Social Network Characteristics of Learners in a Course Forum and

Their Relationship to Learning Outcomes

Zhi Liu

, Lingyun Kang

, Monika Domanska

, Sannyuya Liu

, Jianwen Sun

and Changli Fang

National Engineering Research Center for E-Learning, Central China Normal University,

Luoyu Road 152, 430079 Wuhan, China

Department of Computer Science, Humboldt University of Berlin, Rudower Chaussee 25, 12489 Berlin, Germany

Keywords: Social Network Analysis, Network Structure, Network Position, Learning Outcomes.

Abstract: Recently, learning analytics has become the focus in the interdisciplinary field of education technology.

Among learning analytical approaches, social network analysis (SNA) plays a critical role in examining

collective learning patterns. In this study, we collect the forum data in an undergraduate course from a

university’s online learning system. On the one hand, SNA is adopted to investigate the learners’ social

network characteristics including network structure and network positions. On the other hand, we adopt the

Pearson correlation analysis to identify the relationship between social network positions (e.g., degree

centrality, closeness centrality, betweenness centrality, prestige and influence) and learning outcomes of

learners. The experimental results show that most high-performing learners are located in the core position of

network. Moreover, there is a significantly positive correlation between learner’s social network centrality

and learning outcomes, and high-performing learners have higher prestige and influence in the forum. The in-

depth analyses could help teachers establish effective interactive mechanism that meets knowledge skills of

different individuals, as well as guide learners to help each other in collaborative learning.

1 INTRODUCTION

In recent years, with the increasing fermentation of

educational big data, learning analytics field has

integrated various approaches from multi-fields such

as information science, psychology, sociology, etc.

Among these approaches, social network analysis

have been a critical approach in exploring collective

learning processes (Jo et al., 2014; Kellogg et al.,

2014; Lee & Bonk, 2016). In online learning

environment, learners often spontaneously form

various self-organized learning communities based

on their own learning requirements, interests or tasks.

These communities are built based on the concept of

social relations (Baker-Eveleth, 2003). Meanwhile,

large scale of complex interactive data have been

generated in various online learning systems or social

media. The learners from an on-line learning

community could naturally constitute a network,

where each learner could be viewed as a node.

Interactions among internal members con-tribute to

knowledge construction of each individual in the

network. Therefore, it is worth to investigate that,

what are the structure of social networking and

individual characteristics in interactions, and what is

the relationship between characteristics of learners in

the network and learning outcomes. The exploration

of these questions is beneficial for reshaping

education contexts, teaching methods and optimizing

the learning effects of learners.

This paper aims to adopt social network analysis

method to carry out the empirical research to reveal

the structure and evolution trends of the social

network of learners within online course forum, as

well as the relationship between the network positions

and the learning outcomes of learners. This paper is

organized as follows. In Section 2, we review the

definition of social network analysis and related

research in the online learning environment. The

design of this study is presented in Section 3. Results

are showed in Section 4. Section 5 concludes findings

in this study.

2 RELATED WORKS

Social network analysis (SNA) was derived from the

studies of sociology, psychology and anthropology in

Liu, Z., Kang, L., Domanska, M., Liu, S., Sun, J. and Fang, C.

Social Networ k Characteristics of Learners in a Course Forum and Their Relationship to Learning Outcomes.

DOI: 10.5220/0006647600150021

In Proceedings of the 10th International Conference on Computer Supported Education (CSEDU 2018), pages 15-21

ISBN: 978-989-758-291-2

the 1930s. Social network refers to a collection

including social actors and various relations among

them (Peter et al., 2011; Scott & John, 2000), in

which relationship is the most important research

object in SNA.

In the field of learning analytics, the main

concerns are the relationship among learners, the

relationship between learners and teachers, as well as

the relationship between learners and learning

resources. Learners often utilize social networks to

find the most likely collaborative partner when they

intend to seeking for helps to solve problems. A large

number of studies have confirmed that SNA has

important value in evaluating interactive behaviors of

participants in the online learning environment. Laat

et al. (2006) and Aviv et al. (2003) adopted SNA to

solve a series of questions about calculation of

participant activity degrees, network densities and

identification of core participants. Karina et al. (2015)

used SNA to evaluate the collaboration quality within

collective online learning. Xu & Yang (2015) utilized

social network metrics to characterize the strength of

the relation-ship among learners, which was used to

recommend learning companions for struggling

learners in on-line course. Moreover, many

researchers also investigated the relationship between

social network characteristics and learning outcomes

of learners. For instance, Dowell et al. (2015) and

Tobarra et al. (2014) integrated SNA and discourse

content analysis to jointly explore the associations

among discourse features, learning outcomes and

social centrality. Russo and Koesten (2005) suggested

that learners' cognitive learning results could be

predicted by analysing in-degree and out-degree

indicators of individuals in learning networks.

