Table 3:

Simulation results from 10,000 MC runs when the true model is a near upper hinge linear model, and the data are fit using an upper hinge model. Mean estimate, %bias, and MC SD of the parameter estimate are given. Additionally, coverage and average standard error (SE) are given for model-based and bootstrap-based standard error estimates.

				Model-based		Bootstrap-based
	n	Est(%Bias)	MC SD	Coverage	SE	Coverage	SE
${\mathit{\beta }}_z$
	50	1.00 (0)	0.06	0.944	0.06	0.941	0.06
	100	1.00 (0)	0.04	0.949	0.04	0.946	0.04
	500	1.00 (0)	0.02	0.947	0.02	0.945	0.02
${\beta }_{(x-e)-}$
	50	1.02 (3)	0.22	0.898	0.17	0.959	0.28
	100	1.00 (2)	0.14	0.914	0.12	0.959	0.16
	500	0.99 (0)	0.06	0.916	0.05	0.952	0.06
$e$
	50	3.03	0.53	0.837	0.47	0.948	0.73
	100	3.02	0.34	0.880	0.32	0.958	0.49
	500	3.02	0.15	0.912	0.14	0.952	0.18