GEE with Missingness

As discussed previously, GEE is a useful method for analyzing discrete and continuous longitudinal data when the mean response is of primary interest. Unlike likelihood-based methodology, however, GEE produces biased parameter estimates when data are missing at random. Rotnitzky and Wypij (1994) developed expression which can be used to quantify this bias in various settings. Robins et al. (1995) proposed an extension of GEE that allows for unbiased parameter estimation for MAR data. They propose a class of weighted estimating equations which result in consistent estimation of mean structure parameters with a correctly specified missing data mechanism. Their approach reduces to a weighted version of GEE in which each element of the residual vector (Y¡ — m) is weighted by the inverse of the probability of having a positive response. Paik (1997) discusses alternative GEE based methodology that allows for unbiased parameter estimation for MAR data. She proposes imputation techniques, in which missing observations are imputed in the data set. More recently Rotnitzky, et al. (1998) have proposed GEE estimation for nonignorable missingness.

0 0

Post a comment