6. UAV switch experiment, continued The UAV switch experiment, introduced in Sect. 19.8.3, was run using a 2×2 split-plot design with 16 subjects as whole plots, with two levels of alert type (A) assigned to whole-plots, and with two levels of complexity (B) assigned to split plots. Main effects and interactions for A and B were all significant, but the primary interest is in comparing the effects of the levels of A. Use the approach analogous to the first call of PROC GLM in Table 19.19 to do the following.

(a) Construct a 95% confidence interval for the main effect of alert type, and interpret the results.

(b) For each level of complexity, estimate the difference in effects of the two levels of alert type. Determine the variance of each of the two corresponding pairwise comparison estimators,

compute the corresponding estimated standard errors, and use Satterthwaite’s approximation to compute the approximate number of degrees of freedom associated with each estimated standard error.

(c) Using the results of part (b), construct individual 99% confidence intervals for the difference in effects of the two levels of alert type for each level of complexity. Interpret the results.