Rizzuto et al. (2009) demonstrated that network

density could reflect learners’ understanding levels

on course materials to a large extent. The study of Lee

& Bonk (2016) explored the relationship among

learners in a blended learning environment by

measuring the density, factions and centrality of a

relationship net-work, and suggested that the active

learners tend to be more popular within a relationship

network.

3 EMPIRICAL RESEARCH

3.1 Research Questions

In the online learning environment, the forum inter-

action could help in knowledge sharing and learning

supporting. To enable learners to effectively engage

in knowledge construction in forums, there is a need

to understand the network characteristics of learners

and to demonstrate their relationship to learning

outcomes. This study will be conducted aiming at the

following questions:

(1) How does the social network structure of a

course forum evolve as the course progresses?

(2) What is the relationship between social

network positions of learners in a course forum and

learning outcomes?

3.2 Research Objects and Dataset

The interaction data in this study comes from the

forum of “Literature Translation” course, which was

opened in the spring of 2014 in the online learning

platform of a university. The teachers adopts the

blending learning mode in this course, integrating the

traditional classroom teaching and online

collaborative discussions. The course lasts for 4

months. The learners who take it as an elective course

are all senior undergraduate students majoring in

English Translation. A total of 75 participants (74

students and 1 teacher) engage in discussions in the

course forum. A total of 2982 posts were generated

during their interactions. In order to explore the

relationship between social network positions and

learning outcomes in the course forum, we take the

network characteristics of 75 participants as the

independent variables including degree centrality,

closeness centrality, betweenness centrality, prestige

and influence. The learning outcome of each

participant, as the dependent variable, is represented

by his/her final overall score at the end of semester,

which is composed of two parts, i.e., the usual score

and the final score, each of which accounts for 50%

of overall score.

3.3 Research Design

In order to address the first question, this study first

investigates the evolution trends of learners’

interactive frequency in the forum. In addition, we

analyse the number of monthly posts by continuous

participants and verify the difference of academic

achievement between continuous participants and the

entire population. Then, by SNA method, we

construct the monthly and entire sociogram according

to learners' interactions, and explore the network

characteristics within monthly interaction and the

overall network structure. As for the second question,

the Spearman correlation analysis is used to

demonstrate the relationship between social network

positions of learners in a course forum and their

learning outcomes.

CSEDU 2018 - 10th International Conference on Computer Supported Education

Figure 1: Statistical results of forum participants and posts.

Social network centrality embodies the central

position of individual or organization in a network,

indicating the importance of a member in the network.

This study respectively utilizes degree centrality,

closeness centrality and betweenness centrality to

characterize each learner’s social network position.

Moreover, the learning network formed in forum

interactions is actually a directed social network

graph, which implies individual prestige and

influence in interactions. Here, the individual prestige

in the network is be quantified by in-degree centrality

of the corresponding node, i.e., the number of replies

an individual received. The individual influence is

quantified by the out-degree centrality of

corresponding node, i.e., the number of posts of an

individual sending to other individuals (Shea et al.,

2013). The tools used in this paper are Python's

NetworkX Package and Gephi visual network

analysis software, respectively.

4 RESEARCH RESULTS

4.1 Forum Participation

From Figure 1, we find that the number of monthly

participants in forum tends to be a gradual descending

trend in the teaching progress of the course. There are

72 participants in the first month, which decreases to

50 in last month. The monthly participation rate

ranges from 66.67% to 96.00%. Compared with

monthly participants, the number of posts ascend to

1,012 in the second month, but gradually descends in

the last two months. When the course approaches the

end of the semester, the average times of learners

engaging in discussions will be significantly lower

than before.

