compliers

Compliers are individuals in a randomized controlled trial or experimental study who adhere to the treatment assignment. The concept arises when participation in the treatment is voluntary or constrained, so not all assigned to treatment actually receive it, and some assigned to control may obtain the treatment elsewhere. In this context researchers commonly categorize participants into four compliance types: compliers, never-takers, always-takers, and defiers. Compliers take the treatment if and only if they are assigned to the treatment group.

Under standard assumptions—random assignment, stability, and monotonicity (no defiers)—the local average treatment effect (LATE) can be identified for the compliers using instrumental variable methods. The LATE represents the average causal effect of the treatment on outcomes for the complier subgroup, not for the entire population. The notion emphasizes that estimated effects can depend on the level of compliance and the observed instrument.

Compliers are typically not directly observable; researchers infer their presence through the relationship between assignment and treatment uptake. The distinction contrasts with the intention-to-treat (ITT) effect, which measures outcomes by assignment regardless of actual treatment, and with effects estimated for other latent groups (never-takers, always-takers, defiers).

Limitations include the dependence on assumptions such as monotonicity and random assignment, potential lack of external validity, and the possibility that compliers differ systematically from non-compliers. Nevertheless, the compliers framework provides a formal way to interpret causal effects in the presence of noncompliance and guides the use of instrumental variables in policy evaluation, medicine, and economics.