get_true_means

cross_sectional.simulate.get_true_means(
    target: CausalTarget = CausalTarget.QTE,
    n_oracle: int = 1000000,
    effect_scale: float = 1.0,
    error_var: float = 1.0,
    seed: int = 0,
)

Simulation-based potential-outcome means and the average effect.

With target=QTE the means are over the whole population (effect is the ATE); with target=QTT they are over the treated (effect is the ATT). The truth is recovered by simulating n_oracle units.

Parameters

Name	Type	Description	Default
target	CausalTarget	Estimand to recover, QTE (population) or QTT (treated).	CausalTarget.QTE
n_oracle	int	Size of the simulated population.	1_000_000
effect_scale	float	Treatment effect scale used in simulate_covariate_data.	1.0
error_var	float	Error variance used in simulate_covariate_data.	1.0
seed	int	Seed for the simulated population.	0

Returns

Name	Type	Description
	polars.DataFrame	Columns m_c, m_t and effect (m_t - m_c).