Source-linked AI summary
A Survey on Studying the Social Networks of Students
Akrati Saxena, Pratishtha Saxena, Harita Reddy, Ralucca Gera
TL;DR
The paper asks how student social networks relate to learning, performance, information flow, and behavior. It reviews studies using SNA to organize this research and reports findings on peer knowledge sharing and network-performance relationships. The review identifies limits in generalizing from small groups and in self-reported data.
Problem
Research on student networks addresses relationships between network position and academic performance, classroom information diffusion, collaboration, teamwork, and homophily.
Source: §I · p. 1Method
The paper presents a structured review of studies using SNA to analyze students’ offline and online networks and partitions the research into six fields.
Source: §I · p. 2Results
In CSCL, peer students can become influential information sharers as their knowledge increases and knowledge spreads during a course.
Source: A. The Effectiveness of Collaborative… · p. 2Takeaways & Limitations
The review suggests that understanding student network structure can help inform educational techniques for inclusive learning.
Source: §I · p. 2Findings from small student groups may not generalize, and self-reports about adversarial relationships and habits such as smoking or drinking may be inaccurate because of bias or hesitation.
Source: §III · p. 6
Abstract
from arXiv · showhide
Do studies show that physical and online students' social networks support education? Analyzing interactions between students in schools and universities can provide a wealth of information. Studies on students' social networks can help us understand their behavioral dynamics, the correlation between their friendships and academic performance, community and group formation, information diffusion, and so on. Educational goals and holistic development of students with various academic abilities and backgrounds can be achieved by incorporating the findings attained by the studies in terms of knowledge propagation in classroom and spread of delinquent behaviors. Moreover, we use Social Network Analysis (SNA) to identify isolated students, ascertain the group study culture, analyze the spreading of various habits like smoking, drinking, and so on. In this paper, we present a review of the research showing how analysis of students' social networks can help us identify how improved educational methods can be used to make learning more inclusive at both school and university levels and achieve holistic development of students through expansion of their social networks, as well as control the spread of delinquent behaviors.
I. INTRODUCTION
The survey reviews how SNA has been used to study students’ offline and online networks, including learning, performance, information flow, and behavior. It organizes this work into six fields, highlights limitations, and identifies future research directions.
I. INTRODUCTION
- Student networks have been used to study academic performance, classroom information diffusion, collaborative learning, teamwork, and homophily.This research can help educators assess collaborative learning, identify students isolated from peer help, and examine classroom information flow.Source: §I · p. 1
- Student network datasets can be built from questionnaires about friendships and help-seeking or from online interaction data, yielding friendship, collaboration, and help-seeking networks.Questionnaires can ask students to nominate friends, gurus, and adversaries; platforms such as Moodle can provide interaction data.Source: §I · p. 1
- UCINET, RSiena, SNAPP, Gephi, and NetworkX support different network-analysis tasks, from centrality and subgroup identification to viewing student interactions.SNAPP can help instructors identify community structures and students isolated from others.Source: §I · pp. 1–2
- A structured review examines studies using social network analysis tools to analyze students’ offline and online networks.The review follows an outline of commonly used SNA software programs.Source: §I · p. 2
- A review of SNA methods for studying students’ learning and behavior can clarify the state of research in networks and education.Understanding network structure may help inform educational techniques for inclusive learning.Source: §I · p. 2
- The survey partitions research on networks and education into six fields, highlights limitations in existing work, and suggests future research directions.The fields cover collaborative learning, teamwork, performance and network structure, knowledge dissemination and peer support, homophily and communities, and student habits.Source: §I · p. 2
II. ANALYSIS OF STUDENT SOCIAL NETWORKS
SNA examines student relationships at the individual, group, and collaborative-network levels, revealing how interaction patterns relate to learning and the roles of particular actors or groups.
II. ANALYSIS OF STUDENT SOCIAL NETWORKS
- SNA can examine relationships among students in collaborative networks, groups, and the larger network of individuals.The analysis can indicate how much learners benefit from student interactions and the importance of particular actors or groups in supporting learning.Source: §II · p. 2
- Network structure can indicate learners’ benefit from interaction and the role of particular students or groups in supporting learning.Source: §II · p. 2
- The survey uses these network perspectives to organize its discussion of six education-related aspects of student networks.Source: §II · p. 2
A. The Effectiveness of Collaborative Learning Environments
Studies use SNA to assess participation and communication in collaborative learning environments, and report both shifts in peer interaction and conditions that can limit their effectiveness.
