About InferForge
InferForge makes statistical inference visible. It targets the ideas AP-Statistics students find least intuitive — the sampling distribution of the mean (the Central Limit Theorem) and what a confidence interval actually means — by letting you run the simulation yourself instead of memorising a formula. It picks up where ChanceForge (probability, ages 9–14) leaves off.
Two simulators
- Sampling distribution — pick a population shape and a sample size, predict how spread out the averages will be, then run 300 samples and watch the distribution of the mean build. See it get tighter and bell-shaped as n grows, even from a lopsided population.
- Confidence coverage — build 100 confidence intervals and count how many contain the true average. That count — about 95 — is what “95% confidence” really means.
How it works — and stays honest
- A deterministic engine runs every simulation. The populations are fixed distributions, so the true mean and spread are computed exactly; the samples are seeded and reproducible. No AI is ever near the statistics.
- On-device, no accounts. Nothing is sent anywhere; nothing is stored beyond this page.
- Simulation, not a substitute for the course. This builds intuition for the concepts. It uses a z interval for a clean large-sample demo; a full course would use the t distribution for small samples.
Honest note: this is a tool to see how inference behaves. It is not a claim about test scores, and it is not a replacement for a statistics course.