robust_forCluster {evian}

Robust adjustment function

Description

The robust function computes an adjustment that is applied to the likelihood function to account for the cluster nature of the data.

Usage

robust_forCluster(formula, data, family)

Arguments

formula

a formula specifying the response and possible covariates to keep in the output data frame. This is directly obtained from the evianfunction.

data

data frame; from the output of subsetData.

family

string; the link function for glm.

Details

The robust function is called from within evian functions. It computes a robust adjustment factor that is applied to the likelihood function to account for the cluster nature of the data. The family ID column (FID) specifies the clusters. The robust adjustment factor is the ratio of the regular variance estimator of the maximum likelihood estimate (MLE) to the sandwich variance estimator of the MLE, where the ‘meat’ of the sandwich variance estimator is corrected for clustering in the data (Blume et.al, 2007). If the data is not clustered (i.e. the observations are independent) then the adjustment factor can still be applied to make the working model robust to possible model misspecifications (Royall and Tsou, 2003).

Value

A numeric constant.

Author(s)

Zeynep Baskurt [e-mail]

References

Blume, J. D., Su, L., Remigio, M. O., & McGarvey, S. T. (2007). Statistical evidence for GLM regression parameters: A robust likelihood approach. Statistics in Medicine, 26, 2919-2936.

Royall, R. , Tsou, T. S. (2003). Interpreting statistical evidence by using imperfect models: robust adjusted likelihood functions. J Roy Stat Soc B; 65: 391-404.