两组资料:(1)大样本资料或服从正态分布的小样本资料;(2)若方差齐性,则做成组t检验;(3)若方差不齐,则做t’检验或用成组的Wilcoxon秩和检验;(4)小样本偏态分布资料,则用成组的Wilcoxon秩和检验。多组资料:(1)若大样本资料或服从正态分布,并且方差齐性,则做完全随机的方差分析。如果方差分析的统计检验为有统计学意义,则进一步做统计分析,即选择合适的方法 (如LSD检验、Bonferroni检验等)进行两两比较。(2)若小样本的偏态分布资料或方差不齐,则做Kruskal Two sets of data: (1) Large sample data or small sample data subject to normal distribution; (2) If the homogeneity of variance, then make a group t test; (3) If the variance is not homogeneous, do t ’test or Group Wilcoxon rank sum test; (4) small sample skewed distribution data, the group Wilcoxon rank sum test. Multiple sets of data: (1) If large sample data are subject to normal distribution and homogeneity of variance, do a completely random analysis of variance. If the statistical analysis of variance analysis is statistically significant, further statistical analysis, that is, select the appropriate method (such as LSD test, Bonferroni test, etc.) for any comparison. (2) Kruskal if the skewed distribution data or variance of the small sample is not correct