Assessing external validity in practice

IF 1.2 Q3 ECONOMICS

Research in Economics Pub Date : 2024-05-22 DOI:10.1016/j.rie.2024.100964

Sebastian Galiani , Brian Quistorff

引用次数: 0

Abstract

We review, from a practical standpoint, the evolving literature on assessing external validity (EV) of estimated treatment effects. We review existing EV measures, and focus on methods that permit multiple datasets (Hotz et al., 2005). We outline criteria for practical usage, evaluate the existing approaches, and identify a gap in potential methods. Our practical considerations motivate a novel method utilizing the Group Lasso (Yuan and Lin, 2006) to estimate a tractable regression-based model of the conditional average treatment effect (CATE). This approach can perform better when settings have differing covariate distributions and allows for easily extrapolating the average treatment effect to new settings. We apply these measures to a set of identical field experiments upgrading slum dwellings in three different countries (Galiani et al., 2017).

查看原文本刊更多论文

在实践中评估外部有效性

我们从实用的角度出发，回顾了有关评估估计治疗效果外部有效性（EV）的不断发展的文献。我们回顾了现有的 EV 测量方法，并重点关注允许使用多个数据集的方法（Hotz 等人，2005 年）。我们概述了实际使用的标准，评估了现有的方法，并找出了潜在方法中的差距。我们的实际考虑促使我们采用了一种新方法，即利用 Group Lasso（Yuan 和 Lin，2006 年）来估计基于回归的条件平均治疗效果（CATE）模型。这种方法在具有不同协变量分布的环境中表现更佳，并且可以轻松地将平均治疗效果外推到新的环境中。我们将这些措施应用于一组相同的实地实验，对三个不同国家的贫民窟进行升级改造（Galiani 等人，2017 年）。

本文章由计算机程序翻译，如有差异，请以英文原文为准。

求助全文

约1分钟内获得全文求助全文

来源期刊

Research in Economics ECONOMICS-

CiteScore

1.40

自引率

0.00%

发文量

审稿时长

89 days

期刊介绍： Established in 1947, Research in Economics is one of the oldest general-interest economics journals in the world and the main one among those based in Italy. The purpose of the journal is to select original theoretical and empirical articles that will have high impact on the debate in the social sciences; since 1947, it has published important research contributions on a wide range of topics. A summary of our editorial policy is this: the editors make a preliminary assessment of whether the results of a paper, if correct, are worth publishing. If so one of the associate editors reviews the paper: from the reviewer we expect to learn if the paper is understandable and coherent and - within reasonable bounds - the results are correct. We believe that long lags in publication and multiple demands for revision simply slow scientific progress. Our goal is to provide you a definitive answer within one month of submission. We give the editors one week to judge the overall contribution and if acceptable send your paper to an associate editor. We expect the associate editor to provide a more detailed evaluation within three weeks so that the editors can make a final decision before the month expires. In the (rare) case of a revision we allow four months and in the case of conditional acceptance we allow two months to submit the final version. In both cases we expect a cover letter explaining how you met the requirements. For conditional acceptance the editors will verify that the requirements were met. In the case of revision the original associate editor will do so. If the revision cannot be at least conditionally accepted it is rejected: there is no second revision.