As for the continuous participants, who have been

active (i.e., engaging in discussions each month) in

forum during the entire course, there is a total of 43

learners to continually participate in the forum

discussions, accounting for a half of the entire

population. The statistical results indicate that the

monthly posts of continuous participants show the

same variation trend as the monthly posts of entire

population.

4.2 Social Network Characteristics of

Learners

4.2.1 Evolution of Network Structure

Figure 2 shows the sociograms of learners during four

learning periods. In each diagram, green node denote

teachers, orange node denotes the learners whose

overall scores rank the top 20 in the class, purple node

denotes the learners who participates in the forum and

whose overall scores rank the last 20 in the class.

Each node has a number (the teacher number is 0,

student number is marked in the order of posting).

The node size represents node’s degree, which is the

sum of replied and delivering postings). It can be seen

from Figure 2 that the positions of the learners in the

sociogram gradually vary with the course progresses.

Nevertheless, the majority of the orange nodes are

always at the core position of the network, which

signifies that learners with better academic

performance tend to be more active in the forum. In

addition, in the first month, the green node has the

largest degree and is located at the centre of the

network, indicating that the discussions are mainly

conducted between the teacher and learners. With the

progress of the course, the teacher gradually moves to

the edge of the sociogram, the forum interactions

mainly occurs among learners.

Social Network Characteristics of Learners in a Course Forum and Their Relationship to Learning Outcomes

(a)

(b)

(c)

(d)

Figure 2: Monthly sociogram: (a) first month, (b) second month, (c) third month, (d) forth month.

When interactions among learners tend to be frequent,

the sociogram gradually tends to be more intensive.

4.2.2 Characteristics of Network Structure

The results of network structure characteristics are

shown in Table 1, it can be unfolded that network

density, the average closeness centrality and

betweenness centrality of the learners all show

gradual rising trends. As the course approaches the

end of semester, the interactions begin to gradually

shrink, and the three indicators all exhibit a

descending trend. Specifically, the overall network

density of the course arrives at 0.45, which indicates

that the network structure of the forum in this course

is intensive. Moreover, the learners participating in

the forum are relatively active and have close ties

with each other. Figure 3 displays the overall network

structure, the most orange nodes are located at the

core position of the network, indicating that these

high-performing participants actively participate in

the forum discussions. An interesting phenomenon is

that a small number of nodes such as 3, 17, 20 and 63

are located at the edge of the forum. Actually, in a

blending teaching mode, although the four learners

are rarely involved the forum, they may have the

relatively good English foundation, and they could

deeply engage in classroom learning. Therefore, they

could also gain the higher overall scores.

Figure 3: Overall sociogram.

4.3 Relationship between Learners’

Network Positions and Learning

Outcomes

4.3.1 Relationship between Network

Centrality and Learning Outcomes

Degree centrality signifies the number of learners

with who a learner establishes direct contacts in the

forum. That is to say, the more times of interactions

between a learner and other learners, the higher

degree centrality of this learner. From Table 2, we can

observe that the Spearman correlation coefficient

CSEDU 2018 - 10th International Conference on Computer Supported Education

Table 1: Metrics of monthly and overall network characteristics.

Period

Max./min.

out-degree

Max./min.

in-degree

Degree (std.)

Density

Closeness

centrality (std.)

Betweenness

centrality (std.)

month

25/1

64/0

13.36 (±14.46)

0.19

0.30 (±0.04)

0.01 (±0.04)

month

21/0

43/0

18.65 (±14.51)

0.28

0.32 (±0.08)

0.01 (±0.02)

month

21/0

43/0

18.63 (±13.47)

0.32

0.35 (±0.06)

0.02 (±0.02)

month

11/0

28/0

8.64 (±7.48)

0.18

0.21 (±0.08)

0.02 (±0.04)

Overall

semester

42/0

65/0

34.71 (±26.43)

0.45

0.44 (±0.07)

0.01 (±0.02)

Table 2: Spearman correlation coefficient between social network centrality and learning outcomes.

Number of

samples

Variable

Mean

Standard

deviation

Degree

centrality

Closeness

centrality

Betweenness

centrality

Overall

score

N=74

Degree centrality

27.00

17.27

—

Closeness

centrality

0.44

0.07

0.80**

—

Betweenness

centrality

0.01

0.93**

0.79**

—

Overall score

87.81

2.30

0.46**

0.35**

0.39**

—

**p<0.01

Table 3: Spearman correlation coefficients between prestige, influence and learning outcomes.

