Fractional imputation methods for Longitudinal data analysis /
Abdallah Soliman Abdallah Yaseen
Fractional imputation methods for Longitudinal data analysis / طرق تعويض كسرية لتحليل البيانات الطولية Abdallah Soliman Abdallah Yaseen ; Supervised Ahmed Mahmoud Gad , Abeer Saleh Ahmed - Cairo : Abdallah Soliman Abdallah Yaseen , 2014 - 105 P. ; 25cm
Thesis (M.Sc.) - Cairo University - Faculty of Economics and Political Science - Department of Statistics
The defining characteristic of the longitudinal studies is that sample units are measured repeatedly over time. That is, data are collected for the same set of units for two or more occasions. However, longitudinal studies are plagued with the problem of missing values which complicates the analysis of the data. Dropout pattern occurs when the existence of the missing values means that the subject withdraws from the study. If the missingness is related to the missing values of the outcome variable, the missing mechanism is termed nonignorable. Getting the maximum likelihood estimators in the case of nonignorable missing mechanism requires special procedures. In this Thesis, the parametric fractional imputation method is proposed to handle the missing values problem in the longitudinal studies for nonrandom dropout. Moreover, the Jackknife method is developed to find the standard errors of the parameter estimates. The proposed method is validated by applying it on a real data set in addition to a simulation study
Longitudinal data Mastitis data Missing data
Fractional imputation methods for Longitudinal data analysis / طرق تعويض كسرية لتحليل البيانات الطولية Abdallah Soliman Abdallah Yaseen ; Supervised Ahmed Mahmoud Gad , Abeer Saleh Ahmed - Cairo : Abdallah Soliman Abdallah Yaseen , 2014 - 105 P. ; 25cm
Thesis (M.Sc.) - Cairo University - Faculty of Economics and Political Science - Department of Statistics
The defining characteristic of the longitudinal studies is that sample units are measured repeatedly over time. That is, data are collected for the same set of units for two or more occasions. However, longitudinal studies are plagued with the problem of missing values which complicates the analysis of the data. Dropout pattern occurs when the existence of the missing values means that the subject withdraws from the study. If the missingness is related to the missing values of the outcome variable, the missing mechanism is termed nonignorable. Getting the maximum likelihood estimators in the case of nonignorable missing mechanism requires special procedures. In this Thesis, the parametric fractional imputation method is proposed to handle the missing values problem in the longitudinal studies for nonrandom dropout. Moreover, the Jackknife method is developed to find the standard errors of the parameter estimates. The proposed method is validated by applying it on a real data set in addition to a simulation study
Longitudinal data Mastitis data Missing data