linearmodels.panel.results.PanelEffectsResults.f_pooled¶

property PanelEffectsResults.f_pooled : WaldTestStatistic¶

Test that included effects are jointly zero.

Returns:¶

Statistic value, distribution and p-value. It is an InapplicableTestStatistic if the effects do not use any degrees of freedom, e.g., a single entity with a constant, and an InvalidTestStatistic if the model has no residual degrees of freedom or its residual sum of squares is 0.

Return type:¶

linearmodels.shared.hypotheses.WaldTestStatistic, linearmodels.shared.hypotheses.InapplicableTestStatistic or linearmodels.shared.hypotheses.InvalidTestStatistic

Notes

Joint test that all included effects are zero. Only correct under an assumption of homoskedasticity.

The test statistic is defined as

\[F = \frac{(RSS_{pool}-RSS_{effect})/(df_{pool}-df_{effect})} {RSS_{effect}/df_{effect}}\]

where \(RSS_{pool}\) is the residual sum of squares from a no- effect (pooled) model. \(RSS_{effect}\) is the residual sum of squares from a model with effects. \(df_{pool}\) is the residual degree of freedom in the pooled regression and \(df_{effect}\) is the residual degree of freedom from the model with effects. The test has an \(F_{k,df_{effect}}\) distribution where \(k=df_{pool}-df_{effect}\).