Working Paper: CEPR ID: DP8003
Authors: Sylvian Chassang; Gerard Padro i Miquel; Erik Snowberg
Abstract: We study the design of randomized controlled experiments in environments where outcomes are significantly affected by unobserved effort decisions taken by the subjects (agents). While standard randomized controlled trials (RCTs) are internally consistent, the unobservability of effort provision compromises external validity. We approach trial design as a principal-agent problem and show that natural extensions of RCTs--which we call selective trials--can help improve the external validity of experiments. In particular, selective trials can disentangle the effects of treatment, effort, and the interaction of treatment and effort. Moreover, they can help experimenters identify when measured treatment effects are affected by erroneous beliefs and inappropriate effort provision.
Keywords: Randomized Controlled Trials; Selective Trials; Blind Trials; Incentivized Trials; Marginal Treatment Effects; Mechanism Design; Selection; Heterogeneous Beliefs; Compliance
JEL Codes: C81; C93; D82; O12
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| unobserved effort (D29) | informativeness of RCTs (C90) |
| low true returns (G19) | low effort (D29) |
| low effort (D29) | measured treatment effects (C22) |
| selective trials (C90) | external validity of experiments (C90) |
| agents express preferences for treatments (I11) | implicit selection explicit (D84) |
| selective trials (C90) | identify effects of treatment, effort, and interaction between them (C90) |
| selective trials (C90) | recover distribution of returns as a function of willingness to pay (D39) |
| blind trials (C90) | identify pure treatment effects (C22) |
| incentivized trials (O31) | similar information to blind trials (C90) |