Table 2.

Estimated Type I error rates for Wald and bootstrap tests for two intervention designs

Test	Design 1 (D1)				Design 2 (D2)
	ρ = 0	ρ = .2	ρ = .5	ρ = .7	ρ = 0	ρ = .2	ρ = .5	ρ = .7
Normal errors
Wald	.0674	.0728	.0854	.0808	.0612	.0697	.1000	.1130
Parametric bootstrap
B = 100	.0503	.0503	.0522	.0500	.0513	.0510	.0529	.0540
B = 200	.0513	.0522	.0507	.0544	.0460	.0489	.0508	.0537
Semi‐parametric bootstrap
B = 100	.0513	.0519	.0522	.0541	.0457	.0462	.0465	.0510
B = 200	.0494	.0479	.0505	.0499	.0508	.0475	.0458	.0561
Contaminated normal errors
Wald	.0522	.0625	.0820	.0917	.0547	.0688	.1071	.1223
Parametric bootstrap
B = 100	.0434	.0483	.0475	.0559	.0411	.0448	.0615	.0698
B = 200	.0382	.0411	.0494	.0612	.0391	.0425	.0552	.0677
Semi‐parametric bootstrap
B = 100	.0442	.0436	.0506	.0508	.0392	.0446	.0544	.0577
B = 200	.0428	.0432	.0464	.0518	.0408	.0456	.0587	.0584