Lecture on Learning Analytics

Introduction

• Former high school physics teacher.
• Focus: Educational research and learning analytics.
• Aim: Provide a high-level overview of learning analytics in the next 10-15 minutes.
• Emphasis on actionable data for immediate educational improvement.

What is Learning Analytics?

• Definition (2011): Collection and analysis of data traces related to learning to inform and improve the process and/or outcomes.
• Key Feature: Closing the loop – data should be actionable in real-time - actionable for teachers, students, advisors.

Data in Learning Analytics

• *New Developments:
  • Increased ability to collect diverse data.
  • Higher frequency and granularity of data collection.
  • Opportunities to connect different sources of data.

Types of Data (3 As)

1. Activity: Actions taken by students (log files, physical traces, self-reports).
2. Artifacts: Things created by students (answers, steps in problem-solving).
3. Association: Connections made by students (interactions, temporal linkages).

Caveats

• Big Data Myths:
  • Not neutral – influenced by the creation process.
  • Not natural – requires purposeful processing and planning.
  • Not easy to get good data – design and tracking are crucial.

Design in Learning Analytics

• Data comes second to design.
• Example Study: Wise inquiry learning environment – revisiting steps with dynamic visualizations predicted learning.
• Recent Study Example: Dashboard in inquiry environment – hypothesizing alert for real-time teacher intervention.

Computation in Learning Analytics

1. Prediction:

• Focus: Identifying individuals or features that matter (early alert systems).

2. Structure Discovery:

• Clustering, network analysis, topic modeling (identify patterns).

3. Temporality:

• Lag sequential analysis, hidden markov analysis (understand learning transitions).

4. Visual Analytics:

• Visualize information to discern patterns (heat maps, sand key diagrams).

5. Natural Language Processing (NLP):

• Extract information from student's language (rhetorical moves, argumentation).

Insight and Action

• Model: Learning analytics as a socio-technical system – human decision-making is crucial.
• Process: Questions from instructors, data interpretation, relative and absolute reference points.
• Triangulation and Contextualization: Attributing data accurately; actions based on insight.
• Reflection on Pedagogy: Changing thinking about learning and teaching based on data.

Ethical Considerations

• Ethics & Privacy: Important to handle data responsibly.
• Equity: Addressing algorithmic bias and ensuring fair outcomes.
• Transparency: Ensuring the actionable insights are understood and used correctly.

Conclusion

• New Resource: Learning Analytics Research Network - LA 101 – curated readings, videos, and tutorials.

Final Thoughts

• Potential of learning analytics to change teaching and learning. Importance of informing but not dictating pedagogical decisions.