We checked for possible nonrespons bias and found no significant differences between the first third and last third of the data, nor were the usable responses significantly different from those that were eliminated based on several demographic characteristics. Hence, we concluded that nonrespones bias was not a serious problem.