Welcome to "Using Social Network Analysis to Study Networked Learning in Social Media"


Hi Everyone,

This week let’s discuss what we can or cannot learn when using Social Network Analysis (SNA) to study networked learning in social media.

To start this hotseat, please first complete a quick hands-on exercise:

and then answer the following three questions:

  1. What can we learn about a Twitter-based class and its participants based on the node-level SNA measures such as in-degree and out-degree centrality?

  2. What can we learn about a Twitter-based class and its participants based on the network-level SNA measures such as density, reciprocity, centralization, modularity?

  3. What we cannot learn by using SNA?

My hope is that by the end of this week, you will learn how to

  • discover and visualize online communication networks using Netlytic, a web-based tool designed to discover social networks from online conversations on social media, and
  • apply SNA to study collaborative learning processes in the discovered networks.

Note: If you have worked with social media data and used SNA before, please feel free to skip the hands-on part.


Hi Anatoliy and welcome to the hot seats.

We are very pleased you are hosting this discussion on social network analysis and i think it is great to get some hands on experience with it. This will help further our understanding and to think through how such an approach can help us to research networked LEARNING relationships.

I also hope there will be time to excurse to your lab and play with some of your tools :wink:
Lets have a great week!


Hi everyone,

Thanks to networkedlearning opportunities I am able to follow the hot seats, although I don’t have the chance to participate in the conference this year. Special thanks goes to organizers and contributors: @maarten, jeffreykeefer and the other hidden heroes :smile:

Also, I would like to thank @gruzd for providing a great tool and exercise in this hot seat. As a PHD candidate using Pajek, it was a pleasure for me to experience Netlytic.

When it comes to the answers of the specific questions, here are my answers:

1-What can we learn about a Twitter-based class and its participants based on the node-level SNA measures such as in-degree and out-degree centrality?
profesortbaker and shellterrell seems to have the highest in-degree values. And, cck11feeds have the highest out-degree values.

2-What can we learn about a Twitter-based class and its participants based on the network-level SNA measures such as density, reciprocity, centralization, modularity?
This is a huge question :slight_smile: We are provided with some metrics regarding these in netlytic and it also provides comparison charts. Let’s have a look at the measures.
Diameter: 33
Density: 0.004619360615372
Reciprocity: 0.083700440528634
Centralization: 0.101458179732940
Modularity: 0.665297156277477

It is not a very dense network. It has some isolates at the periphery. It shows some reciprocal relationship. Moreover, Netlytic points 5 different clusters and a relatively high modularity score. Of course we can say more. On the other hand, in order to talk more about these concepts, we should know more about the context! Maybe at the begining of the course some people are determined as a group?

3-What we cannot learn by using SNA?
Although we can learn much about the interaction and communication in the networked learning environments with SNA, the question of “what forms a learning tie?” is not answered yet. (see reference) I should also underline the importance of the assumptions when forming a network. Both the SNA related assumptions (how you form the network) and the underlying theoretical assumptions shape the conclusions gathered via SNA. I also think that SNA has some shortcomings when we try to embed the qualitative data within a network.

That’s all for now. :wink:
Hope to have a great discussion :smile:

Middle East Technical University (METU)

Haythornthwaite, C., & De Laat, M. (2010). Social networks and learning networks: Using social network perspectives to understand social learning. In 7th International Conference on Networked Learning.


Hi Didem,

Thank you for your detailed reply. I absolutely agree about the need to know the context (e.g., syllabus, lesson plan and activities) to be able to interpret the network properly. For example, how would your interpretation change (or wouldn’t) if you knew that Twitter was only one of at least two other CMCs used in the class?

I also like your (and Caroline’s) point about needing to know what constitutes a learning tie. I think it is especially important for Twitter data, as not every interaction may lead to “learning”. Without doing an advanced NLP to analyze exchanges, we may be able to limit the effect of noisy data by only considering “strong” ties.

For example, we can establish a threshold for what can be considered as a “tie” between two participants if and only if they had more than K interactions (K>1, empirically derived). You can also add a requirement that for a tie to be a tie, it must be reciprocal. Of course, this approach should not be used indiscriminately and should be applied and justified on a case-by-case basis.



Hello everyone!
Some nice discussions about SNA here. Tried the Netlytic software - easy-to-use and great online tool for SNA.
I have previously used SNAs to explore social participation in mobile Web 2.0 learning settings. The software used was NodeXL.

I was wondering if anyone has encountered any SNAs or any network diagram analysis software that could be used to analyze MOOCs? As students in such platforms are massive, it would be interesting to discover learning patterns (in form of network diagrams) of MOOCs.


