基于中国湿地CVM研究的Meta分析  

A meta-analysis of wetland CVM studies in China

作　　者：李庆波 敖长林[1] 袁伟 高琴[1,2] LI Qingbo;AO Changlin;YUAN Wei;GAO Qin(Department of Management Science and Engineering,College of Engineering,Northeast Agricultural University,Harbin 150030,China;International Education Institute,Shandong Institute of Business and Technology,Yantai 264005,China)

机构地区：[1]东北农业大学管理科学与工程系,哈尔滨150030 [2]山东工商学院国际教育学院,烟台264005

出　　处：《资源科学》2018年第8期1634-1644,共11页Resources Science

基　　金：国家自然科学基金资助项目(71171044);山东省自然科学基金项目(ZR2016GM21)

摘　　要：Meta分析作为效益转移的一种有效方法,在湿地资源评估方面得到广泛应用,但由于样本数据中出版选择偏差的存在,使评估结果的准确性受到影响。为解决该问题,本文构建可消除出版选择偏差影响的Meta回归模型,通过收集中国应用CVM对湿地进行价值评估的研究结果,建立价值转移数据库,通过检验数据库中出版选择偏差的存在与否,选择恰当的Meta回归模型,进而建立价值转移模型,并对模型样本外价值转移的有效性进行检验。FAT-PET的检验结果表明,数据库中不存在出版选择偏差,说明本文所选取文献的样本代表性较好,样本数据并没有受到出版选择偏差的影响;Meta回归结果表明,湿地服务类型、湿地位置、湿地类型、湿地面积、受益人口数以及问卷引导方式的差异,都会导致湿地价值估值的差异,而问卷调查方式、人均GDP、研究时间和期刊质量等对湿地价值估值没有显著影响;Meta回归模型的样本外效益转移的误差范围为0.08%~39.02%,误差均值为9.58%,模型可用于样本外效益转移。Meta-analysis is an effective method of benefit transfer, which has been widely used in the evaluation of wetland resources. However, due to the existence of publication selection bias in the sample data, the accuracy of the evaluation result is affected. In order to solve this problem, we constructed a meta-regression model which can eliminate the influence of publication selection bias. The value transfer database was established through collecting the empirical research results of the application of CVM to evaluate the value of wetlands in China and an appropriate metaregression model was chosen by examining the existence of the publication selection bias in the database or not. Then, the model of value transfer was established, and the effectiveness of out-of-sample value transfer was examined. FAT-PET test results show that there is no publication selection bias in the database, which demonstrates that the sample of the literature selected in this paper is more representative and the sample data are not affected by publication selection bias. Meta-regression results indicate that the types of wetland services, wetland location, wetland type, wetland area, number of beneficiaries and the difference in the format of the questionnaire will lead to the difference in the valuation of wetland value. However, the way of questionnaire survey, per capita GDP, research time and periodical quality had no significant impact on the valuation of wetland value. The error range of out-of-sample benefit transfer in meta-regression model was 0.08%-39.02%, and the average error was 9.58%. Thus, the meta-regression model can be used for out-of-sample benefit transfer.

关 键 词：出版选择偏差 META分析 湿地 效益转移 中国 

分 类 号：X37[环境科学与工程—环境工程]