An industrial engineer presented two types of stimuli (two-dimensional and three-dimensional films) of two different jobs (1 and 2) to each of five analysts. Each analyst was presented each job

stimulus film twice, and he order of the whole experiment was considered to be completely randomized. The engineer was interested in the consistency of analyst ratings of four sequences within each job stimulus presentation. Since log variance is more likely to be normally distributed than variance, the log variance was recorded for the four sequences. The following tabulation resulted.