Modelled research study - synthetic data
Schools increasingly have access to embedded analytics tools like Microsoft Teams Reflect and Insights, but limited guidance exists on how to use them for wellbeing-first decision making rather than purely performance tracking. This project investigates how these tools could be leveraged to support student emotional wellbeing, engagement tracking, and digital collaboration in a secondary school context. Using a structured synthetic dataset modelled on real-world patterns from NSW Showcase Schools, I analysed realistic analytics across 100 modelled participants, students, teachers, and coordinators to evaluate trends in emotional check-ins, engagement levels, and collaboration patterns.
Synthetic data design
Designed a structured synthetic dataset of 100 modelled participants across three roles, students, teachers, and coordinators informed by real-world usage patterns from NSW Showcase Schools, while avoiding any use of actual student data.
Metric framework definition
Defined four core metrics for analysis, weekly Reflect usage, average wellbeing score, engagement percentage, and collaboration instances to enable meaningful comparison across roles and over time.
Analysis in Excel
Applied descriptive statistics, outlier identification, and persona comparison techniques in Microsoft Excel to surface patterns across the modelled dataset.
AI-assisted interpretation
Used ChatGPT and Bing AI to support interpretation of patterns and to stress-test findings against existing research on student wellbeing and digital engagement.
Recommendations development
Translated findings into actionable, ethically grounded recommendations for school leadership consideration, balancing emotional tracking with student agency.
Average wellbeing score across the modelled dataset was 7.14 out of 10
Teachers showed the highest Reflect use and collaboration levels of the three personas
Students with no Reflect usage had the lowest engagement, below 50%
Increased check-ins correlated with better engagement and collaboration patterns
Findings pointed to a need for staff training and boundary-setting around digital load
Embed Reflect use into weekly school routines
Offer data literacy training to teachers and coordinators
Use Insights dashboards to identify early signs of disengagement
Introduce wellbeing-led alerts for at-risk students
Design policies that balance emotional tracking with student agency
This project sharpened my thinking about the ethics of analytics in education, specifically, how to use data-driven design to inform decision-making without compromising student privacy or agency. Working with a modelled rather than real dataset forced me to think rigorously about what good wellbeing analytics should actually look like before any real implementation, rather than retrofitting ethics after the fact. It's a methodology I'd bring to any future learning design role where data informs design decisions model first, validate carefully, and always design with student agency at the centre.