Comparison of Different LGM-Based Methods with MAR and MNAR Dropout Data

Abstract

The missing not at random (MNAR) mechanism may bias parameter estimates and even distort study results. This study compared the maximum likelihood (ML) selection model based on missing at random (MAR) mechanism and the Diggle–Kenward selection model based on MNAR mechanism for handling missing data through a Monte Carlo simulation study. Four factors were considered, including the missingness mechanism, the dropout rate, the distribution shape (i.e., skewness and kurtosis), and the sample size. The results indicated that: (1) Under the MAR mechanism, the Diggle–Kenward selection model yielded similar estimation results with the ML approach; Under the MNAR mechanism, the results of ML approach were underestimated, especially for the intercept mean and intercept slope (μi and μs). (2) Under the MAR mechanism, the 95% CP of the Diggle–Kenward selection model was lower than that of the ML method; Under the MNAR mechanism, the 95% CP for the two methods were both under the desired level of 95%, but the Diggle–Kenward selection model yielded much higher coverage probabilities than the ML method. (3) The Diggle–Kenward selection model was easier to be influenced by the non-normal degree of target variable's distribution than the ML approach. The level of dropout rate was the major factor affecting the parameter estimation precision, and the dropout rate-induced difference of two methods can be ignored only when the dropout rate falls below 10%.

Sleep disturbances and dementia risk: A multicenter study

“…One common problem with longitudinal study designs is the risk of dropouts, and that these dropouts are not random (e.g., missing due to worsened somatic health or morbidity). Li et al (2017) have suggested that the level of dropouts was the strongest factor for affecting the parameter estimates using the maximum likelihood estimation. In their analyses, level of dropouts could only be ignored if the dropout rate falls below 10%.…”

Differential Patterns of the Division of Parenthood in Chinese Family: Association With Coparenting Behavior

“…Mother’s working hours per week were negatively correlated with mother engagement and responsibility, whereas father’s working hours per week and child gender were not correlated with parental involvement. The kurtosis and skewness values for all the variables were lower than 3 (Li et al, 2017), indicating the normal distribution of these variables. The multi-collinearity was identified by the values of variance inflation factor (VIF).…”

Depression and Anxiety Symptoms of British Adoptive Parents: A Prospective Four-Wave Longitudinal Study

“…Due to this, data analysis employed selection models for both MAR analysis using maximum likelihood with robust standard errors (MLR) and MNAR using the Diggle-Kenward selection as recommended [50]. Studies have found that the Diggle-Kenward performs better than the MAR based maximum likelihood estimator method under the MNAR mechanism, especially with smaller sample sizes and samples with more than 10% attrition [51]. As the sensitivity analysis showed minimal difference between the MAR and MNAR analyses, only the MAR are reported, and MNAR analyses are available upon request.…”