Simulation Lab guide
What the simulator is for, what model it uses, and how to interpret its outputs without over-claiming.
What it does
The Simulation Lab is a model-based interface for exploring how three factors
(adult education, attendance, and income)
relate to predicted school-level outcomes. It is useful when visual inspection alone cannot separate overlapping influences.
Current scope note: this interface uses the income-based model for continuity. Poverty-aware model checks are reported in
the Statistical Findings and results overview pages.
Open Simulation Lab
Model behind the interface
- Participant-weighted linear model at school-row level.
- Pairwise interactions: education x attendance, education x income, attendance x income.
- Prediction and delta outputs include 95% uncertainty intervals.
- All outputs are associational summaries of aggregate data, not student-level causal effects.
How to read it
- Curves/Bars/Heatmap/Small multiples are alternate views of the same fitted model.
- Scenario prediction is the model estimate at your selected slider values.
- Scenario delta compares your scenario against baseline medians.
- Local sensitivity reports estimated local change in outcome per +1 point in each input.
- If the delta is small relative to uncertainty, treat it as inconclusive.
Guardrail: simulation outputs are best used to test hypotheses and compare relative signal strength,
not to make deterministic school judgments.