Skip navigation
All Places > Data and Analytics > Blog > 2016 > January

Data and Analytics

January 2016 Previous month Next month

We reached out to Matt Lewis, who posted information about an LTI tool designed to help teachers and students visualize their participation in Canvas discussions. The tool is called Threadz.


Q: What was the motivation for developing an LTI like this?


Threadz was developed out of the faculty's desire to leverage more from their online discussions. "Conducting a discussion online was a little out of their comfort zones and they were trying to understand more about what was actually happening besides what Canvas was showing in Speedgrader," Lewis said.


In their search to find a solution, the development team looked at integrating SNAPP (Social Network Adapting Pedagogical Practice), a tool that performs real-time social network analysis and visualization of discussion forum activity within popular commercial and open source Learning Management Systems (LMS). "For various reasons, this tool proved too difficult to integrate into Canvas, so we built our own," Lewis concluded. "We knew we wanted to tap into the API to extract all the discussion data, and a LTI would help facilitate that." Although the SNAPP tool was never used, the team relied on SNAPP research regarding the use of social network data and learning analytics.


Once enabled in a course, Threadz can be utilized by any student and/or teacher enrolled in that course. "It provides valuable information about the constructs of the social network patterns for individual discussions," says Lewis. "These patterns can help highlight successful student behavior characteristics about student interactions within a discussion as well as assist the instructor with adjustments to the course’s learning environment to improve student success."


Q. What information are teachers finding to be most valuable?


By assessing the social network connections in a discussion, teachers can hone in on patterns that emerge related to the activities and behaviors of individuals and groups. An instructor can use these patterns to identify behaviors and characteristics within the course, such as:

  • learner isolation
  • non-integrated groups
  • instructor-centric discussions
  • key integration (power) users and groups.



"I think the classic network diagram provides a good overview about who is influencing the discussion and to a certain level the amount a student is engaged with others," Lewis points out. "It is easy-to-read at-a-glance and is intuitive to most instructors to understand."



The chord diagram however provides more information about the direction and count of the interactions between two people. "While maybe not as familiar, the chord diagram can provide a more complete look at a discussion," says Lewis.


Q: How are teachers leveraging this information about their discussions?


Threadz provides information about which students might be isolated within the group or which groups are dominating a discussion. Instructors use this information to determine when to intervene and interact with individuals or groups. "A forgotten element within discussions is the impact of instructor interactions," stresses Lewis. "Threadz assists the instructor in understanding if they are too involved in leading the discussion or too hands-off."


While Threadz does not provide any qualitative insights on whether a discussion is good or bad, it certainly adds another layer of assessment to course discussions. Teachers are empowered to leverage this data to design and/or modify discussions, accordingly. On the flip side, students can also self-assess their own involvement in course discussions and determine the frequency and intensity in which they are participating in their course discussions.


Q: What's next for Threadz?


Sentiment analysis is certainly not off the table, but may be a bit ambitious at this stage. "Exploring the language of the posts would be very interesting even at a rudimentary level," shared Lewis. "But this is getting a little ahead of the game."


At this time, the development team is looking to add features related to data file downloads and include additional statistics around the ratio of posts and word counts. (Cool, right?!)


Q: How can the Canvas Community be helpful to your team?


The Threadz team is interested in making this tool easy-to-use. Canvas admins should be able to easily install the LTI and students and teachers should be able to intuitively use it. Their team would sincerely appreciate any feedback about it's application and functionality. And welcome all feature suggestions for Threadz.


Thanks to Matt Lewis for your insights. Feel free to reach out to him for additional information.



As teachers, it’s probably safe to assume students prioritize their assignments based on deadlines, right? But how close to deadlines are students submitting their work? And what does this insight reveal about student behaviors and teacher expectations?


One higher ed institution* decided to find out.


At the beginning, this school’s analytics team set out to explore student course participation patterns in Canvas by looking at a typical week and gathering discussion, assignment, and quiz submission activity. But preliminary results revealed patterns, which led to further examination the relationship between submission time and assignment due time.


What parts of Canvas Data were used?

The analytics team leveraged assignment submission dimension and assignment dimension table. Further analysis results were derived from course assignment information and submission data. Online quizzes and the assignments that had an ‘online’ submission type were included in the analysis. The assignments that did not have a due date/time were excluded from the analysis.


What were the results of the analysis?

Most students submitted assignments before due date/time than past due. (That’s good, right?) However, the median submission time before due date was 30 minutes and the median past due submission time was 1.2 hours.


What does this say about student behavior and teacher expectations?

After a few outliers were identified and removed, the median submission time before due date was 30 minutes and the median submission time past due time was 1.2 hours. The chart below shows that the number of before due submissions was much greater than the total number of past due submissions. Also, the variation in past due submission hour was wider than before due submission hour.


The results imply that majority students tended to submit assignments more often before due time than past due time; and, the likely assignment submission time is 30 minutes prior to assignment due time.



The bar chart below revealed that a number of assignments contained due date/time that were set between midnight and 7am Eastern time, which led to some students working overnight to submit the assignment right around its due time. Taking all terms in the year of 2015 into consideration, the evening period from 8 pm to 10 pm was a popular time for assignment submissions, and 10 pm was the peak assignment due time (when assignments were due).


Therefore, a discussion with faculty to carefully consider their assignments’ deadlines might be warranted.


What’s next?

Some interesting follow-up questions might include:


  • Does submission time impact the quality (points/grade) of submission?
  • Does submission time change over the course of the semester?
  • Are more people submitting on time if deadline is morning, night, evening?
  • What is submission grading time? In other words, how quickly are teachers grading assignments after submission?


What other analytics around assignments do you think might be interesting to pursue?


Post your thoughts below!




* School requested to remain anonymous for this post.