Power for an independent-measures t-test

Choose two normal populations, sample sizes, alpha, and test type. Each simulated experiment draws two samples, runs a t-test, stores the p-value, and updates the running rejection rate.

Experiments

Last p-value

—

Simulated power

—

Predicted power

—

Effect size (d)

—

Population distributions. Dots on the x-axis show the most recent samples.

Evolving distribution of p-values. Red bars are rejected tests; green bars are non-rejected tests.