import pyfixest as pf
data = pf.get_data()
fit = pf.feols("Y ~ X1 + X2", data=data, vcov="hetero")
fit.evalue()Intercept 6.193976e+12
X1 5.321643e+29
X2 2.465287e+12
Name: e_value, dtype: float64Feols.evalue
evalue(mixture_precision=1.0)Compute coefficient-wise SAVI e-values.
Parameters
| Name | Type | Description | Default |
|---|---|---|---|
| mixture_precision | float | Positive mixture precision fixed before sequential monitoring. Defaults to 1. Use pyfixest.optimal_mixture_precision() to minimize confidence-sequence width at a target sample size. |
1.0 |
Returns
| Name | Type | Description |
|---|---|---|
| pd.Series | One e-value per coefficient. |
Notes
SAVI currently supports unweighted, non-IV feols models without absorbed fixed effects. The covariance estimator must be iid or heteroskedasticity robust (hetero, HC1, HC2, or HC3). Note that for HC2/HC3, pyfixest’s default small-sample correction scales the variance by n / (n - k) while the R implementation in avlm does not. Inference is pointwise / by coefficient.