Confidence Interval Explorer
The population mean μ is fixed. Each sample produces one interval; the interval is the random thing. Draw many and count how often the procedure captured μ.
Press Draw to take a sample and compute its interval.
Expected half-width vs n (this level, this population)
How to read this
Larger n → narrower intervals (half-width shrinks with 1/√n). Higher level → wider intervals. Over many samples about 95 % of the intervals cover μ when the method's assumptions hold.
A single interval either contains μ or it does not. '95 %' describes the procedure over repeated samples, under the stated model and design — not the probability that this interval is right.
Synthetic finite population N = 2 000 (seeded; same generator as the Sampling Universe). Coverage shown is a finite-simulation estimate: with 100 intervals the count itself has a standard error of about 2 percentage points at 95 %.
0 intervals, coverage —.
Formula
A single interval either contains μ or it does not. '95 %' describes the procedure over repeated samples, under the stated model and design — not the probability that this interval is right.