Book Chapter10.1016/B978-0-12-801342-7.00014-9
Methods for handling missing data
Xian Liu
- 01 Jan 2016
pp 441-473
27
TL;DR: It is displayed that the selection model only has a very limited capacity to handle non-ignorable missing data, whereas the pattern-mixture and the nonparametric models are promising approaches for handling MNAR.
read more
Abstract: Chapter 14 is devoted to the description of various models and methods for handling missing data. First, I supply the mathematical definitions of three missing-data mechanisms: missing completely at random (MCAR), missing at random (MAR), and missing not at random (MNAR). While MCAR and MAR are ignorable, MNAR cannot be ignored in performing longitudinal data analysis. Next, a variety of methods handling missing at random are introduced, including some simplistic approaches (such as list-wise deletion, mean substitution, and hot deck imputation), the last observation carried forward approach, and multiple imputations. Lastly, three statistical models for handling MNAR are delineated: the two-step parametric selection model, pattern-mixture modeling, and the two-stage nonparametric mixed model. An empirical example is provided to illustrate and compare those three methods. It is displayed that the selection model only has a very limited capacity to handle non-ignorable missing data, whereas the pattern-mixture and the nonparametric models are promising approaches for handling MNAR.
read more
Chat with Paper
AI Agents for this Paper
Find similar papers on Google Scholar, PubMed and Arxiv
Write a critical review of this paper
Analyze citations of this paper to find unaddressed research gaps
Citations
Longitudinal Association Between Teen Sexting and Sexual Behavior
Jeffrey Temple,Hye Jeong Choi +1 more
TL;DR: This study extends cross-sectional literature and supports the notion that sexting fits within the context of adolescent sexual development and may be a viable indicator of adolescentSexual activity.
173
Detecting cyberattacks using anomaly detection in industrial control systems: A Federated Learning approach
Truong Thu Huong,Ta Phuong Bac,Dao Minh Long,Tran Duc Luong,Nguyen Minh Dan,Le Anh Quang,Le Thanh Cong,Bui Doan Thang,Kim Phuc Tran +8 more
TL;DR: An anomaly detection architecture for the IIoT-based SM is proposed to deploy one of the top most concerned networking technique - a Federated Learning architecture - that can detect anomalies for time series data typically running inside an industrial system.
142
Dissociation Predicts Later Attention Problems in Sexually Abused Children.
TL;DR: Findings from this study indicate that children who experienced sexual abuse at an earlier age, by someone within their family, and/or report symptoms of dissociation during disclosure are especially likely to benefit from intervention.
79
Developing the energy profile of mountainous areas
TL;DR: In this article, the authors quantified the increase of energy needs with altitude, in an attempt to highlight the greater vulnerability of mountainous areas to energy poverty, and suggested mathematical models to estimate heating and cooling energy demand in a simple and reliable way.
26
An In-Home Medication Dispensing System to Support Medication Adherence for Patients With Chronic Conditions in the Community Setting: Prospective Observational Pilot Study
19 May 2022
TL;DR: In this article , the authors examined the use of a smart medication dispenser as a medication adherence and self-management support tool for community dwelling adults over a 6-month period, in addition to usability, usefulness, satisfaction, and impact on caregiver support.
15
References
Missing Data: Our View of the State of the Art.
JL Schafer,JW Graham +1 more
TL;DR: Statistical methods for missing data have improved, but misconceptions persist. The authors review methods, offer advice, and recommend maximum likelihood and Bayesian multiple imputation, while discouraging older procedures and discussing newer developments for non-MAR data.
Analysis of Incomplete Multivariate Data.
David E. Booth
TL;DR: This paper analyzes incomplete multivariate data, exploring EM and data augmentation methods for normal and categorical data, as well as loglinear models for mixed data, providing a comprehensive overview of statistical techniques for handling incomplete data.