生物多样性 ›› 2026, Vol. 34 ›› Issue (1): 25308.  DOI: 10.17520/biods.2025308  cstr: 32101.14.biods.2025308

• 生态学数据分析方法专题 • 上一篇    下一篇

Meta分析应用中应注意的几个关键问题

张霜1, 宋波2*   

  1. 1. 中国科学院生态环境研究中心区域与城市生态安全全国重点实验室, 北京 100085; 2. 中国科学院昆明植物研究所云南省极小种群野生植物综合保护重点实验室, 昆明 650201
  • 收稿日期:2025-08-04 修回日期:2025-12-09 接受日期:2025-12-11 出版日期:2026-01-20 发布日期:2026-01-21
  • 通讯作者: 宋波
  • 基金资助:
    国家自然科学基金面上项目(31971481); 中国科学院西部之光交叉团队项目(xbzg-zdsys-202319)

Several key questions when conducting a meta-analysis

Shuang Zhang1, Bo Song2*   

  1. 1 State Key Laboratory of Regional and Urban Ecology, Research Center for Eco-Environmental Sciences, Chinese Academy of Sciences, Beijing 100085, China 

    2 Yunnan Key Laboratory for Integrative Conservation of Plant Species with Extremely Small Populations, Kunming Institute of Botany, Chinese Academy of Sciences, Kunming 650201, China

  • Received:2025-08-04 Revised:2025-12-09 Accepted:2025-12-11 Online:2026-01-20 Published:2026-01-21
  • Contact: Bo Song

摘要: Meta分析是通过对不同案例数据进行加权整合分析, 得到普适性结论的重要统计工具, 在生态学领域具有广阔的应用价值。但长期以来, 科研人员对Meta分析的基本理念和方法体系具有较多的认识误区, 一定程度上造成了该方法的误用甚至错用。本文从Meta分析的操作步骤出发, 从其基本特征、文献的查询与筛选、效应值的构建、模型的选取、特殊数据结构的纳入、解释变量的引入、结果可靠性的判定、软件工具介绍等几个方面, 介绍了Meta分析的基本理念和应用中应注意的问题。相关概念和技术要点的厘清, 将为我们构建更为精准、合理的Meta分析模型, 提升结果可靠性提供帮助。Meta分析技术的不断更新进步, 必将为生态学领域众多基础科学问题的回答提供更有力、可靠的技术支撑。

关键词: 效应值, 异质性, 随机效应模型, 层次模型, Meta回归, 发表偏倚性

Abstract

Background & Aims: Meta-analysis is a crucial statistical tool for deriving generalized conclusions through the weighted analysis of data from case studies, with broad applications in ecology. However, for a long time, researchers have significant misconceptions regarding the fundamental principles and methodological framework of meta-analysis, which has contributed to its misuse, even erroneous application. 

Review Results: According to the standard steps of conducting a meta-analysis, this article summarizes the basic feature of meta-analysis and highlights critical considerations for its application, encompassing aspects such as: defining key concepts, literature search and screening, construction of effect sizes, selection of models, incorporation of special data structures, inclusion of explanatory variables, assessment of result reliability, and relevant software tools. 

Conclusion: The clarifications of related concepts and key points will aid in constructing more precise and appropriate meta-analysis models, thereby enhancing the reliability of results. Furthermore, continued advancements in meta-analysis methodology is poised to offer more robust and reliable technical approaches for addressing numerous fundamental scientific questions in ecology.

Key words: effect size, heterogeneity, random effect model, hierarchical model, meta-regression, publication bias