常用统计学方法：分类（名义分类或等级分类）资料用χ2检验、Ridit分析，区间变量均数组间比较用t检查、F检验、秩和检验（适用于呈偏态分布的区间变量）。χ2检验的适应范围：当总体样本含量（n）>40，理论频数（T）>5时，用χ...常用统计学方法：分类（名义分类或等级分类）资料用χ2检验、Ridit分析，区间变量均数组间比较用t检查、F检验、秩和检验（适用于呈偏态分布的区间变量）。χ2检验的适应范围：当总体样本含量（n）>40，理论频数（T）>5时，用χ2检验的基本公式；若n>40.1<T<5时，用校正χ2检验；若n<40或T<1时，用χ2检验的确切概率法。t检验、F检验均适用于呈正态分布的区间变量，t检验常用于配对设计的差值均数与总体均数0的比较（如治疗前后比较）、成组设计的两小样本均数的比较，成组设计的两大样本均数的比较用U检验；F检验是一揽子比较（只能发现两组以上区间变量均数的差异是否存在统计学意义,而不能发现是哪两组差异存在统计学意义），多个样本均数间的两两比较可用事后检验（Post Hoc），常用用LSD法、SNK法、Dunnett’t法、Dunnett’s T 3(方差不齐)、Tammhane’s T 2(方差不齐)等。显著性水准只取P<0.05和P<0.01,若用SPSS、SAS软件等进行统计学处理，也可给出具体的P值。展开更多
In this study, it was shown that, same comparisons can be made by using contrast coefficients instead of Dunnett＇s test in the experiments with control groups. It was also shown that, in situations with an ordinal sc...In this study, it was shown that, same comparisons can be made by using contrast coefficients instead of Dunnett＇s test in the experiments with control groups. It was also shown that, in situations with an ordinal scale and equal spacing quantitative grouping, a trend investigation could be done by contrast coefficients. For this purpose, a small part of the data from a TUBITAK project in the Soil Science Department, Agriculture Faculty, Kahramanmaras Sutcu Imam University, was used with permission. The soils were absorbed to natural zeolite in concentration of 0, 2.5, 5 and 10 mg/kg, and after two years, the available Zinc （Zn） amounts in the soil were analyzed. As a result, in both Dunnett＇s test and contrast methods, the Zn amounts in control and 2.5 mg/kg concentration groups were found similar （P 〉 0.01）; but were different （P 〈 0.01） between control and 5 mg/kg concentration groups, and control and 10 mg/kg concentration groups. Furthermore, when orthogonal polynomial contrast coefficients were investigated, linear effects were found significant （P 〈 0.01） and cubic effects were obtained significant （P 〈 0.05）, but quadratic effect was obtained insignificant （P 〉 0.05）.展开更多
文摘常用统计学方法：分类（名义分类或等级分类）资料用χ2检验、Ridit分析，区间变量均数组间比较用t检查、F检验、秩和检验（适用于呈偏态分布的区间变量）。χ2检验的适应范围：当总体样本含量（n）>40，理论频数（T）>5时，用χ2检验的基本公式；若n>40.1<T<5时，用校正χ2检验；若n<40或T<1时，用χ2检验的确切概率法。t检验、F检验均适用于呈正态分布的区间变量，t检验常用于配对设计的差值均数与总体均数0的比较（如治疗前后比较）、成组设计的两小样本均数的比较，成组设计的两大样本均数的比较用U检验；F检验是一揽子比较（只能发现两组以上区间变量均数的差异是否存在统计学意义,而不能发现是哪两组差异存在统计学意义），多个样本均数间的两两比较可用事后检验（Post Hoc），常用用LSD法、SNK法、Dunnett’t法、Dunnett’s T 3(方差不齐)、Tammhane’s T 2(方差不齐)等。显著性水准只取P<0.05和P<0.01,若用SPSS、SAS软件等进行统计学处理，也可给出具体的P值。
文摘In this study, it was shown that, same comparisons can be made by using contrast coefficients instead of Dunnett＇s test in the experiments with control groups. It was also shown that, in situations with an ordinal scale and equal spacing quantitative grouping, a trend investigation could be done by contrast coefficients. For this purpose, a small part of the data from a TUBITAK project in the Soil Science Department, Agriculture Faculty, Kahramanmaras Sutcu Imam University, was used with permission. The soils were absorbed to natural zeolite in concentration of 0, 2.5, 5 and 10 mg/kg, and after two years, the available Zinc （Zn） amounts in the soil were analyzed. As a result, in both Dunnett＇s test and contrast methods, the Zn amounts in control and 2.5 mg/kg concentration groups were found similar （P 〉 0.01）; but were different （P 〈 0.01） between control and 5 mg/kg concentration groups, and control and 10 mg/kg concentration groups. Furthermore, when orthogonal polynomial contrast coefficients were investigated, linear effects were found significant （P 〈 0.01） and cubic effects were obtained significant （P 〈 0.05）, but quadratic effect was obtained insignificant （P 〉 0.05）.