A. The Effectiveness of Collaborative Learning Environments
- In CSCL, network analyses of participation and message exchange can assess student roles and whether communication is centered on the instructor.Measures include node degree, betweenness and closeness centrality, and student roles.Source: A. The Effectiveness of Collaborative… · p. 2
- Fewer student-to-student interactions indicate a poor collaborative environment, while intervention can help students with high betweenness relay information across the network.Source: A. The Effectiveness of Collaborative… · p. 2
- In BSCW, doubt-solving was the most decentralized activity, while information sharing relied on a few highly central participants.Students central in pre-existing ties showed lower change propensity in CSCL networks, while peripheral students tended to show greater willingness to form ties over time.Source: A. The Effectiveness of Collaborative… · p. 2
- As CSCL courses progress and knowledge spreads, students can emerge as peer More Knowledgeable Others after the instructor initially fills that role.Source: A. The Effectiveness of Collaborative… · p. 2
- In online threaded discussions, high-achieving students tend to bridge groups, and proper use of discussions increases participation and connections while reducing average closeness centrality.Meerkat-ED can dynamically display interactions and identify central students by discussion topic.Source: A. The Effectiveness of Collaborative… · p. 2
- Collaborative methods such as OTDs can increase interaction, but participation may fade with repeated use, and one higher-education PAL scenario found prior performance more predictive than network characteristics.The passage also notes that success depends on students’ motivation and academic abilities.Source: A. The Effectiveness of Collaborative… · p. 3
B. Studies on the Teamwork Among Students
Team outcomes are associated with communication balance, cohesion, and network roles, while course structure can shape whether students connect or remain disconnected.
B. Studies on the Teamwork Among Students
- High communication within teams is positively correlated with team effectiveness, while balanced contributions and fewer adversarial relationships are associated with better team results and satisfaction.One reported observation links more workload sharing with lower grades; the passage suggests successful teams may have a few members taking on most of the work.Source: B. Studies on the Teamwork Among… · p. 3
- Cohesion and centrality were the most important predictors of team performance across networks of advice, leadership, social, and obligation ties.Cohesion and group conflict were measured as ratios of positive and negative relationships, respectively; negative ties were inferred from low teammate rankings in a positively worded questionnaire.Source: B. Studies on the Teamwork Among… · p. 3
- Less-structured classes produced connected groups alongside disconnected students, whereas more-structured classes had a more even distribution of interactions.SNA can help instructors identify disconnected students and encourage their participation in teamwork.Source: B. Studies on the Teamwork Among… · p. 3
- Dynamic SNA-based team formation in MOOCs was reported to enhance participation and reduce attrition through active engagement.Teams can support task completion, doubt clarification, discussion, and knowledge brokerage between groups.Source: B. Studies on the Teamwork Among… · p. 3
- Team success is associated with communication level and balance, cohesion, brokers, and the absence of disconnected students.The passage relates these factors to both team grades and student engagement within teams.Source: B. Studies on the Teamwork Among… · p. 3
C. Academic Performance versus Social Networks
Across several settings, student network position is associated with academic performance, and high-achieving students tend to occupy central positions and form persistent ties. These relationships vary by network type, and static analyses may mislead when performance and network participation affect one another.
C. Academic Performance versus Social Networks
- Communication- and advice-network centrality are positive indicators of academic performance, while adversarial-network centrality is negatively correlated with student satisfaction.Study-environment centrality in university also relates to learning and employability.Source: C. Academic Performance versus Social… · p. 3
- Because good academic performance may itself increase network participation and centrality, static network analysis can lead to misleading conclusions [41] [42].Source: C. Academic Performance versus Social… · p. 3
- High-performing college students tend to form persistent ties with other high-performing students from early in a course, and advanced-course students often have larger ego networks and more interactions.The review concludes that high achievers tend to be central and establish persistent ties with classmates.Source: C. Academic Performance versus Social… · p. 4
- In MOOCs, Eigenvector centrality is reported as one of the most useful predictors of academic achievement, with high-performing students occupying central network positions.Source: C. Academic Performance versus Social… · p. 4
- Moodle behavior similarity correlates positively with grade similarity, but the study does not support the hypothesis that higher centrality causes better performance.Across two biology-course exams, more complete triads indicated more study groups and greater network density; second-exam performance also correlated with betweenness and degree centrality.Source: C. Academic Performance versus Social… · p. 4
- Offline closeness centrality is significantly correlated with academic performance, whereas the correlation is weaker for online closeness centrality; online network data require rigorous analysis before applying sociological metrics.Source: C. Academic Performance versus Social… · p. 4
D. Knowledge Dissemination and Peer Support
Knowledge diffusion depends on network structure, while collaboration-demanding activities and peer-help tools are associated with denser, more inclusive networks and expanded opportunities for support.
