StaggeredDifferenceInDifferences.plot_tau_surface#

StaggeredDifferenceInDifferences.plot_tau_surface(hdi_prob=0.94)[source]#

Plot the ETWFE effect surface, one panel per adoption cohort.

The aggregated event study collapses every cohort into a single line, which is exactly the information that the extended two-way fixed effects parametrisation exists to preserve. This figure puts each cohort’s tau[g, k] profile in its own panel, so cohort heterogeneity – effects that grow faster for later adopters, say – is visible directly.

Parameters:

hdi_prob (float) – Probability mass for the credible band on the Bayesian path. Defaults to causalpy.constants.HDI_PROB. Ignored on the OLS path, which draws a 95% normal-approximation interval from the coefficient standard errors.

Returns:

The figure and one axis per cohort, in cohort order.

Return type:

tuple[plt.Figure, list[plt.Axes]]

Raises:

ValueError – If the experiment was run with estimator="imputation", which produces no tau_surface_.