A site for Critical Data and Education (an MSCDE course)

Tag: Learning (page 1 of 1)

On ‘Learning’

Block: ‘Learning’ with Data / Summary

This block, I tracked different bodily data while learning and working – my posture, movement, and reflected gaze. I wanted to experiment with the encoding of the body through data as a way to explore the turn towards bodily and emotional data in edtech (Knox et al. 2020). If the edtech imaginary treats the failings of educational systems as “engineering problems to be solved at scale” (Friesen 2019, p. 144), then the turn towards bodily data posits our bodies as part of the ‘problem’ to be ‘solved’ by edtech and learning analytics. This is an obvious and problematic endpoint of data practices based on behavioural approaches – or, at least obvious to anyone who has experienced similar exertions of control over their bodies based on their gender, disability, sex or race.

I didn’t set any clear questions to be answered in this block, choosing to explore and think about the construction and curation of bodily data instead. But my experimenting so far feels only surface level and, despite inspiration from the ‘Dear Data’ project, I struggled to track and visualise data in ways that I felt reflected the complexities of intra-action between the mind-body and data while learning or working (Rogowska-Stangret 2017). Still, I posited that a ‘knowing’ awareness of the methods of data collection and visualisation could provide a site of performative resistance to surveillance technologies and learning analytics, in a similar way to how a knowing awareness of gender performance informs drag acts.

There are some obvious problems with this position. Firstly, the increasing complexity of learning algorithms work to obscure from students the ways in which their learning activities and bodies are datafied and shaped, as does the increasing insertion of behavioural interventions and nudging across the educational landscape (Knox et al. 2020). Secondly, existing power relationships and imbalances in complex educational systems (Tsai et al. 2020) limit the ability of students to enact their agency in even performative ways. My position comes from three weeks of self tracking data, a learning activity that encouraged my own agency and power over the data I curated. There are evidently different power dynamics at play when a student is sitting a high stakes exam with mandatory online proctoring, or having to agree to terms of use in order to matriculate (Tsai et al. 2020), and this limits their ability to resist or question the ways in which their data will be collected and used.

The act of self-tracking made me think about how I ‘pay attention’ to learning activity and data production and when I’m happy to let a machine ‘pay attention’ on my behalf. Can an awareness of the rules of the game, so to speak, help develop learners’ empowerment over their data? Although increased transparency and personalisation of data (for example, through personalised data dashboards) might help to develop a ‘knowingness’ of data practices, we cannot automatically assume that this will lead to student or teacher empowerment over their learning (Bulger 2016, Tsai et al. 2020). Assuming also that teachers should explain to students how their data is being collected ignores the ways in which power dynamics in educational systems and complex data processes obscure teachers’ comprehension of data practices as well. 

It’s these tensions and complexities I’ll carry in the back of my mind as we move into our next block on ‘teaching’ with data.

References

Bulger, M, 2016, Personalized learning: The conversations we’re not havingData and Society22(1), pp. 1-29.

Friesen, N 2019, “The technological imaginary in education, or: Myth and enlightenment in ‘Personalised Learning.” In M. Stocchetti (Ed.), The digital age and its discontents. University of Helsinki Press, pp. 141-149.

Knox, J, Williamson, B & Bayne, S 2019, ‘Machine behaviourism: Future visions of “learnification” and “datafication” across humans and digital technologies‘, Learning, Media and Technology, 45(1), pp. 1-15.

Rogowska-Stangret, M 2017, Body, viewed 13 February 2020, <https://newmaterialism.eu/almanac/body/body.html>.

Tsai, Y-S, Perrotta, C & Gašević, D 2020, Empowering learners with personalised learning approaches? Agency, equity and transparency in the context of learning analytics, Assessment & Evaluation in Higher Education, 45(4), pp. 554-567, DOI: 10.1080/02602938.2019.1676396

Three

Block: ‘Learning’ with Data / Week 5

This week, I recorded all the times my image was reflected back to me by software while I worked. This reflection is through Zoom and Microsoft Teams while I work from home. I attempted to classify these instances by the type of activity, whether my camera was on or off (with profile picture on display), and audience/participants by number and type (internal team members or faculty). For background, I am a learning technologist, and these measures are a personal codification of my work this week.

I plotted these instances along a subjective “fidget meter”. After each Zoom or Teams session, I noted how often I thought I fidgeted (made movements with my hands, arms, face and back) during that time.

Figure 1: The self, reflected and fidgeting.

Some things I learned:

  1. I appear to fidget the most during internal team meetings.
  2. I appear to fidget the least when I’m working with faculty members in consultations or leading workshops.
  3. The main way I appear to meet with faculty is through consultations and workshops, not meetings.
  4. I do not turn my camera off while leading consultations or workshops.

Fidgeting is used as an indicator of lower engagement by some algorithms (Chang et al. 2018), so you could say that I’m not very engaged in most meetings. However, the session in which I fidgeted the most was a meeting of our team’s “journal club”. I might have fidgeted a lot, but I was also actively participating in the discussion. I think I fidgeted a lot because I was with people I was comfortable with, discussing a more casual topic. Compared to working with faculty academics in workshops or consultations, where I have to act a certain way, or be more ‘switched on’.

I use the phrase ‘switched on’ to imply a different level of engagement, but also a performance. By paying attention to my fidgeting this week I likely fidgeted more or less often at times, performing for my data visualisation, my reflection and my real time audience. The nature of meeting people through software like Zoom and Teams, which reflect your image back to you (even with your camera off, you may display a profile picture or your name), also appear to encourage an awareness of self image that is different from offline meetings.

