Missing Data Mechanisms for Analysing Longitudinal Data with Incomplete Observations in Both Responses and Covariates

Summary Missing observations in both responses and covariates arise frequently in longitudinal studies. When missing data are missing not at random, inferences under the likelihood framework often require joint modelling of response and covariate processes, as well as missing data processes associated with incompleteness of responses and covariates. Specification of these four joint distributions is a nontrivial issue from the perspectives of both modelling and computation. To get around this problem, we employ pairwise likelihood formulations, which avoid the specification of third or higher order association structures. In this paper, we consider three specific missing data mechanisms which lead to further simplified pairwise likelihood (SPL) formulations. Under these missing data mechanisms, inference methods based on SPL formulations are developed. The resultant estimators are consistent, and enjoy better robustness and computation convenience. The performance is evaluated empirically though simulation studies. Longitudinal data from the National Population Health Survey and Waterloo Smoking Prevention Project are analysed to illustrate the usage of our methods.

[1]  Grace Y. Yi,et al.  A pairwise likelihood approach for longitudinal data with missing observations in both response and covariates , 2013, Comput. Stat. Data Anal..

[2]  David Feeny,et al.  The natural history of health-related quality of life: a 10-year cohort study. , 2009, Health reports.

[3]  G. Yi,et al.  A Pairwise Likelihood Method For Correlated Binary Data With/withoutMissing Observations Under Generalized Partially Linear Single-indexModels , 2011 .

[4]  Joseph G Ibrahim,et al.  Maximum Likelihood Methods for Nonignorable Missing Responses and Covariates in Random Effects Models , 2003, Biometrics.

[5]  Richard J. Cook,et al.  A robust pairwise likelihood method for incomplete longitudinal binary data arising in clusters , 2011 .

[6]  K. Brown,et al.  Effectiveness of a social influences smoking prevention program as a function of provider type, training method, and school risk. , 1999, American journal of public health.

[7]  LIKELIHOOD-BASED INFERENCE WITH NONIGNORABLE MISSING RESPONSES AND COVARIATES IN MODELS FOR DISCRETE LONGITUDINAL DATA , 2006 .

[8]  Richard J. Cook,et al.  Weighted Generalized Estimating Functions for Longitudinal Response and Covariate Data That Are Missing at Random , 2010 .

[9]  S. Lipsitz,et al.  Missing-Data Methods for Generalized Linear Models , 2005 .

[10]  Ying Yuan,et al.  Model‐based estimates of the finite population mean for two‐stage cluster samples with unit non‐response , 2007 .

[11]  D. Bates,et al.  Fitting Linear Mixed-Effects Models Using lme4 , 2014, 1406.5823.

[12]  Roderick J. A. Little,et al.  Statistical Analysis with Missing Data: Little/Statistical Analysis with Missing Data , 2002 .

[13]  Xiao-Hua Zhou,et al.  Doubly Robust Estimates for Binary Longitudinal Data Analysis with Missing Response and Missing Covariates , 2011, Biometrics.

[14]  J. Ware,et al.  Random-effects models for longitudinal data. , 1982, Biometrics.

[15]  A. Gelman,et al.  Multiple Imputation with Diagnostics (mi) in R: Opening Windows into the Black Box , 2011 .

[16]  Grace Y Yi,et al.  Estimation methods for marginal and association parameters for longitudinal binary data with nonignorable missing observations , 2013, Statistics in medicine.

[17]  Xin Gao,et al.  Composite Likelihood EM Algorithm with Applications to Multivariate Hidden Markov Model , 2009 .

[18]  D. Cox,et al.  A note on pseudolikelihood constructed from marginal densities , 2004 .