I don't understand the notation of the Average Treatment effect on the Treated

48 Views Asked by At

The Average Treatment effect on the Treated is a formula in causal inference which calculates the effect of some treatment on the treated group.

$ATT = E[Y(A{=}1)-Y(A{=}0)| A{=}1]$

I don't understand how the notation of this expression relates to how it is calculated. For instance, given the sample data:

Y(A=1) | Y(A=0)
--------------------
10.0   | 
20.0   | 
30.0   |
       | 1
       | 2
       | 3

The $ATT = 54=(10+20+30)-(1+2+3)$, where the values under $Y(A=1)$ correspond to the effect for a few samples with treatment $A=1$, and the values under $Y(A=0)$ correspond to the effect for a few samples with treatment $A=0$.

The expression $|A=1$ in the original expression makes me think $({=}1)−({=}0)$ should only be calculated over values in which $A=1$. To me, the following expression makes more sense:

$ATT = E[Y(A{=}1) | A{=}1 - Y(A{=}0)| A{=}0]$

or maybe even

$ATT = E[Y(A{=}1) | A{=}1] - E[Y(A{=}0)| A{=}0]$

If you interpret $E$ as the calculation of the average effect across a distribution. What am I misunderstanding about the relevant notation?