Missing data can happen in longitudinal clinical trials due to various reasons. Even in the setting of smaller well-controlled studies, like a multiple-dose titration experiment, the missing data pattern may not be so clear. When the data is missing not at random because of a possible link to treatment intolerability, for example a down dosing due to drug-related heart rate increase, some statistical models may not be applied appropriately due to potential violation of model assumptions. The performance of the modeling is likely related to the sample size and the extent of the missing rate observed. In this presentation, we will set up the different scenarios under the different sample sizes, missing rates, and treatment effects in the simulation data. Under the null or alternative testing situations, we will compare the mixed effect model and some multiple imputation methods in terms of the biasness, variability of the estimation, coverage, type 1 error, and corresponding power. A real case study will also been used for the comparison in terms of variability of the corresponding treatment estimates.