Number of

samples

Variable

Mean

Standard

deviation

Prestige

Influence

Overall score

N=74

Prestige

33.47

46.02

—

Influence

35.22

26.09

0.67**

—

Overall score

87.81

2.30

0.41**

0.40**

—

**p<0.01

between individual degree centrality and learning

outcome reaches 0.47 (p<0.01), indicating a highly

positive correlation between the two variables in the

network.

Closeness centrality indicates the average

closeness degree between a learner and other learners

in the forum, describing the degree of dependence of

the learner to other learners in a network. The higher

of closeness centrality of a learner, and the less he/she

depends on other learners when seeking for helps. It

can be seen from Table 2 that the correlation

coefficient between the closeness centrality and

learning outcome in this course reaches 0.35 (p

<0.01), indicating that there is a significantly positive

correlation between the two variables. This also

implies that the close degree of a learner to others may

predict his/her learning outcome to an extent.

Betweenness centrality represents the extent of a

learner being an “intermediary” in interactions and

describes the learner’s ability to adjust the social

interactions. The “intermediary” not only can control

the direction and manner of an information flow, but

also can coordinate relationship of any other two

individuals or organizations. Therefore, the

“intermediary” could play a bridge role in the

learning network. It can be observed from Table 2

that the correlation coefficient between the two

variables equals 0.39 (p<0.01), indicating a

significantly positive correlation between the two

variables. It can be assumed that, the higher the

degree of “intermediary” role of a learners in the

learning network, the higher his/her learning outcome.

We also observe that, the betweenness centrality of

learners is quite low, ranging from 0.00 to 0.07, while

the teacher’s betweenness centrality could reach the

highest level of 0.11. This implies that the teacher

actually plays a major leading role in coordinating

interactive relationship within the forum, and guides

learners to follow specific topics for enhancing the

understanding of knowledge.

4.3.2 Relationship among Prestige, Influence

and Learning Outcomes

In a directed social network, the in-degree centrality

and out-degree centrality could be jointly used to

indicate a learner’s prestige and influence in the

network (Shea et al., 2013). The in-degree centrality

refers to the number of posts of a learner receiving

from others. The high in-degree learner is considered

Social Network Characteristics of Learners in a Course Forum and Their Relationship to Learning Outcomes

to have a high prestige in the network since the views

and ideas expressed by prestigious learners are

considered more important than other members. On

the other hand, the out-degree centrality is used to

measure learners’ influence in a network, which can

be measured by the number of posts of a learner

delivering to others, indicating that the learner’s

activity degree in engaging in a forum. As shown in

Table 3, there are significant positive correlations

among learners’ learning outcomes, prestige and

influence. In other words, the learners with high

academic performance tend to have a high prestige or

influence. The influential learners typically could

receive more replies. Additionally, the result also

shows that the learners participating in interactions

averagely gain the overall score of 87.81, and there is

a quite low average deviation among the learners’

scores (standard deviation is 2.3). As for those

learners who never participated in the forum (a total

of 5), their overall scores rank relatively backward

such as 26, 27, 73, 74 and 77, respectively, the

average score of which is less than learners who

participated in the forum.

5 CONCLUSIONS

This study utilizes the social network analysis to

investigate the evolution trends of network structure

of learners within a course forum in a university’

online learning platform, as well as further analyze

the correlation between individual position features

and learning outcomes in the forum. We could draw

the following conclusions:

Social network structure of learners would

dynamically vary as the progresses of course. In the

first three months of the course, the network density,

number of participants, number of posts and network

centrality all show gradual upgrading trends. That is,

the interactions among learners tend to be

increasingly frequent while links among them

become closer. However, in the last month, when the

course approaches the end of the semester, both the

numbers of participants and posts decrease, as well as

the sociogram also becomes relatively sparse.

Within interactions of the course forum, the

positions of learners in the network are partially

correlated to learning outcomes. Social network

centrality metrics have significantly positive

correlations with learning outcomes. The learners

with higher prestige or influence in social network

could typically gain higher learning outcomes. And

the factor that is most correlated to learning outcome

is degree centrality, followed by betweenness

centrality, the last one is closeness centrality. Finally,

learner’s prestige and influence in a forum are

significantly positively correlated to their learning

outcomes. This also indicates that the high-achieving

learners generally have the high prestige and

influence.

