Computation of the degrees of freedom of the chi-squared distribution relative to the robust-based variance
dfSigmaRobust(contrast, vcov, score)[numeric vector] the linear combination of parameters to test
[numeric matrix] the variance-covariance matrix of the parameters.
[numeric matrix] the individual score for each parameter.
When contrast is the identity matrix, this function compute the moments of the sandwich estimator and the degrees of freedom of the approximate t-test as described in (Pan, 2002) section 2 and 3.1.
Wei Pan and Melanie M. Wall, Small-sample adjustments in using the sandwich variance estiamtor in generalized estimating equations. Statistics in medicine (2002) 21:1429-1441.