A parametric fractional imputation method for intermittent missingness in longitudinal data analysis / Hanan Emad Galal Ahmed ; Supervised Ahmed Mahmoud Gad

Thesis (M.Sc.) - Cairo University - Faculty of Economics and Political Science - Department of Statistics

Longitudinal data analysis had been widely developed in the past 20 years. This type of data is common in public health, medical, biological and social sciences. Such data have special nature as the individual may be observed during a long period of time. So there are many analysis techniques developed especially for such data. Missing values are common in longitudinal data. The missed values lead to biased results and complicate the analysis. Hence, there is a need for approaches capable of dealing with such missingness. The missing values have two patterns: intermittent and dropout. The missing data mechanisms are missing completely at random (MCAR), missing at random (MAR), and missing not at random (MNAR). The appropriate analysis relies heavily on the assumed mechanism and pattern. Imputation techniques have been emerging in recent years to deal with missing values. In imputation methods we are looking for the appropriate imputation for the missing value(s). Many imputation methods are available such as; the single imputation methods, the multiple imputation methods, the fractional imputation, and the parametric fractional imputation. Our objective in the thesis is to develop an imputation technique for the non-random intermittent missingness in longitudinal data using the parametric fractional imputation method, assuming that the response variable of interest follows standard normal distribution. The jackknife method is used to find the standard errors for parameter estimates. A simulation study is conducted to evaluate the performance of the proposed technique. Also, the proposed technique is applied on a real data set

Issued also as CD