A basketball coach asks half his players to stretch before each game; the other players do not stretch. After a month of games, she analyzes their playing stats. She concludes the stretching helped improve their performance on the court. How can the coach test the means to be sure the results not likely to happen by chance? She can pick players at random and analyze their playing stats without calculating means.

She can analyze the two groups by calculating the means and the difference of the means.

She can rerandomize the results by separating the two groups randomly, then calculate the means and the difference of the means.

She can pick the players she wants to analyze based on the best stats.