Next chapter: Measuring effect size

Using the formula for the binomial distribution, the probability of obtaining 7 or more "successes" with N = 8 and π = 0.50 can be calculated to be 0.0352 as follows:

Therefore, P(7) + P(8) = 0.03125 + 0.00391 = 0.0352. The one-tailed probability value is therefore 0.03516; The two-tailed probability value is (2)(0.03516) = 0.0732.

Technically speaking, the sign test is a test of whether the median is a particular value. In this example, the test is whether the median difference between conditions is zero. The sign test is generally less powerful than a t test of the difference between dependent means. For the example data, the t test would be: t(7) = -3.21. The two-tailed probability value for this t is 0.015.

Next chapter: Measuring effect size