Thesis (Ph. D.)--Duke University, 2007.

Thesis (M. Sc.)--Carleton University, 2001. / Includes bibliographical references (p. 108-110). Also available in electronic format on the Internet.

The effectiveness of seven methods of handling missing data was investigated in a factorial design using random samples selected from the National Education Longitudinal Study of 1988 (NELS-88). Methods evaluated were listwise deletion, pairwise deletion, mean substitution, Buck's procedure, mean regression, one iteration regression, and iterative regression. Factors controlled were number of variables (4 and 8), average intercorrelation (0.2 and 0.4), sample size (200 and 2000), and proportion of incomplete cases (10%, 20%, and 40%). The pattern of missing values was determined by the pattern existing in the variables selected from NELS-88 data base. Covariance matrices resulting from the use of each missing data method were compared to the 'true' covariance matrix using multi-sample analysis in LISREL 7. Variable means were compared to the 'true' means using the MANOVA procedure in SPSS/PC+. Statistically significant differences (p≤.05) were detected in both comparisons. The most surprising result of this study was the effectiveness (p>.05) of pairwise deletion whenever the sample size was large thus supporting the contention that the error term disappears as sample size approaches infinity (Glasser, 1964). Listwise deletion was also effective (p>.05) whenever there were four variables or the sample size was small. Almost as surprising was the relative ineffectiveness (p<.05) of the regression methods. This is explained by the difference in proportion of incomplete cases versus the proportion of missing values, and by the distribution of the missing values within the incomplete cases. / Ph. D.

published_or_final_version / abstract / Mathematics / Master / Master of Philosophy

by Siu-fung Chan. / Thesis (M.Phil.)--Chinese University of Hong Kong, 1991. / Bibliography: leaves 38-41. / Chapter Chapter 1 --- Introduction --- p.1 / Chapter Chapter 2 --- Analysis of the Structural Equation Model with Continuous Data --- p.6 / Chapter §2.1 --- The Model --- p.6 / Chapter §2.2 --- Mehtods of Handling Incomplete Data --- p.8 / Chapter §2.3 --- Design of the Monte-Carlo Study --- p.12 / Chapter §2.4 --- Results of the Monte-Carlo Study --- p.15 / Chapter Chapter 3 --- Analysis of the Structural Equation Model with Polytomous Data --- p.24 / Chapter §3.1 --- The Model --- p.24 / Chapter §3.2 --- Methods of Handling Incomplete Data --- p.25 / Chapter §3.3 --- Design of the Monte-Carlo Study --- p.27 / Chapter §3.4 --- Results of the Monte-Carlo Study --- p.31 / Chapter Chapter 4 --- Summary and Discussion --- p.36 / References --- p.38 / Tables --- p.42 / Figures --- p.78

Leung Man-Kit. / Thesis (M.Phil.)--Chinese University of Hong Kong, 2003. / Includes bibliographical references (leaves 37-41). / Abstracts in English and Chinese. / Chapter 1 --- Introduction --- p.1 / Chapter 1.1 --- Basic concepts of survival data --- p.3 / Chapter 1.2 --- Missing Complete at Random (MCAR) --- p.8 / Chapter 1.3 --- Missing at Random (MAR) --- p.9 / Chapter 2 --- The maximaization of the marginal likelihood --- p.11 / Chapter 2.1 --- Survival function --- p.11 / Chapter 2.2 --- Missing covariate pattern --- p.13 / Chapter 2.3 --- Set of survival time with rank restrictions --- p.13 / Chapter 2.4 --- Marginal likelihood --- p.14 / Chapter 2.5 --- Score function --- p.15 / Chapter 3 --- The MCMC stochastic approximation approach --- p.17 / Chapter 4 --- Simulations Studies --- p.22 / Chapter 4.1 --- MCAR : Simulation 1 --- p.23 / Chapter 4.2 --- MCAR : Simulation 2 --- p.24 / Chapter 4.3 --- MAR : Simulation 3 --- p.26 / Chapter 4.4 --- MAR : Simulation 4 --- p.27 / Chapter 5 --- Example --- p.30 / Chapter 6 --- Discussion --- p.33 / Appendix --- p.35 / Bibliography --- p.37

Most statistical surveys and data collection studies encounter missing data. A common solution to this problem is to discard observations with missing data while reporting the percentage of missing observations in different output tables. Imputation is a tool used to fill in the missing values. This dissertation introduces the missing data problem as well as traditional imputation methods (e.g. hot deck, mean imputation, regression, Markov Chain Monte Carlo, Expectation-Maximization, etc.). The use of artificial neural networks (ANN), a data mining technique, is proposed as an effective imputation procedure. During ANN imputation, computational effort is minimized while accounting for sample design and imputation uncertainty. The mechanism and use of ANN in imputation for complex survey designs is investigated. Imputation methods are not all equally good, and none are universally good. However, simulation results and applications in this dissertation show that regression, Markov chain Monte Carlo, and ANN yield comparable results. Artificial neural networks could be considered as implicit models that take into account the sample design without making strong parametric assumptions. Artificial neural networks make few assumptions about the data, are asymptotically good and robust to multicollinearity and outliers. Overall, ANN could be time and resources efficient for an experienced user compared to other conventional imputation techniques. / Graduation date: 2005

Thesis (Ph. D.)--Ohio State University, 2006. / Title from first page of PDF file. Includes bibliographical references (p. 233-237).

Not available / text

Thesis (Ph.D.) - Carleton University, 2005. / Includes bibliographical references (p. 252-258). Also available in electronic format on the Internet.