Using Adjusted Bootstrap to Improve the Estimation of Variance Components and Their Variability for Generalizability Theory
Guangming Li, Minqiang ZHANG
Abstract
Open-access reader
Guangming Li, Minqiang ZHANG
Abstract
Open-access reader
摘要: Bootstrap方法是一种有放回的再抽样方法, 可用于概化理论的方差分量及其变异量估计。用Monte Carlo技术模拟四种分布数据, 分别是正态分布、二项分布、多项分布和偏态分布数据。基于p×i设计, 探讨校正的Bootstrap方法相对于未校正的Bootstrap方法, 是否改善了概化理论估计四种模拟分布数据的方差分量及其变异量。结果表明:跨越四种分布数据, 从整体到局部, 不论是“点估计”还是“变异量”估计, 校正的Bootstrap方法都要优于未校正的Bootstrap方法, 校正的Bootstrap方法改善了概化理论方差分量及其变异量估计。
OpenAlex reports 1 citations for this work. Citation counts describe recorded attention and do not establish research quality.
A contribution statement is not available in the OpenAlex record.
Method details are not available in the OpenAlex metadata.
Findings are not separately available in the OpenAlex metadata.
Limitations are not available in the OpenAlex metadata.
Application details are not available in the OpenAlex metadata.
摘要: Bootstrap方法是一种有放回的再抽样方法, 可用于概化理论的方差分量及其变异量估计。用Monte Carlo技术模拟四种分布数据, 分别是正态分布、二项分布、多项分布和偏态分布数据。基于p×i设计, 探讨校正的Bootstrap方法相对于未校正的Bootstrap方法, 是否改善了概化理论估计四种模拟分布数据的方差分量及其变异量。结果表明:跨越四种分布数据, 从整体到局部, 不论是“点估计”还是“变异量”估计, 校正的Bootstrap方法都要优于未校正的Bootstrap方法, 校正的Bootstrap方法改善了概化理论方差分量及其变异量估计。
Key concepts: Jackknife resampling, Statistics, Markov chain Monte Carlo, Mathematics, Confidence interval, Econometrics, Bootstrapping (finance), Generalizability theory