生态学研究中常见的统计学问题分析

在当代生态学研究中统计学方法的应用日益广泛,对于生态科学的发展和研究水平的提高起到了积极的作用。但是不容忽视的是在生态学研究应用统计学方法的过程中存在若干问题,主要表现在:1)回归分析方面的问题。直线回归方程用相关指数R²来描述直线回归的显著性;曲线回归方程往往用相关系数r来表示显著性;多元线性回归方程只对方程进行显著性检验,没有对每一个回归系数进行显著性检验。2)方差分析方面的问题。当处理数超过2时,不恰当地使用t_检验比较平均数的差异显著性。该文分析了产生这些问题的原因,提出了改进的对策。

ANALYSIS ON SOME MISTAKES IN THE APPLICATION OF STATISTICS IN ECOLOGICAL RESEARCHES

The application of statistical methods in ecological researches is more and more extensive, and it is very important for improvement and development of ecologica science. However, there are some mistakes when statistical methods are applied to ecological researches, 1) Regression analysis: correlation index, R~2, was incorrectly used to test whether a simple regression equation was significant or not. Correlation coefficient, r, was inappropriately used to describe the significance of curve regression equations. Moreover, multiple linear regression equation was only tested by analysis of variance (ANOVA), but each regression coefficient in the equation was not tested; 2) ANOVA. t_test was applied inappropriately to test significance of the means of all treatments when the treatments in an experiment were more than 2. The reasons for these mistakes were investigated and methods to avoid these mistakes were introduced.