【摘要】 北京地区秋白菜上蚜虫种群在其一定发展阶段的空间格局是可以用Pearson Ⅲ型分布拟合的。用蒙特卡洛试验研究这种分布的三个参数——平均数Ex,变异系数Cv和偏态系数Cs的估计误差同样本容量的关系是有效的途径。本文介绍了该试验的设计思想和实施步骤,并得出初步结论:以一株菜为一个样本单位计数白菜上的蚜虫头数,则在允许误差不大于5％的情况下,样本容量为50时,用矩法便可足够准确地估计Ev和Cv值了;但对于估计Cv值,则样本容量应为500左右。更多还原

【Abstract】 For mixed populations of Myzus persicae (Sulzer) and Lipaphis erysimi. (Kaltenbach) on Chinese cabbage in autumn in the area of Beijing, some of their spatial distributions can be simulated w;th the type III of Pearson curve system (P-III distribution), whose variables can be generated in terms of the Monte Carlo method on TRS-80 microcomputer. The P-III distribution is definited with three shape parameters: Ex, the mean; Cv, the coefficient of variation and Cs, the coefficient of skewness. The present paper deals with relation between accuracy of the parameter estimates and sample size from the P-III distribution population generated by the method in Whittaker (1973).Let η denote the estimates of Ex, Cv and Cs; Eη, aη respectively mathematical expects and variances of the estimates; L the number of samples to enter into statistical calculation; and e the permissible relative error, thus.whereIn order to ensure the precision of Monte Carlo test, the L value should be determinated. So firstly, an exploratory test is made:Let L = 50 and sample size has five levels, that is EN = 50, 100, 200, 300, 400. At the end of the test, Cvη-values are estimated by the moment method and then L values are gained whene=5%. Secondly, the L values are verified.Finally, relative errors in the parameter estimation, corresponding to the levels of sample size, are calculated byThe result shows that in the case of e = 5%, the moment method can quite exactly estimate Ex and Cv even when EN = 50; but for Cs, the method requires EN = 500 at least.更多还原