生态学区组试验设计的方差分析及P值探讨  

Analysis of Variance and P-value of Block ExperimentalDesign in Ecology

作　　者：吕世杰 闫宝龙 王忠武 李治国 康萨如拉 刘红梅[3] Lv Shijie;Yan Baolong;Wang Zhongwu;Li Zhiguo;Kang Sarula;Liu Hongmei(College of Grassland,Resources and Environment/Key Laboratory of Grassland Resources of the Ministry of Education/Key Laboratory of Forage Cultivation,Processing and High Efficient Utilization of the Ministry of Agriculture and Rural Affairs/Inner Mongolia Key Laboratory of Grassland Management and Utilization,Inner Mongolia Agricultural University,Hohhot 010019;College of Agriculture,Inner Mongolia University for Nationalities,Tongliao 028043;Inner Mongolia Academy of Forestry Science,Hohhot 010010,China)

机构地区：[1]内蒙古农业大学草原与资源环境学院/草地资源教育部重点实验室/农业农村部饲草栽培、加工与高效利用重点实验室/内蒙古自治区草地管理与利用重点实验室,呼和浩特010019 [2]内蒙古民族大学农学院,通辽028043 [3]内蒙古自治区林业科学研究院,呼和浩特010010

出　　处：《草原与草业》2020年第4期40-45,共6页Grassland and Prataculture

基　　金：国家自然科学基金项目(32060384);教育部草地资源可持续管理科技创新项目(IRT_17R59)。

摘　　要：为保证生态学区组试验设计数据分析的科学性,对区组试验设计的方差分析进行了比较全面的阐述,并对P值进行了初步探讨。认为单因素区组试验设计的方差分析模型应该是双因素固定效应方差分析模型,且不能考虑试验处理效应与区组效应的交互作用;在方差分析过程中,如果存在区组效应干扰时,可调整为单因素方差分析模型;这一过程均需要对指标数据进行正态性、方差同质性检验,结合样本容量和线性拟合率综合分析方差分析的可靠性和科学性。在进行多重比较时,需要给定具体的P值,对应的方差分析模型、处理效应检验可以根据研究情况对P值做出合理的调整;建议方差分析模型、处理效应检验和多重比较检验的P值最好一致。In order to ensure the scientific validity of data analysis of block experimental design in ecology,this study elaborated a comprehensive method for variance analysis in block experimental design,and provided a preliminary discussion on the P-value.It is considered that the analysis of variance model for single-factor block test designs should be a two-factor fixed-effects analysis of variance model,and the interaction between the test treatment effect and block effect can be ignored.In the process of analysis of variance,if there is block effect interference,it can be adjusted to a single factor analysis of variance model.This process requires normality and variance homogeneity tests on the indicator data,combined with analysis of sample size and linear fit rate,to comprehensively assess the reliability and scientific validity of analysis of variance.When performing multiple comparisons,specific P-values needed to be given.The corresponding analysis of variance model and processing effect test can make reasonable adjustments to the P-value according to the research situation.It is recommended that the analysis of variance model,treatment effect test and multiple comparison tests should have the same P-value.

关 键 词：生态学 单因素区组设计 方差分析运用 P值选用 

分 类 号：Q141[生物学—生态学]