论文部分内容阅读
土壤肥力梯度对随机区组试验的影响有时未能或不能通过适当的田间设计除去。若直接对含土壤肥力梯度效应的资料进行方差分析,会使试验结论出现偏差或错误。作为补救措施,本文提出一种消除梯度效应的统计方法:首先对资料中是否存在梯度效应作出统计推断;若存在,则对资料进行矫正,除去此种效应。和原资料相比,矫正资料降低了试验误差,突出了处理效应,提高了分析精度。对本方法在理论上的合理性也进行了讨论。
The impact of soil fertility gradients on randomized block trials has sometimes failed or can not be removed by appropriate field designs. An analysis of variance (ANOVA) directly on data with a gradient effect of soil fertility can lead to deviations or errors in the test results. As a remedial measure, this paper presents a statistical method to eliminate the gradient effect: firstly, make statistical inferences about whether there is a gradient effect in the data; if it exists, correct the data to eliminate such effect. Compared with the original data, the correction data reduces the experimental error, highlights the processing effect and improves the analysis accuracy. The theoretical rationality of this method is also discussed.