Activity Number:
|
284
|
Type:
|
Contributed
|
Date/Time:
|
Tuesday, August 4, 2009 : 8:30 AM to 10:20 AM
|
Sponsor:
|
Section on Survey Research Methods
|
Abstract - #305284 |
Title:
|
Missing Data and Complex Samples: The Impact of Listwise Deletion vs. Subpopulation Analysis on Statistical Bias and Hypothesis Test Results When Data Are MCAR and MAR
|
Author(s):
|
Bethany A. Bell*+ and Jeffrey D. Kromrey and John M. Ferron
|
Companies:
|
University of South Carolina and University of South Florida and University of South Florida
|
Address:
|
College of Education, Columbia, SC, 29208,
|
Keywords:
|
survey research ; complex samples ; missing data ; variance estimation
|
Abstract:
|
Secondary data analysis of complex sample survey results is common among social scientists. Yet, the degree to which unbiased estimates and accurate inferences can be made from complex samples depends on the care researchers take when analyzing the data, including strategies for the treatment of missing data. Several studies have illustrated that the results of subpopulation analysis may diverge from those obtained through listwise deletion. However, given the paucity of simulation work in this area, it is not clear how frequently discernable discrepancies will arise. This Monte Carlo study focuses on the impact of listwise deletion versus a subpopulation analysis, when the data are MCAR and MAR, in the context of multiple regression analysis of complex sample data. Results are presented in terms of statistical bias in parameter estimates and both confidence interval width and coverage.
|