Table 3. Type I error rates for bootstrap simulations.

Under 1,000 simulations with a binary outcome of accept or reject, deviations in type I error rates beyond a Wald confidence interval [0.036, 0.064] suggest deviations from a level 0.05 test not due to sampling. Scenarios where the error rates exceeded the Wald confidence bounds are bolded.

Conditions								Type I error rates
ρXC	ρXM	ρCM	ρXY	ρCY	ρMY	α1	γ3	PSRM	IEFM	PB	BCB	BCAB
n = 30
0.15	0	0	0.15	0.15	0.6	0	0.6	0.071	0.086	0.065	0.115	0.109
0.6	0	0	0.6	0.6	0.6	0	0.6	0.061	0.079	0.064	0.100	0.084
n = 100
0.15	0	0	0.15	0.15	0.6	0	0.6	0.052	0.056	0.060	0.090	0.075
0.6	0	0	0.6	0.6	0.6	0	0.6	0.051	0.059	0.060	0.082	0.065