Table 1.

Proportion of True Values to Left and Right of 95% Confidence Intervals - One or Two Zero Paths, Study 1

			Sample Size
Effect Size			50		100		200		500		1000
α	β	Test	true to left	true to right	true to left	true to right	true to left	true to right	true to left	true to right	true to left	true to right
0	0	z	0*	0.0001*	0*	0*	0*	0.0001*	0.0002*	0*	0.0003*	0*
		M	0.0007*	0.0009*	0.0012*	0.0009*	0.0013*	0.0005*	0.0013*	0.0007*	0.0012*	0.0006*
0	.14	z	0.0003*	0.0001*	0.0013*	0.0008*	0.0023*	0.0022*	0.0064*	0.0046*	0.0129	0.0127
		M	0.0024*	0.0018*	0.0063*	0.0040*	0.0079*	0.0079*	0.0079*	0.0140	0.0194	0.0190
0	.39	z	0.0047*	0.0042*	0.0116*	0.0078*	0.0201	0.0177	0.0218	0.0249	0.0256	0.0241
		M	0.0129	0.0122*	0.0153	0.0165	0.0214	0.0201	0.0228	0.0235	0.0249	0.0244
0	.59	z	0.0126	0.0119*	0.0189	0.0169	0.0211	0.0213	0.0266	0.0232	0.0248	0.0244
		M	0.0202	0.0179	0.0205	0.0181	0.0211	0.0215	0.0263	0.0223	0.0248	0.0241
Average		z	0.0044*	0.0041*	0.0080*	0.0064*	0.0109*	0.0103*	0.0138	0.0132	0.0159	0.0153
		M	0.0091*	0.0082*	0.0108*	0.0099*	0.0129	0.0125	0.0146	0.0151	0.0176	0.0170

Note. z refers to confidence limits based on the traditional z test. M refers to the confidence limits formed based on the distribution of the product tables in Meeker et al. (1981). Values marked with * are outside Bradley (1978) robustness criteria.