Therefore, if designed appropriately, discussion

activities may not only enhance the interactions

among learners, but also facilitate collaborative

inquiry learning and knowledge construction among

learners. To improve the activity levels of learners

among interactions, teachers may design some high-

quality interactive activities like inquiry-based

discussions, questions and answers, literature reviews

and knowledge brainstorms. Also, these activities

should be designed to be appropriate for knowledge

skills and interests of learners as well as have a certain

difficulty to drive learners to actively conduct

collaborative inquiries and discussions.

ACKNOWLEDGEMENTS

This work was supported by the Research Funds from

National Natural Science Foundation of China (Grant

No. 61702207), MOE (Ministry of Education in

China) Project of Humanities and Social Sciences

(Grant No. 16YJC880052), China Scholarship

Council (Grant No. 201706775022), National Social

Science Fund Project of China (Grant No.

14BGL131), Ministry of Education-China Mobile

(Grant No. MCM20160401).

REFERENCES

Aviv, R., Erlich, Z., Ravid, G., & Geva, A. (2003). Network

analysis of knowledge construction in asynchronous

learning networks. Journal of Asynchronous Learning

Network, 7(3): 1–23.

Baker-Eveleth, L. J. (2003). An online third place:

Emerging communities of practice. Ph.D. Dissertation.

United States: Washington State University.

Cela, K. L., Sicilia, M. Á., & Sánchez, S. (2015).

Comparison of collaboration and performance in

groups of learners assembled randomly or based on

learners’ topic preferences. Educational Technology &

Society, 18(4): 287–298.

Dowell, N. M., Graesser, A. C., Hennis, T. A., et al. (2015).

Modeling Learners’ Social Centrality and Performance

through Language and Discourse. In Proceedings of the

8th International Conference on Educational Data

Mining, pages 250-257. ERIC.

Jo, I. H., Kang, S., & Yoon, M. (2014). Effects of

Communication Competence and Social Network

CSEDU 2018 - 10th International Conference on Computer Supported Education

Centralities on Learner Performance. Journal of

Educational Technology & Society, 17(3): 108-120.

Kellogg, S., Booth, S., & Oliver, K. (2014). A social

network perspective on peer supported learning in

MOOCs for educators. International Review of

Research in Open and Distance Learning, 15(5): 263-

289.

Laat, M. D., Lally, V., Lipponen, L., & Simons, R. J.

(2006). Analysing student engagement with learning

and tutoring activities in networked learning

communities: a multi-method approach. International

Journal of Web Based Communities, 2(4): 394-412.

Lee, J., & Bonk, C. J. (2016). Social network analysis of

peer relationships and online interactions in a blended

class using blogs. Internet and Higher Education, 28:

35-44.

Peter J. & Scott, J. (2011). The Sage Handbook of Social

Network Analysis. UK: SAGE Publications Ltd.

Rizzuto, T. E., Ledoux, J., & Hatala, J. P. (2009). It’s not

just what you know, it’s who you know: Testing a

model of the relative importance of social networks to

academic performance. Social Psychology of

Education, 12(2): 175–189.

Russo, T. C., & Koesten, J. (2005). Prestige, centrality, and

learning: A social network analysis of an online class.

Communication Education, 54(3): 254–261.

Scott & John (2000). Social Network Analysis: A

Handbook. UK: SAGE Publications Ltd.

Shea, P., Hayes, S., Smith, S. U., Vickers, J., et al. (2013).

Online learner self-regulation: Learning presence

viewed through quantitative content and social network

analysis. International Review of Research in Open and

Distance Learning, 14(3): 427–461.

Tobarra, L., Robles-Gómez, A., Ros, S., Hernández, R., &

Caminero, A. C. (2014). Analyzing the students’

behavior and relevant topics in virtual learning

communities. Computers in Human Behavior, 31(1):

659-669.

Xu, B., & Yang, D. (2015). Study partners recommendation

for xMOOCs Learners. Computational Intelligence and

Neuroscience, 2015: 832093.

Social Network Characteristics of Learners in a Course Forum and Their Relationship to Learning Outcomes