We intend to analyze the online coversation patterns between our Learning Analytics course students looking at the Herodotou et al., 2019 reading on predictive LA and a video webinar on networks.
The key data we have here are: the names of students, their specific comments, which gives us the number of messages posted, number of replies for each of those, and the number of upvoters for each of those. We also know the range (where the comment was made) and the date of post.
To analyze the conversation patterns, we consider the nodes as students, and the edges to be online interactions between them on the reading or video.
Summary of the Herodotou Reading Online Conversations
Messages_Reading
Summary of the Video Webinar Online Conversations
Messages_Video
Now, let's take a look at the Reading Network visualization to analyze the patterns there.
Reading Network Visualization
In the visualization, the node sizes denote the number of messages, with the biggest nodes implying the maximum number of messages, while the node colors denote the number of replies. The visualization is undirected, capturing replies from both sides.
As we can see from the visualization, the key participants in the reading network based on their number of posts on the reading are: Wilhelm Schumann, Madhu G, Iron Kim, Pranali Mansukhani, and Ethan Moltz. The graph seems pretty connected. The central nodes in the connected components indicate that Wilhelm Schumann, Pranali Mansukhani, Madhu G, and Xiaomeng Huang have a high number of interactions in the form of either replies made to others' posts, or replies made to them. For example, while Iron Kim and Ethan have a high number of messages, the number of interactions are fewer.
Now, let's take a look at the Video Network visualization to analyze the patterns there.
Video Network Visualization
In the visualization, the node sizes denote the number of messages, with the biggest nodes implying the maximum number of messages, while the node colors denote the number of replies. The visualization is undirected, capturing replies from both sides.
As we can see from the visualization, the key participants in the reading network based on their number of posts on the reading are: Fanjie Li, Pranali Mansukhani, Xiaomeng Huang, and Madhu G. The central nodes in the connected components indicate that the same students Pranali Mansukhani, Fanjie Li, Xiaomeng Huang, and Madhu G have a high number of interactions in the form of either replies made to others' posts, or replies made to them.
Compared to the reading, there are both fewer posts and fewer interactions on the video overall. This could be possibly because the video seemed more straightforward and there were fewer points of debate and discussion on the video, compared to the reading.
Let's also take a look at the Overall Network visualization using the total messages and replies to analyze the patterns there.
Overall Network Visualization
The overall network visualization shows similar central nodes in the connected components, indicating that student behaviors did not change much relative to their own behavior in either video or reading. Students who participated centrally in both, also made a relatively high number of comments.
There are two types of student reading behaviors here:
Students who made a high number of relevant posts that evoked a high number of replies. These behaviors could often lead to these students to be considered as the "initiators," who initiated conversations early on encouraging others to participate.
The second type could be students who made a high number of replies to already existing posts. These behaviors could lead to students being considered as "listeners" exhibiting an enhanced listening behavior and contributing to the online talk by adding agreements or counterarguments. This type of behavior would logically be exhibited by students who started contributing the online talk, later in time.