A Deep Dive into �Student Behaviour
years
behind in Maths by �the end of Grade 9
Spaull & Kotze, 2015
4
How can they �catch up?
Our Intervention Process
6
Impact evaluation
5
Continual monitoring
2
Individual feedback
4
Support & Recognition
1
Diagnostic assessment
3
Personalised content
Identify individual learning backlogs
Activate student engagement
Encourage students �and sustain motivation
Target gaps and�build metacognition
Impactful
tool
Consistent usage
X
=
Improved understanding
Increased grade-level marks
time
4
3
2
1
Our Impact Model
Can we build a �predictive measure of �learning outcomes?
Ave. learning time | Ave. grades of improvement |
10 hours | 0.5 |
20 hours | 1.0 |
30 hours | 1.4 |
40 hours | 1.8 |
Predicting learning outcomes
Correlation = 0.85
95% confidence interval
1.8 hours = 1 month improvement
Ave. learning time | Ave. grades of improvement |
10 hours | 0.6 |
20 hours | 1.1 |
30 hours | 1.5 |
40 hours | 2.0 |
Predicting learning outcomes
Correlation = 0.67
95% confidence interval
1.6 hours = 1 month improvement
How can we use data to �support students to engage?
Student Profiling
Learning time
(minutes)
% Learning time of App time
10
30
60
100%
50%
80%
Timewaster
Avoider
Dedicated
Uncommitted
High-flyer
Focused
No show
Example (Class 8D, Sep 2024)
Learning time
(minutes)
% Learning time of App time
10
30
60
100%
50%
80%
Timewaster
Avoider
Dedicated
Uncommitted
High-flyer
Focused
No show
12%
6%
18%
27%
12%
6%
18%
Responding to Profiles
Ave. learning time | Ave. grades of improvement |
10 hours | 0.6 |
20 hours | 1.1 |
30 hours | 1.5 |
40 hours | 2.0 |
Predicting learning outcomes
Correlation = 0.67
95% confidence interval
1.6 hours = 1 month improvement
A Deep Dive into �Student Behaviour