Publications by authors named "Alexander Ratnikov"
Psychol Methods
September 2014
Article Synopsis
- Treatment noncompliance in randomized experiments can impact the accuracy of causal conclusions and the interpretation of treatment effects.
- The article reviews 7 methods to address this issue, dividing them into 3 traditional and 4 newer statistical strategies.
- Traditional methods include intention-to-treat analysis, as-treated analysis, and per-protocol analysis, while newer methods encompass complier average causal effect, dose-response estimation, propensity score analysis, and treatment effect bounding.
- The discussion analyzes each approach's applications, estimates, assumptions, strengths, and weaknesses.