This awareness or gaze uncovers a performative aspect to the surveillance of bodily data that resists the idea that ‘correct’ forms of engagement can be measured through the body. If we are aware that bodily data, like other learning analytics measures, serve as “proxies for, but not accurate representations of, attentional focus” (Bulger 2016, p.16), we can perhaps find ways to perform focus to surveillance algorithms in queer, resisting or dissembling ways.

References

Bulger, M., 2016. Personalized learning: The conversations we’re not havingData and Society22(1), pp.1-29.

Chang, C., Zhang, C., Chen, L. and Liu, Y., 2018, October. An ensemble model using face and body tracking for engagement detection. In Proceedings of the 20th ACM International Conference on Multimodal Interaction (pp. 616-622).

Two

Block: ‘Learning’ with Data / Week 4

This week, I continued my exploration of “bodily” learning data by recording the times I took a break from learning, working, leisure or caring. These needed to be breaks that I would pay attention to and record (I might not notice that I look out of a window, or open a new tab in my browser). So I recorded each time I stopped an activity to stretch.

This week was another sedentary week. Before recording, I sorted the stretches I expected to perform into three categories: hands/wrists, back/neck/shoulders, and hips/legs/feet. These are represented by shapes. I also sorted the expected activities I would perform this week into 4 categories: work/study, leisure, care (for self and others), and rest. These are represented by colours. The time of the stretches and activities were recorded throughout the week.

Figure 1. Tracking my stretches over a week.

On Tuesday, I downloaded a Stretch Reminder app to see if an app could “nudge” me to stretch more often. Shapes with a line represent the times I was “nudged” to stretch. Reminders were sent via push notifications every hour. You can see I often ignored the reminders.

At the end of the week, I drew a gender-neutral body to plot my stretches on. While I considered using an outline of my own body, to avoid an assumption of a ‘typical’ human body, or perhaps an “ideal learner” Eynon describes (2015, p. 408), I’m not quite ready to share my own silhouette here. But by drawing over the body with colourful representations of my stretches I think I was still able to reflect the messiness and individuality of embodied activity.

Like last week, I like how the act of animating data visualisations can help draw attention to the construction of the visualisation.

Video 1: I hand draw stretch tracking data on an outline of a body.

As I worked on this I reflected upon my activity. Firstly, I learnt that “nudging” via app doesn’t have any impact on the number of times I stretch per day. Would “nudging” from learning analytics algorithms impact my learning activity? I’m not sure it would after this. Secondly, I felt the need to collect more comparative data – I would like to compare this sedentary week to a more active week. Lastly, I heed Enyon’s warning about interpreting this data with an understanding of what it is actually capturing and not to invest too much into it (2015, p. 409). This data counts the stretches I performed, but not the quality of the stretching.

This data could be interpreted in different ways by humans and machines. To me, it’s perhaps a subjective count of the times I stopped to care for myself, but also the pains I felt while performing different activities.

To different algorithms, the data could look different again. If I were being monitored by proctoring software, it could be the times I potentially cheated; MS Teams might interpret this as the times I stopped learning or being productive; to the Stretch Reminder app, it’s the times I did or didn’t complete desired user behaviour.

Data is never neutral, in its production or construction, and the quantifying of self through data can feel empowering and restrictive at the same time.

References

Eynon, R., 2015, The quantified self for learning: critical questions for education, Learning, Media and Technology, 40(4), pp. 407-411.

One

Block: ‘Learning’ with Data / Week 3

This week, I decided to start recording some very simple data. I recorded the general position of my body while I read articles or books (for leisure, work and study) throughout the week, using stick figures as my data points.

Below is my data collection between January 25-29 (including some test runs), a legend, and my totals at the end of the week.

As the week went on, I was tempted to record more data to explain the stick figures I was recording. Very hot weather and busy work days shaped this data, but this isn’t always explained in the data. The temptation to record more data to explain the data on hand is perhaps not an uncommon temptation.

Another frustration I found was that, by adhering to plain stick figures and recording 30-minute blocks of time, my reading body appears to be more static than it is. My data couldn’t show the all the fidgeting and readjusting my body made. I felt like I had chosen a very reductive way of recording the human body – one that doesn’t show my weight, ability, gender or skin. This reduction feels both liberatory and restrictive to me, tensions which are similarly explored in John Phillip Sage’s Data Drag project. While this exercise allows me to choose how I track my body through data, I was still unhappy with the representation I chose. Nevertheless I tried to stick to the original parameters I had set out, of simple stick figures, to sit in that ambivalent feeling.

When it came to visualising the data, I didn’t want to create a static image. So I made a quick flip book animation (below). If I kept recording these stick figures I could create a longer and more engaging animation/visualisation.

Knox et al. suggest that “a substantial interest in new education technology development appears to be towards ‘bodily’ and ’emotional’ data” (2019, p. 42), and in light of this I want to spend the rest of this block continuing to visualise data around the embodied experience of learning. When I think about machine tracking of bodily data in education, I think immediately of proctoring software that tracks eye and body movements for surveillance purposes. Can I record and visualise data about my body that doesn’t replicate this kind of surveillance, but is more in the spirit of Data Drag and a (queer) quantified self? That might require some unlearning for me.

References

Knox, J, Williamson, B & Bayne, S 2019, ‘Machine behaviourism: Future visions of “learnification” and “datafication” across humans and digital technologies‘, Learning, Media and Technology, 45(1), pp. 1-15.

Sage, J.P. 2018, Data drag, viewed January 30 2021, <https://www.johnphilipsage.com/datadrag.html>.