Categories → Causal Inference , Statistics , Study Design , Causal Effect
LATE measures the causal effect of a treatment on compliers, who are individuals who comply with their assigned treatment group.
The Local Average Treatment Effect (LATE) is a statistical concept used in causal inference to measure the causal effect of a treatment or intervention on a specific subgroup of individuals who are referred to as compliers.
Compliers are individuals who comply with the treatment assignment they are given, either by receiving the treatment if assigned to the treatment group, or by not receiving the treatment if assigned to the control group.
Individuals might not comply with their assigned group by e.g. refusing to accept the treatment, or by obtaining the same treatment somewhere else if assigned to the control group.
LATE is also known as the Complier Average Causal Effect (CACE) because it estimates the causal effect of the treatment on the compliers only. LATE is based on the assumption that there are some individuals who are only affected by the treatment if they comply with the assigned treatment group. This is known as the "compliance effect".
LATE is calculated by comparing the outcome of the compliers who receive the treatment to the outcome of the compliers who do not receive the treatment (i.e. the populations considered are filtered). This is done in order to estimate the causal effect of the treatment for the subset of the population that is most responsive to the treatment.
The concept of LATE is particularly relevant in situations where the treatment assignment is not completely random and some individuals may choose not to comply with their assigned treatment. In such cases, the average treatment effect estimated from the entire population may not accurately reflect the true causal effect of the treatment. LATE provides a way to estimate the causal effect of the treatment for the subgroup of individuals who would be most affected by the treatment if they comply with their assigned treatment group.
It is important to note that the validity of LATE relies on the assumption that there are no other confounding variables that affect the compliers differently than the non-compliers. This is known as the "exclusion restriction" assumption, which states that the only factor that differentiates the compliers from the non-compliers is their compliance with the treatment assignment. If this assumption is not valid, the estimate of LATE may be biased.