Loading the tool…

Experiment
Simulator

Run repeated trials, compare observations
with theory, and share an experiment
someone else can reproduce.

Ten flips, a thousand times.

Success = heads
xorshift32 generator

Exactly 5 heads

Observed · 235 of 1000

23.50%

Theoretical probability

24.6094%

Finite runs vary.

No forced match to theory.

Observed frequencyTheoretical probability

First experiment:

6 heads in 10 flips

Simulated results · seed 2026 reproduces this run; finite observations may differ from theory.

Share a run someone can repeat.

Keep the settings, seed and finite-run observations together.

Make the most of your experiment.

1

Keep a model in mind

Check fixed n, independence
and the same p each trial.

2

Expect variation

A finite run does not exactly
equal theoretical probabilities.

3

Record the seed

Save settings and generator
so someone can repeat the run.

Questions, Answered.

Why does a run differ from the exact probability?

A finite set of simulated observations varies. Exact theory describes the model distribution; it is not a promise about a particular finite run.

What does the seed control?

The nonzero unsigned32 seed starts the xorshift32 generator. Keeping the same n, percentage, repetitions, seed and generator reproduces the same outcomes.

Does more repetition remove all variation?

No. More repetitions often make observed proportions more stable, but a finite run is not forced to match theory.

Can this check independence in real data?

No. This educational simulation assumes independent Bernoulli trials. It does not establish independence or model fit for measured data.