Hi Anatoliy,

Really grateful for the questions you’ve posed here, since my research involves exploring learning through Twitter. How useful SNA is in helping us better understand and interpret the learning which is (or isn’t?) mediated by Twitter is a question I keep posing to myself. Is this a method that’s going to help me with my research questions? What this hotseat has done for me then, is to focus my attention and hopefully provided an opportunity to learn from people more experienced in using SNA in different contexts. (I am very much a beginner!).

I’ve only recently started to look at tools which help visualise interactions which take place; although I’ve experimented with a few, Netlytics is new to me. Since I lack the coding skills to use those released through GitHub and elsewhere, I’m grateful for those which reduce the barrier to access, as Netlytics appears to do. The key then of course, is being able to interrogate the information that the visualisations and analytics provide to extract meaning from the data … one more area I’ve only recently begun to explore!

In general terms then, and before using Netlytics to answer the questions, I’d expect to be able to focus either on individuals or the group as a whole. I would hope to be able to explore individual learner behaviour, in terms of their communication within the class, and the extent and nature of interactions across the group as a whole. I suspect that the picture will only be partial, since the sample was probably gathered using #cck11? Any tweets related to the class, perhaps between participants, but not using the hashtag, will of course be missing.

SNA will help us to see structure and pathways within the class, and to some extent what is moving through the network, but I’m not yet sure how it might help us see the effects of those exchanges. What are the outcomes? I’m also not sure about visualizing the aggregate of all those exchanges to provide a snapshot. Seeing the communications unfold over time, and how transfers occur dynamically might be illuminating … or perhaps inflict the viewer with too much complexity!

Thanks once again and please forgive the naivety of my observations; I’m very much a #noob!


Hello everyone.

Based on @Dido’s reply I see that the measures you get a pretty standard for network analysis. I’m wondering if we can use SNA to provide input on students learning. For example:

  1. How can we measure student’s engagement with the topic? Is there a measure already that best capture it or should we create one?
  2. What advice can we give to teachers based on the network?

Maastricht University (Netherlands)


Hi again :smile:

Let me do a bit brainstorming about the @gruzd’s question “How would your interpretation change (or wouldn’t) if you knew that Twitter was only one of at least two other CMCs used in the class?” This means that there are other networks regarding the other CMS’s. (Maybe, we should add one more if the lesson has a face to face class opportunity.) Which means that we have a multiple relations network (MRN) with different types of relations regarding different communication mediums. Then we can examine these networks and we can reach some important results, which would definitely change my interpretation. (There might be some people who doesn’t interact at all by using twitter but using the other communication medium.)

I want to add also a concern about the @katerinabohlec uestion. Although the measures seem to be standardized, the assumptions regarding network formation stays as a hidden variable. So, I think that it is not possible to standardize SNA measures for learning even for MOOC’s. I really wonder your opinions.

I am researching on corporate training environments. I would appreciate any “must read” recommendations regarding SNA in training environments (not the mooocs or the formal education context).

Best regards,


Hi Everyone,

Thank you for contributing to our discussion thus far! I really enjoyed reading all of your comments and replies.

If you are just joining us today, there is still time to contribute to the hotseat until this weekend.

However, if you don’t have time to complete a hands-on part of the hotseat, please feel free to use the following live network visualization of cMOOC Twitter network (#cck11) to answer my questions:




Hello Dido,

I’m not so sure. Students who post often (thus have high outdegree) messages containing “high level processing features”(e.g., summarizing of discussion, elaborating on contradictory points etc). have higher grades. But that doesn’t take the lurkers into account who might also be learning. Other studies have shown that degree doesn’t relate to grade.

Regarding your point about network formation, I’m not sure I understand it. I think as a teacher it would be a good tool to see how the interaction network changes. THis could provide input on how to change the lessons.

If we have different communication mediums, you gonna get clusters with every student using his/her preferred medium. It is gonna be more difficult for the teacher to provide guidance as the teacher needs to be visible on all communication channel he/she offers to the students. In addition, I’m not sure these are multiple relation. Isn’t the relationship between the two people still the same (communicating with each other)?

About advice on must reads, do you research formal training environments (e.g., classes?) If not there is a lot of research on advice seeking you might want to look into.



I really like your point about Multiple Relations Network. As far as I can see, Social Networking sites like Facebook and Twitter aren’t in the business of complying with common standards for the common good. They share data how and when it suits them to satisfy user experience and maximise economic value of user contributions. They are what Meijas calls monopsonies - he explains this in the video at the end of this post - Twitter wants to be ‘the’ micro-blogging site and Facebook wants to be ‘the’ social networking site.