Controlling the error:

• Goal: As little error as possible in our decision making procedure.
• For a fixed sample size we can’t control both Type I error (α) and Type II error (β).
• Problem: Type I error and Type II error vary inversely.
• Consequence: We have to choose which type of error we want to control. Therefore, we want to control the error whose mistake will be more “costly”, where “cost” is reflected in more negative consequences.