- [[unbiased estimator]], [[conditional average treatment effect]], [[covariate adjustment]]
# Idea
The average treatment effect (ATE) is also known as the average causal effect (ACE).
The letter $D$ is often used to indicate treatment status:
$
D_{i}= \begin{cases}1 & \text { if } i \text { is insured } \\ 0 & \text { otherwise }\end{cases}
$
For each subject or unit $i$, the causal effect of the treatment $\tau_i$ is defined as the difference between two [[potential outcomes]]:
$
\tau_{i} \equiv Y_{i}(1)-Y_{i}(0)
$
# ATE
$
\mathrm{ATE} \equiv \frac{1}{N} \sum_{i=1}^{N} \tau_{i}
$
$
A T E=\mathbb{E}\left[Y_{1}-Y_{0}\right]=\mathbb{E}_{x \sim p(x)}[\operatorname{CATE}(x)]
$
Average treatment effect is the expected effect of treatment $T$ on $Y$:
$ATE:=\mathbb{E}\left[Y_{1}-Y_{0}\right]$
If we assume [[ignorability assumption|ignorability]], we can derive/use the [[adjustment formula]].
# Averages conditional on treatment condition
Average among treatment subjects:
$
A v g_{n}\left[Y_{i} \mid D_{i}=1\right]
$
Average among control subjects:
$
A v g_{n}\left[Y_{i} \mid D_{i}=0\right]
$
These quantities are averages conditional on insurance status.
# References
- [[Gerber 2012 field experiments]]