Variability in longitudinal outcomes is often perceived as a nuisance parameter in statistical models and is not usually estimated. However, within-subject variability may be informative. For example, pediatric systemic lupus erythematosus (SLE) is a chronic disease commonly involving kidney inflammation and is characterized by periods of flare and periods of disease quiescence. This makes modeling variability of kidney function clinically interesting. In this project, we aim to model outcome variability in the presence of irregular observation. We propose several estimating functions, each of which comprises a marginal mean model and variability models. The variability models to consider are sample variance and median absolute deviation (MAD) about mean. We obtain closed form expression of sandwich variance estimator. A simulation study compared variability estimators for estimating equations and confirmed a good performance in terms of their biases, variances and 95% coverages.