摘要
研究了74个小麦品种（系）的38个数量性状的表型分布,同时比较了6种正态性检验方法。结果表明:仅有少数性状符合正态分布,为使分析结果准确可靠,在作以正态性假设为基础的统计分析前对数量性状表型值进行正态性检验非常必要;6种检验方法中,频数图解法较直观,偏度系数定向检验、峰度系数定向检验、偏度与峰度系数联合检验和达戈斯提诺检验较严格,且相互间差异小,卡方拟合检验较宽松,检验方法的选用取决于试验者的目的要求与试验数据的性质。
This paper studied the distribution of phenotypic value for thirty-eight quantitative characters over seventy wheat varieties and made a comparison of six normality test methods. Only a few of the 38 characters studied obey normal distribution; it was very necessary to carry out normality test for experimental data before statistical analyses based on normality hypothesis; frequency graph method is comparatively visual; Chi-square fittest test was not very strict; skewness test, kurtosis test, joint test of skewness and kurtosis, and D'Agostino test were all strict and showed no obvious differences among them; the selection of test method showed depends on the purposes and the nature of experimental data.
出处
《西南农业学报》
CSCD
1990年第2期33-38,共6页
Southwest China Journal of Agricultural Sciences
关键词
小麦
数量性状
表型值
正态性检验
Wheat, Quantitative character, Distribution of phenotypic value, Normality test, Comparison of test method
