Table 1 Hypothetical data for the toy example comparing the SATE (\(\beta\)) to subgroup-specific ATEs (\(\tau _A, \tau _B, \tau _C\)). Bolded values indicate the lowest absolute difference between subgroup-specific effect and SATE in each row
From: Quantifying and reducing inequity in average treatment effect estimation
Study | \(\tau _A\) | \(\tau _B\) | \(\tau _C\) | \(\beta\) | \(|\tau _A - \beta |\) | \(|\tau _B - \beta |\) | \(|\tau _C - \beta |\) |
|---|---|---|---|---|---|---|---|
1 | -3.51 | -2.96 | 4.04 | -2.65 | 0.87 | 0.31 | 6.69 |
2 | -4.39 | -0.13 | 4.62 | -2.64 | 1.75 | 2.51 | 7.26 |
3 | -1.84 | -1.08 | -0.26 | -1.53 | 0.31 | 0.45 | 1.27 |
4 | -1.16 | -4.71 | -4.22 | -2.18 | 1.02 | 2.53 | 2.05 |
5 | -1.75 | 0.85 | -1.38 | -1.19 | 0.56 | 2.05 | 0.19 |