Two events are independent when P(AB) = P(A)P(B). For three events, mutual independence asks for this for each of the three pairs and P(ABC) = P(A)P(B)P(C): four equations, none of which follows from the others. Click outcomes to put them into A, B and C (all outcomes are equally likely), read off which equations hold exactly, and watch simulated relative frequencies: the differences between the frequency of an intersection and the product of the frequencies settle on the exact differences, which are 0 exactly when the equation holds.
The grey band is ±0.5/√n. For two independent events the frequency difference P̂(AB) − P̂(A)P̂(B) has standard deviation close to √(pA(1−pA)pB(1−pB)/n) ≤ 0.25/√n, so for independent pairs the curves stay inside the band about 95% of the time; a difference that leaves it for good is a real dependence.
code/ch01_probability.py (independence_table, classify_independence, bernstein_events, triple_product_events, indep_demo_sim, supplementary Figures 1.S3 and 1.S7).