Question

g Suppose that twenty different hypothesis tests for whether jellybeans cause acne are conducted. In order that the probability of one or more type I error between these should be at most 0.05, at most what significance level should be used for each of them?

Answer 1

The level of significance to be used is α = 0.0025

Explanation:

Here, we are interested in calculating the the level of significance which at most must be used for each of the hypothesis test

We proceed as follows;

P(type 1 error) = α

From the question, n = number of hypotheses = 20

P( of one or more type one error) ≤ 0.05

1- P(no type one error) ≤ 0.05

Hence;

1- (1-α)^20 ≤ 0.05

(1-α)^20 ≥ 0.95

1- α ≥ 0.997438621223

α ≤ 0.00256

Thus α = 0.0025