Saddle-point Approximation for General Chi-squared-type Mixed Distribution
Xizhi Wu · Tongji yu xinxi luntan · 2008
Random variables of general chi-squred-type mixed distribution is often used in nonparametric testing procedure.Its normal approximation performs not well in small or moderate sample size situations.In this article,we take the saddlepoint approximation approach,which outperforms the normal and chi-squared approximations,especially in the tail areas of the distribution.