Critical value functions for likelihood-ratio tests for normality

The asymptotic distributions of likelihood ratio tests for normality are unknown; only a few critical values have been tabulated using Monte Carlo simulations.This study aims to develop the critical value functions for these likelihood ratio tests using response surface regressions.In these regressions, the simulated critical values depend on sample size; however, practitioners can easily compute the finite-sample critical values for a number of sample sizes using a hand calculator.An extensive Monte Carlo simulation shows that the proposed critical value functions perform very well for both small and large samples.