As part of some recent research and a subsequent publication, I was asked by reviewers to indicate how “temporality played a role in participant comments.” I quickly went back to my stats books and notes from my doctoral work and found…nothing. 🙂
This led me to the start of a broader research project over the summer and into the current year as I explore digitally native research methods, and social media analytics. I enrolled (and finished!!!) a MOOC on Social Media Analytics from the great researchers at the Digital Media Research Center at Queensland University of Technology.
In this post I wanted to share some of what I learned and created as I’ve been playing over the last week. What I learned and built in the MOOC is only as good as the time spent playing and building. So..I dug into some social media to see what I could learn and create.
Equity Unbound is described as an “emergent, collaborative curriculum which aims to create equity-focused, open, connected, intercultural learning experiences across classes, countries and contexts.” This project is led by Equity Unbound was initiated by Maha Bali (American University in Cairo, Egypt), Catherine Cronin (National University of Ireland, Galway), and Mia Zamora (Kean University, NJ, USA) for use in their courses this term (September-December 2018).
For those of you that have never heard of this idea before, this is an idea that I played with in the past with the CLMOOC, and my work with the WalkMyWorld Project. To level up my own skills, and make sure I was always better with tech that my students, I would play each spring/summer with others in the CLMOOC. WalkMyWorld originated as a group of colleagues in teacher ed would share a common curriculum, or series of assured learning experiences that we’d have students complete and share with each other online. In these experiences, we’re bridging the distance between local and global learning in a third space created by hybrid and online technologies.
Data Collection & Analysis
For this original attempt as social media analytics, I used the TAGS explorer to pull all content from Twitter using the #UnboundEq hashtag. I decided to not include the Twitter username for UnboundEq as well as the website for the learning experience. I also may have lost some messages and content if users didn’t include the hashtag, or did not use it correctly. Each of these are decisions that impact my data collection, and ultimately my findings.
This process resulted in 990 tweets collected, with the first tweet happening on 9/10/2018 and the last tweet (in this initial dataset) being collected on 9/21/2018. In terms of full transparency (and to promote open scholarship and open data) you can view this dataset here. This initial output is not that interesting to me (IMHO), other than to share with others to discuss the digital residue we leave behind in these social interactions.
I then pulled this dataset into Tableau in order to further unpack the data through the process of data visualization. Data visualization describes a process and a set of tools used to help individuals understand the significance of data by placing it in a visual context. As an example, looking at the Google Spreadsheet with the output from TAGS I described as “not being that interesting.” It was not until I visualized that data did several elements pop out immediately that I wanted to drill down into to make sense of events. My initial scroll through a spreadsheet, comments, and time stamps was mind-numbing. But, visually, I could see another picture. Sorry for the lame pun…I’m moving on.
In subsequent posts, I’ll share more about my process and settings in Tableau. I’m still learning how to use these tools, and that is not the purpose of this post.
I brought the output from TAGS into Tableau and visualized this information and broke this down based on (DIMENSIONS) Distinct Counts of Tweets [id str (count (distinct))] and (COLUMNS) time broken down to the hour [HOUR(Time)]. This gave me the following visualization.
As I soon as I pulled up this graph, my first question was…”what happened in those two spikes on 9/17/2018?” Using Tableau, I was able to pull the raw data from these time points and export for further analysis. In terms of full transparency, I shared this data set here. I could bring this dataset into Tableau (or other tools) and continue to unpack the findings. I did a quick visual scan to look at the comments, links, and tweets. Once again, my focus in this work was to play, explore temporality in context, and conduct some social media analytics research.
To further explore temporal patterns, I used the same descriptors listed above for Tableau, I then created a new way to process data by creating a “calculated field” for Tweet types. This would allow me to differentiate between original tweets, mentions, and retweets (RTs). I added this new calculated field as a new color and line in the Dimension output. This gave me the following visualization.
This exploration of temporality in the data and discussion helps me think a bit more about the role of these new tools and lenses in my research. My key focus is on looking at time and interactions in this discussion. As a researcher, I would next start to look at content in these discussions, but once again I’m wondering if I’m digitizing traditional research methodologies, or exploring digitally native research techniques.
My next step will be to continue to explore temporality in this dataset and future datasets. I’ve modified TAGS to continue to collect data during UnboundEq, and I’ll continue to share.
I’ll use Tableau to examine retweets and their role in the discourse community. I’ll also examine user activity and visibility, as well as follower and followee metrics in the community.
For now, this has been a great opportunity to think about my own practice, and try to share my learning with others to think about what opportunities are available. I think this work also provokes questions about the digital residue and connections we leave behind.
Also published on Medium.