...
【24h】

Alternatives to Randomized Experiments

机译:随机实验的替代方法

获取原文
获取原文并翻译 | 示例
           

摘要

Randomized experiments are preferred for making inferences about causality when they can be implemented and their assumptions are met. Yet assumptions can fail (e.g., attrition, treatment noncompliance) or randomization may be unethical or infeasible. I describe alternative design and statistical approaches that permit testing causal hypotheses and present current empirical evidence related to alternative designs. Alternative designs permit a wider range of research questions to be answered and permit more direct generalization of causal effects; however, when using such designs, estimates of the magnitude of the causal effect may be more uncertain.
机译:当可以实施和满足假设时,最好使用随机实验推断因果关系。然而,假设可能会失败(例如,人员流失,治疗不依从),或者随机化可能是不道德或不可行的。我描述了替代设计和统计方法,这些方法和方法可以检验因果假设,并提供与替代设计相关的当前经验证据。替代设计可以回答更广泛的研究问题,并可以更直接地推广因果关系;但是,当使用此类设计时,因果效应的大小估计可能会更加不确定。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号