linearmodels.shared.hypotheses.NormalTestStatistic¶

class linearmodels.shared.hypotheses.NormalTestStatistic(stat: float, null: str, *, name: str | None = None, two_sided: bool = True)[source]¶

Test statistic holder for tests with an asymptotic standard normal law.

Parameters:¶

stat: float¶

The test statistic.

null: str¶

A statement of the test’s null hypothesis.

name: str | None = None¶

Name of test.

two_sided: bool = True¶

Flag indicating whether p-values and critical values are computed using a two-sided test. The default is True.

Notes

A two-sided test rejects the null if the absolute value of the statistic \(s\) is large. The p-value is \(2\Pr(Z > |s|)\), where \(Z\) has a standard normal distribution, and the critical values for tests of size 10%, 5% and 1% are the 0.95, 0.975 and 0.995 quantiles of the standard normal.

A one-sided test rejects the null if the statistic is large. The p-value is \(\Pr(Z > s)\), which is close to 1 if the statistic is large and negative, and the critical values are the 0.9, 0.95 and 0.99 quantiles.

Methods

Properties

critical_values	Critical values for common test sizes
null	Null hypothesis
pval	P-value of test statistic
stat	Test statistic