|
Abstract:
|
When we have data with missing values, the assumption that data are missing at random is very convenient. It is, however, sometimes questionable because some of missing values could be strongly related to the underlying true values. We recently introduced methods for nonignorable multivariate missing data, called a Latent-Class Selection Model (LCSM). In the LCSM, the missingness is assumed to be related to the variables in question, and to additional covariates, through a latent variable measured by the missingness indicators. The methodology developed here is useful for investigating the sensitivity of one's estimates to untestable assumptions about the missing data mechanism. A small simulation study is conducted to compare to its performance of the proposed method to that of MAR-based alternatives.
|
