概化理论偏态分布数据方差分量标准误估计
The Estimating Standard Error of Variance Component for Skewed Distribution Data in Generalizability Theory

利用 GH 分布性质,采用 Monte Carlo 数据模拟技术,模拟生成一定偏度的偏态分布数据,运用Traditional方法、Jackknife方法、Bootstrap方法和MCMC方法估计概化理论偏态分布数据的方差分量标准误,探讨了数据的不同偏度对概化理论方差分量标准误估计的影响.研究结果显示: Jackknife 方法估计偏态分布数据的方差分量标准误性能较差, Traditional和MCMC方法尚可, Bootstrap方法标准误偏差相对较小,且偏态分布数据的偏度对概化理论方差分量标准误估计有影响, Bootstrap方法对于偏态分布数据表现出良好的“适应性”,偏度对其影响较小.
To explore how skew has effect on estimating standard error of variance component for Generalizability Theory. Using nature of Generalized Hyperbolic distribution, the study adopts Monte Carlo data simulation technique to simulate skewed distribution data. Traditional method, bootstrap method, jackknife method and Markov Chain Monte Carlo (MCMC)method were used to compare estimating standard error of variance component for skewed distribution data in Generalizability Theory. Jackknife method is not good to estimate standard error of variance component for skewed distribution data. Traditional method and Markov Chain Monte Carlo (MCMC)method were not very suitable, but can be accepted and bootstrap method is better. Skew of skewed distribution data have a effect on estimating standard error of variance component. Bootstrap method is a good adaptability to estimate standard error of variance component for Generalizability Theory. Skew has less effect on Bootstrap method