FixVarianceCovarianceMatrix: FixVarianceCovarianceMatrix

Description Usage Arguments Details

View source: R/regression.R

Description

Makes some adjustments to permit a covariance-marix to be inverted, if required.

Usage

1
FixVarianceCovarianceMatrix(x, min.eigenvalue = 1e-12)

Arguments

x

A variance-covariance matrix of parameter estimates.

min.eigenvalue

Minimm acceptable eigenvalue.

Details

Sandwich and sandwich-like standard errors can result uninvertable covariance matrices (e.g., if a parameter represents a sub-group, and the sub-group has no residual variance). This function checks to see if there are any eigenvalues less than min.eigenvalue, which defaults to 1e-12. If there are, an attempt is made to guess the offending variances, and they are multiplied by 1.01.


Displayr/flipRegression documentation built on Aug. 14, 2018, 10:24 p.m.