Abstract: A method for multi-way heteroscedastic analysis is presented,which detailedly discusses the heteroscedastic analysis of one,two and multiple factors,in the condition of the same coefficient of variation in all states.It can test whether the multiple populations have a certain kind of different means,and estimate their parameters.With small sample,it can test the results of simulation or digital design and analyze the influence of each factor in the results.And then,it can point out which factors have been simulated correctly and which haven’t.Furthermore,this method can correct the results of simulation or digital design according to the experimental data.