;We propose new goodness fit of test approaches that are easy to implement with the bootstrapping techniques. The techniques are instituted by taking advantage of the fact that the mean regression parameters can be consistently estimated by using normal, gamma and Poisson models even when model fails. We test the appropriateness of the Weibull, log-normal and gamma model assumptions to illustrate the merit of our new methods. We also compare our novel approaches with several commonly used and implemented existing methods with simulations and real data analyses.
