Development Economics X Paper Model Fourteen
We extend the use of pattern graphs–graphical tools to represent nonmonotone missing data mechanisms–to two quasi-experimental settings: regression discontinuity design (RDD) and difference-in-difference (DID). We call these credibility graphs. We show how to use pattern graphs to derive estimators for RDD and DID parameters under nonignorable missingness, and how to choose and justify the cutoffs for RDD and the parallel trends assumption for DID. We also provide some theoretical results on the validity and robustness of our approach.
Opoku-Agyemang, Kweku (2023). "Credibility Graphs for Missing Data in Program Evaluations." Development Economics X Paper Model Fourteen.
© 2023 Development Economics X Corporation. All Rights Reserved.