PRISE STUDY: A missing data problem Open Access

Abstract

The PRISE study was an intervention to encourage African-American women of Grady Healthcare to improve their fitness. This study was subject to large drop outs, presenting a particular problem for analysis. This problem with missing data inspired drop-out analysis and a comparative analysis of imputation. We will analyze the drop-out process as an outcome and determine if any factors measured in the study are associated with drop-out. For imputation, the methods we compare are Average (Avg), Previous (Prev), Post (Post), Last and Next (LaN), Last Value Carried Forward (LVCF), and Next Value Carried Backward (NVCB). We will compare the methods given varying amounts of missingness. We give evidence to support that simple imputation methods can be an effective way to deal with MAR data. The best imputation methods as found in this study were LaN and Post. As the amount of missingness increased both LaN and Post maintained relatively low bias, (proportionate variance) PV close to 1, and consistent variance.

Table of Contents

Table of Contents List of Tables.....................................................................................................vi List of Figures....................................................................................................vii 1. Introduction ...................................................................................................1 1.1. Background: PRISE Study................................................................................1 1.2. Missing Data.................................................................................................2 2. Methods.........................................................................................................5 2.1. Data Collection.............................................................................................5 2.2. Analysis.......................................................................................................6 2.2.1. Overall Analysis..........................................................................................6 2.2.2. Drop-out Analysis.......................................................................................7 2.2.3. Imputation................................................................................................7 2.2.4. Analysis on Complete Data...........................................................................8 3. Results..........................................................................................................9 3.1. Study Population..........................................................................................9 3.2. Physiological Measures.................................................................................11 3.3. Psychological Measures................................................................................12 3.4. Comparative Analysis...................................................................................12 3.5. Complete Analysis.......................................................................................14 3.6. Drop-Out Analysis.......................................................................................16 3.7. Imputation................................................................................................17 4. Discussion...................................................................................................22 References......................................................................................................24

About this Dissertation

Rights statement

• Permission granted by the author to include this thesis or dissertation in this repository. All rights reserved by the author. Please contact the author for information regarding the reproduction and use of this thesis or dissertation.

School	Laney Graduate School
Department	Biostatistics
Degree	MS
Submission	Dissertation
Language	English
Research Field	Biology, Biostatistics Health Sciences, Public Health
Keyword	Imputation Fitness Intervention African-American Women Missing Data
Committee Chair / Thesis Advisor	Hertzberg, Vicki S, Emory University
Committee Members	Bowman, F Dubois, Emory University

Last modified

Primary PDF

Thumbnail	Title	Date Uploaded	Actions
	PRISE STUDY: A missing data problem ()	2018-08-28 12:05:50 -0400	Press to Select an action Download

Supplemental Files

Thumbnail	Title	Date Uploaded	Actions