calcSigma {tmle}

Calculate Variance-Covariance Matrix for MSM Parameters (calcSigma)

Description

An internal function called by the tmleMSM function to calculate the variance-covariance matrix of the parameter estimates based on the influence curve of the specified MSM.

Usage

calcSigma(hAV, gAVW, Y, Q, mAV, covar.MSM, covar.MSMA0, covar.MSMA1, I.V, 
     Delta, ub, id, family)

Arguments

hAV

values used in numerator of weights applied to the estimation procedure

gAVW

P(A=a | V,W,T)*P(Delta=1 | A,V,W,T)

Y

continuous or binary outcome variable

Q

estimated P(Y | A, V, W, T, Delta=1), typically targeted values Q* are passed in

mAV

predicted values for EY1 from the MSM using the targeted estimates for psi

covar.MSM

covariate values used as predictors for the MSM when A=a

covar.MSMA0

covariate values used as predictors for the MSM when A=0

covar.MSMA1

covariate values used as predictors for the MSM when A=1

I.V

indicator that observation is in stratum of interest

Delta

indicator of missing outcome. 1 - observed, 0 - missing

ub

upper bound on weights

id

subject identifier

family

‘gaussian’ for continuous outcomes, ‘binomial’ for binary outcomes

Value

sigma

influence-curve based variance-covariance matrix. See Rosenblum&vanderLaan2010 for details.

Author(s)

Susan Gruber

See Also

tmle, estimateQ, estimateG, tmleMSM