How many
is enough?
Sampling Shala helps education and social-sector teams design defensible samples, and teaches the statistics underneath by letting you play with them. Free, offline-capable, and honest about the maths.
You run a programme in 1,200 schools.
State, district, block, school. The cascade is real, and every layer adds people you are responsible for understanding.
You cannot visit them all.
Measuring everyone is a census, and almost nobody can afford one. Time, staff and budget all run out long before the last school does.
Picking the convenient ones lies to you.
A sample drawn from whoever is easiest to reach tends to miss in the same direction every time. A random sample does not.
The maths knows when to stop.
Confidence and margin of error trade off against sample size in a precise, checkable way. This is exactly what Build mode computes for your programme, in five steps. Start with your own numbers
A defensible plan in five steps.
Question, scale, who, precision, strategy. Multi-stakeholder designs, an LQAS decision-rule engine, and the binding constraint, computed for you.
Play until it clicks.
Interactive playgrounds, a concept library with a visual for every idea, stories from real evaluations, and a living practice district.
Drill until it sticks.
Short timed modules that turn concepts into reflexes, built for busy field teams.
Describe your programme in plain words.
AI reads your context and drafts the setup. It never computes your sample size; the verified formulas do that.
Enter →- MEL practitioners who need to defend a sample to a donor by Friday.
- Young and emerging evaluators who want the intuition, not just the formula.
- Grassroots NGOs doing serious work on modest budgets and modest phones.
What question is this evaluation answering?
This is the start of the pathway. Each picture shows how that design looks in the field; your choice shapes every step that follows.
How big is the programme?
Education programmes live on a cascade: nation → state → district → block → school → classroom. Pick a starting scale, then fine-tune every number. This sets the universe and pre-fills the stakeholder populations you can later override.
Who must this evaluation hear from?
An education evaluation is several samples, not one. Tap everyone who matters. For each, Sampling Shala recommends a method and pre-fills a population estimate from your cascade. Edit the population to your real frame, and choose whether that group needs a full statistical sample or a lighter plan.
Precision: tuned once for rigor, once per group for reality
Your integrated sampling strategy
Every group, every method, reconciled into one field plan. Go back and nudge anything to see it respond.
Facilitator training
Ready-made sessions for running a workshop or orientation on sampling. Pick a module, project the slides, and use the built-in discussion prompts and scenarios to guide the room.