D. Knowledge Dissemination and Peer Support
- Effective classroom knowledge diffusion is associated with leaders, optimal network density, and subgroups with adequate connections between them.Degree and betweenness centrality identify leaders, while cohesion measures inter-group diffusion; bridge students are academically outstanding.Source: D. Knowledge Dissemination and Peer… · p. 4
- Collaboration-demanding assignments can create a useful information network, increasing density and average degree centrality while reducing average geodesic distance.Intensive collaboration also improves diffusion efficiency and inclusiveness and alleviates collapse risk through the absence of cutpoints.Source: D. Knowledge Dissemination and Peer… · p. 4
- Centrality in peer-information networks is associated with greater learning, while project-group members often seek subject-knowledgeable peers and online forums expand graduate students’ ego networks.Project-group formation is highly dependent on pre-existing friendships.Source: D. Knowledge Dissemination and Peer… · p. 4
- Student relationships may stabilize over time, and stochastic actor-based analyses find that high-achieving students gain more incoming connections as time passes.The review says this may reflect peers seeking study help.Source: D. Knowledge Dissemination and Peer… · p. 4
E. Study on Subcultures and Homophily Among Students
Student networks show homophily by attributes such as academic performance and gender, while studies also identify limits to measuring multiple similarities and trade-offs in homophilic grouping.
E. Study on Subcultures and Homophily Among Students
- Students tend to form relationships with people sharing similar traits, and studies report academic- and gender-based homophily in high-school friendship and consultation networks.Working while studying is another reported similarity associated with denser connections.Source: E. Study on Subcultures and Homophily… · pp. 4–5
- Students tend to select friends with similar academic performance rather than adapt their own performance to current friends; girls exhibit more selection.Source: E. Study on Subcultures and Homophily… · p. 5
- Most studies examine similarity dimensions separately, while a multidimensional adolescent-network study found ties between people sharing multiple attributes might not evolve further.The review suggests students may seek friends with different thoughts and knowledge.Source: E. Study on Subcultures and Homophily… · p. 5
- Degree and closeness centrality positively indicate students’ sense of community, whereas betweenness is negatively correlated; the reported study measured only online discussion.Source: E. Study on Subcultures and Homophily… · p. 5
- Homophilic groups may foster inclusiveness but can also group poorly performing students or reflect racial and ethnic segregation; inter-group ties are proposed as a countermeasure.A seat-reservation model for backward-class students found the gap with upper-class students reduced over time, but used basic network modeling needing real-life parameters [62].Source: E. Study on Subcultures and Homophily… · p. 5
F. Studies on Adolescent Behavior
The review links student friendship networks to behavioral change, including aggression and substance use, while also examining health habits and academic misconduct.
F. Studies on Adolescent Behavior
- Students connect with peers showing similar aggression, while increased network importance and friends’ aggression are linked to adolescent aggression, except among very highly central students.Friends’ delinquent behavior also has a significant impact beyond school involvement.Source: F. Studies on Adolescent Behavior · p. 5
- Smoking alignment with friends and alcohol onset when many friends drink are reported; marijuana initiation influence appeared in only one of two schools.SABM analyses also report adolescents being attracted to drinkers where drinking confers high status; these analyses can help identify schools for potential marijuana interventions.Source: F. Studies on Adolescent Behavior · pp. 5–6
- Adolescents adjust BMI and physical activity toward friends’ levels, connect through homophily on these traits, and show friendship-network influence on junk-food consumption.The review suggests institutions may use such findings to design policies and activities that encourage better decisions when choosing friends.Source: F. Studies on Adolescent Behavior · p. 6
- Cliques and weak faculty ties are positively correlated with cheating, while genetic algorithms can optimize exam seating to reduce matches between friendship and lattice links [73].Source: F. Studies on Adolescent Behavior · p. 6
III. CONCLUSION AND FUTURE DIRECTIONS
The conclusion presents student-network analysis as a way to understand behavior, information diffusion, and academic-performance correlations, and to inform educational methods and policies. It also identifies limits in current evidence and future directions for data collection and network research.
III. CONCLUSION AND FUTURE DIRECTIONS
- Student-network analysis connects network structure with behavior, information diffusion, and habit acquisition, while academic performance is correlated with students’ network position.These observations can inform corrective behavior, alternate teaching methods, group activities, and policies affecting students.Source: §III · p. 6
- Findings from small student groups may not generalize, and self-reports of adversarial ties and habits may be inaccurate because of bias or hesitation.The authors also call for more frequent network evaluations and delayed follow-up to assess whether participation persists after new learning methods are introduced.Source: §III · p. 6
- Combined teacher–student networks remain little studied, with further real-life case studies needed to examine unethical practices such as cheating.The review identifies potential uses for such networks in studying teachers’ lasting influence, group formation, and students’ faculty connections.Source: §III · p. 6
- Future comparisons across universities with different diversity levels and intermingling programs could track homophily changes and identify ways to disrupt detrimental segregation.Source: §III · p. 6
- CHUNK learning represents content and recommends learning modules based on students’ differing abilities and skills, potentially complementing collaborative learning with personalized experiences.Source: §III · p. 6
- The authors favor practical, low-burden analysis and automatic online data collection, with SNA complementing rather than replacing human observation of student interactions.They propose easy-to-use tools for monitoring networks, participation, and disconnected students so faculty can consider appropriate interventions.Source: §III · pp. 6–7