论文部分内容阅读
目的 研究非独立数据组内相关系数 ρ的抽样分布。 方法 用MonteCarlo方法从已知组内相关系数的假定总体中重复抽样 ,用RIGLS法拟合二水平模型 ,分别计算 ρ值以观察其分布类型。 结果 组内相关系数抽样分布类型随 ρ的数值大小而变化 ,小于 0 5时为正偏态 ,大于 0 5时为负偏态 ,接近 0 5时为近似正态分布。其估计误差与 1水平和 2水平的样本含量有关 ,总样本含量相同时 ,2水平的样本含量越大 ,估计误差越小。结论 组内相关系数的大小反映非独立数据的共变关系大小 ,其分布类型的研究对非独立数据分析时的实际应用具指导意义
Objective To study the sampling distribution of the correlation coefficient ρ in non-independent datasets. Methods The Monte Carlo method was used to repeatedly sample from the assumed population of known intergroup correlation coefficients. The RILLS method was used to fit the two-level model and the ρ values were calculated to observe the distribution types. RESULTS: The sampling distribution of the correlation coefficient within the group varied with the value of ρ, and was positively skewed when it was less than 0 5, negatively skewed when it was greater than 0 5, and nearly normal distributed when it was close to 0.5. The estimation error is related to the level 1 and level 2 samples. When the total sample size is the same, the larger the sample level 2 level, the smaller the estimation error. Conclusion The size of the intra-group correlation coefficient reflects the size of the covariation of non-independent data, and the study of its distribution type is instructive for the practical application of non-independent data analysis.