Using Learning Analytics to Illuminate Student Learning Pathways in an Online Fraction Game
-
Upload
nicole-forsgren -
Category
Data & Analytics
-
view
502 -
download
1
description
Transcript of Using Learning Analytics to Illuminate Student Learning Pathways in an Online Fraction Game
Using Learning Analytics to Illuminate Student Learning Pathways in an Online Fraction Game
Taylor Martin, Nicole Forsgren VelasquezActive Learning Lab, Huntsman School of Business
Utah State University
The Opportunity
• The new microscope• Rich and growing streams of digital
learning data• Better measures of learning and teaching
Teaching Fractions
http://games.cs.washington.edu/Refraction/
Visualizing Game States
Learning Gains
• Results: Students improve (pre to post) after playing game
• But… – Visualizations suggest different strategies– What about personalized learning?
• To investigate different strategies, we use cluster analysis
Cluster Analysis
• Variables– Number of unique board states– Total number of board states– Average time per board state– Number of moves until initial 1/3 board state– Success on game level
• Results: 5 clusters (fussing strategies)– Duncan’s Multiple Range Test used to
interpret
Cluster 1: Minimal
• Clustering variables– Number of unique board states: Low– Total number of board states: Low– Average time per board state: Very High– # moves until initial 1/3 board state: Very High– Success on game level: Low
Minimal
Cluster 2: Haphazard
• Clustering variables– Number of unique board states: Medium– Total number of board states: Very High– Average time per board state: Low– # moves until initial 1/3 board state: Very High– Success on game level: Low
Haphazard
Cluster 3: Explorer
• Clustering variables– Number of unique board states: High– Total number of board states: Medium– Average time per board state: High– # moves until initial 1/3 board state: High– Success on game level: Medium
Explorer
Cluster 4: Strategic Explorer
• Contrast to Haphazard• Clustering variables
– Number of unique board states: Very High– Total number of board states: High– Average time per board state: Very Low– # moves until initial 1/3 board state: Medium– Success on game level: High
Strategic Explorer
Cluster 5: Careful
• Contrast to Minimal• Clustering variables
– Number of unique board states: Low– Total number of board states: Very Low– Average time per board state: Medium– # moves until initial 1/3 board state: Low– Success on game level: Very High
Careful
Learning Gains: Transfer
• Posttest transfer score not associated with strategy
• Strategy used is related to learning• If prior knowledge is medium or better:
– Explorer strategy learned the most – All high-fussing strategies (strategic explorers,
explorers, haphazard) were good
• If prior knowledge is low:– Minimal strategy was better than Haphazard– High fussing is counterproductive
Initial Conclusions
• Fussing at a medium level productive• Careful (non fussing) strategies can be
productive, particularly with low prior knowledge
• Students with low prior knowledge may benefit from directed activities or hints
Next Steps• Towards Adaptivity
– What degree of fussing? – When?– For whom?
• Process Analytics– Identify exploration sequences
Thank You!
• activelearninglab.org• [email protected]• [email protected]