Full Record

Author | Dipita, Theophile B |

Title | Missing Data in Clinical Trial: A Critical Look at the Proportionality of MNAR and MAR Assumptions for Multiple Imputation |

URL | https://digitalcommons.georgiasouthern.edu/etd/1483 |

Publication Date | 2016 |

Degree | Doctor of Public Health in Biostatistics (Dr.P.H.) |

Discipline/Department | Department of Biostatistics (COPH) |

Degree Level | doctoral |

University/Publisher | Georgia Southern University |

Abstract | Randomized control trial is a gold standard of research studies. Randomization helps reduce bias and infer causality. One constraint of these studies is that it depends on participants to obtain the desired data. Whatever the researcher can do, there is a possibility to end up with incomplete data. The problem is more relevant in clinical trials when missing data can be related to the condition under study. The benefits of randomization is compromised by missing data. Multiple imputation is a valid method of treating missing data under the assumption of MAR. Unfortunately this is an unverified assumptions. Current practice advise the use of sensitivity analysis to adjust for departure from the MAR missingness. Data collectors’ knowledge, researchers’ insight, and statisticians’ experience can improve assumptions of missing data mechanisms. In practice, a mixture of possible assumptions can be made about missingness. In an attempt to exploit supplemental knowledge for the amelioration of inference from data with missing values, this dissertation explores the possibility of combining various proportion of MAR and MNAR assumptions. This exploration will be done by simulating data having normal, chi-square, and t distributions with varying proportion of MAR and MNAR assumptions. We propose influential exponential tilting in which the model for the non-respondents correspond to an exponential tilting of the model for respondents, and the specified function in the tilted model is the influential function of the parameter to be estimated. The proposed method will be combined with MI to overcome the issue of MNAR. |

Subjects/Keywords | Missing at random; Missing not at random; Influential exponential tilting; Multiple imputation; Biostatistics; Clinical Trials; Statistical Methodology; Jack N. Averitt College of Graduate Studies, Electronic Theses & Dissertations, ETDs, Student Research |

Contributors | Hani Samawi; Haresh Rochani |

Rights | License: http://creativecommons.org/licenses/by/4.0/ [Always confirm rights and permissions with the source record.] |

Country of Publication | us |

Record ID | oai:digitalcommons.georgiasouthern.edu:etd-2580 |

Repository | gsu |

Date Retrieved | 2020-01-02 |

Date Indexed | 2020-01-06 |

Created Date | 2016-01-01 08:00:00 |

Sample Search Hits | Sample Images | Cited Works

…advantages of multiple *imputation* are that a *random* draw of imputations
increases the efficiency of the estimation and it takes into account variability due to missing data,
providing valid inference under MAR. MI also allows researchers to easily study the…

…33
METHOD OF MULTIPLE *IMPUTATION* ............................................................................33
3.1 Development ................................................................................................................33
3.1.1…

…Schematization of Multiple *Imputation*. The question marks indicate missing values
and m is the number of imputations (Schafer & Graham, 2002)..........................................10
Figure 2: Normal distribution of the population with 20% stochastic…

…a systematic difference between the observed and the missing
data could bias the results (Tabachnick & Fidel, 2007; Carpenter & Kenward, 2013). Although
no *imputation* method can equally compensate for the missing values, researchers have…

…complete case analysis, which retains
only observations with no missing occurrence, and available-case analysis, which considers all
available data for each analysis. Furthermore, there were several applications of single
*imputation* methods where the…

…missing values are replaced by some chosen values.
An important step in the missing data resolution came with the idea of multiple
*imputation* (MI), initially proposed by Rubin (1978). MI is a principled method to deal with
missing data…

…x29;. The method ends by analyzing each of the m complete datasets using the
usual methods for complete data and combining the results to obtain one single effect.
Figure 1: Schematization of Multiple *Imputation*. The question marks indicate missing…

…the increased problems
of nonresponse in surveys, the lack of satisfaction with methods used, and the inflated number of
computational tools for analysis with missing data. Complete case analysis and single *imputation*
methods usually assume missing…