【摘要】 本文通过蒙特卡罗模拟产生的八个模拟资料比较了我国常用的方差分析法(不考虑场年季效应)、Henderson方法1、Henderson方法3、最大似然法、改进最大似然法与约束最大似然法。结果表明:方差分析法估计值偏差最大,而约束最大似然法的估计值最准确。更多还原

【Abstract】 Eight simulated data sets are generated by Monte Carlo simulation to compare the relative accuracy of methods for estimating variance components. These methods selected are Analysis of variance ( ANOVA ) which does not take the fixed herd-year-season effects into account, Henderson method 1 , Henderson method 3, Maximum Likelihood ( ML ) , Modified Maximum Likelihood(MOML ) and Restricted Maximum Likelihood ( REML ) . The simulation results indicate that the estimates of REML are most accurate and the estimates of ANOVA are most biased, compared with those of the other methods.更多还原