paper by scores, is correlated with the following variables • at a 99% confidence level: indegree, weighted indegree, eigencentrality; • at a 95% confidence level: weighted degree. In the case of Class V, the results in Table 3 show that the Sig. Value is greater than 5%, meaning that there is no significant correlation between the independent variable (indicators measuring a student's connection in class) and the second variable (the student's learning outcomes). It means that the number of other peers' connections, whether high or low, a student's centrality is not correlated with academic performance. In other words, there is no evidence that a student's level of connection with classmates positively or negatively affects learning. This finding is similar to many case studies showing a small or moderate correlation between academic connection and achievement (Osterman, 2000). Time-length studies have also found a very modest correlation between school social connectivity and academic performance (Archambault et al., 2009;Wang & Holcombe, 2010).
On the contrary, many interesting and diverse results were obtained from Class P, as presented in Table 2. A student with many people who viewed him/her as a close friend appears to have good grades. According to Chen et al. (2002) and Sheskin (2020), this is demonstrated by the strong correlation, i.e., .605, between 'grade' and 'weighted indegree.' To a certain extent, this finding is in line with those reported by Samdal et al. (1999) and Wilkins et al. (2016) that students have good results if they are satisfied with their school, teachers, and fellow students. Surprisingly, the eigenvector centrality also seems to be the significant predictor of good grades. This finding substantiates previous literature on associations between student social networks and academic performance (Gomes Jr, 2019; Ortiz et al., 2004). Taken together, these results would seem to suggest that the characteristics of the network should be judged on a case-by-case basis, and large-scale SNA analyses have been rarely reported. Although the large sample size experiments provide generalizable outcomes, they may be timeconsuming, error-prone, and expensive. Further work needs to be performed to improve the representativeness of the limited sample, to remain doable, and avoid excessive jargon.

CONCLUSION
This paper used a case study and questionnaire survey to measure and model the connections between members of two undergraduate classes and make relationships with learning outcomes. The two networks of interest were proved to be generally dense and well-tied through modeling and computing without any isolates. To test whether there is a correlation between score and intersection, the paper used the correlation analysis function in SPSS. The results show similarities with many previous studies that student centrality and connectedness positively affect scores in one class. In other words, the results show a contrasting situation as there is no evidence that students' connectedness attributes impact learning performance.
Given that the results obtained from the two cases are contradictory, the interpretations should be treated with caution. Educators can also use this research to generate ideas about classroom organization, group work improvement, and creating collaboration mechanisms. It is also hoped that this work will be constructive in solving the difficulty of revealing possible patterns of in-class social networks and academic performance. This study has few limitations. Some students had not participated in the survey, which caused the data to be incomplete in the 'real-life class' meaning. The data only reflects the state at the time of collection, while the network status and academic performance are more likely to change over time. Using a similar approach to this article, it is believed that there is ample opportunity to study the association between classmate connectedness and career success. Also, researchers can use the ego-centric network to study students' social connection with classmates and other schoolmates, teachers, and friends from other institutions. In general, the application range of SNA in examining relational data, including education, is very broad. Research techniques and approaches around SNA are expected to evolve further shortly, mainly as it is not limited to human-human relationships during the surge of multidisciplinary collaboration.