Implications of Survey Sampling Design for Missing Data Imputation

Previous studies that analyzed multiple imputation using survey data did not take into account the survey sampling design. The objective of the current study is to analyze the impact of survey sampling design missing data imputation, using multivariate multiple imputation method. The results of the current study show that multiple imputation methods result in lower standard errors for regression analysis than the regression using only complete observation. Furthermore, the standard errors for all regression coefficients are found to be higher for multiple imputation with taking into account the survey sampling design than without taking into account the survey sampling design. Hence, sampling based estimation leads to more realistic standard errors.


Issue Date:
2013-05
Publication Type:
Conference Paper/ Presentation
PURL Identifier:
http://purl.umn.edu/149679
Total Pages:
16
Series Statement:
Paper
1870




 Record created 2017-04-01, last modified 2017-08-04

Fulltext:
Download fulltext
PDF

Rate this document:

Rate this document:
1
2
3
 
(Not yet reviewed)