get_ssc

get_ssc(
    ssc_dict,
    N,
    k,
    k_fe,
    k_fe_nested,
    n_fe,
    n_fe_fully_nested,
    G,
    vcov_sign,
    vcov_type,
)

Compute small sample adjustment factors.

Parameters

Name	Type	Description	Default
ssc_dict	dict	A dictionary created via the ssc() function.	required
N	int	The number of observations.	required
k	int	The number of estimated parameters (as in the first part of the model formula)	required
k_fe	int	The number of estimated fixed effects (as specified in the second part of the model formula).	required
k_fe_nested	int	The number of estimated fixed effects nested within clusters.	required
n_fe	int	The number of fixed effects in the model. I.e. ‘Y ~ X1 \| f1 + f2’ has 2 fixed effects.	required
n_fe_fully_nested	int	The number of fixed effects that are fully nested within clusters.	required
G	int	The number of clusters.	required
vcov_sign	array - like	A vector that helps create the covariance matrix.	required
vcov_type	str	The type of covariance matrix. Must be one of “iid”, “hetero”, “HAC”, or “CRV”.	required

Returns

Name	Type	Description
	tuple of np.ndarray and int	A small sample adjustment factor and the effective number of coefficients k used in the adjustment.

Raises

Name	Type	Description
	ValueError	If vcov_type is not “iid”, “hetero”, or “CRV”, or if G_df is neither “conventional” nor “min”.

Examples

Called internally by the estimation functions. Use it directly to reproduce an adjustment factor by hand.

import pyfixest as pf
from pyfixest.utils.utils import get_ssc

# cluster-robust adjustment: 1000 observations, 3 coefficients, 20 clusters
adj, k_used, G_used = get_ssc(
    ssc_dict=pf.ssc(),
    N=1000,
    k=3,
    k_fe=0,
    k_fe_nested=0,
    n_fe=0,
    n_fe_fully_nested=0,
    G=20,
    vcov_sign=1,
    vcov_type="CRV",
)
adj, k_used, G_used

(array([1.05474318]), 3, np.int64(19))

Configure the behaviour with ssc(). See On Small Sample Corrections for the formulas.