|
KMID : 0379919940190010080
|
|
Journal of The Korea Socity of Health Informatics and Statistics
1994 Volume.19 No. 1 p.80 ~ p.90
|
|
|
A Review Study on the Statistical Techniques for the Treatment of Nonresponse in Sample Surveys
|
|
|
|
|
Abstract
|
|
|
This paper described the decision making procedure for the nonresponse treatments by review of the literatures and compared the performance of three methods in estimating mean parameters for the situations that have varying rates of nonresponse, patterns of incomplete data, and degrees of correlational structure among the dependent and independent variables.
The adequte solution to the nonresponse problem requires that a set of decisions be made. The crucial issue of nonresponse in sample surveys was the nature of the underlying mechanisms which cause certain values to be missing. The results of comparison were that under the assumption of MCAR the ML methods, Complete cases method and Available cases method performed equally when the nonresponse rate was 10 percent. With the exception of this case, ML methods was generally appropriate. Moreover, in the case of not MAR all methods yielded inconsistent estimates, but ML me hods reduced the size of the bias.
|
|
|
KEYWORD
|
|
|
|
|
|
FullTexts / Linksout information
|
|
|
|
|
|
Listed journal information
|
|
|
|