Finally, to further evaluate the stability of model parameters, bootstrapping procedures were employed within AMOS. The procedure generated no unused samples and 1000 useable samples. The results strongly supported the original parameter values generated via maximum likelihood, and reflected robust SE estimates. Bias, the differences between bootstrap generated means and the original estimates, ranged from .000 to .004 for standardized parameters, from .000 to .065 for variances, and from .000 to .006 for squared multiple correlations. The bias was .000 for the relationship between SEX and AGE.
