研究登记册与可信度危机：一项实证与理论研究

中文摘要

我们分析了针对实验研究可信度危机的一项突出政策方案——研究登记册——主要关注 AEA RCT 登记册。我们发现，AEA RCT 登记册迄今影响有限。近期经济学实验大多未登记，已登记的实验中有仅约一半是在干预之前登记的，且这些预注册大多缺乏足够细节。将这些模式与 ClinicalTrials.gov 相比较，我们发现二者存在广泛的相似性。随后，我们提出一个关于登记的经济模型，以探讨登记册总体上的潜在改进。该模型表明，禁止事后登记可以显著提高登记册的有效性。

Abstract

Abstract We analyse one prominent policy solution to the credibility crisis in experimental research—research registries—with primary focus on the AEA RCT Registry. We find that the AEA RCT Registry has had a limited impact to date. A majority of recent economics experiments are not registered, only about half of those registered did so prior to intervention and most of these preregistrations lack sufficient detail. We find broad similarities when comparing these patterns to ClinicalTrials.gov. We then advance an economic model of registration to explore potential improvements to registries generally. The model shows that banning late registration can significantly increase registry effectiveness.