|
KMID : 0371019990320010080
|
|
Journal of Preventive Medicine and Public Health
1999 Volume.32 No. 1 p.80 ~ p.87
|
|
|
A study ou the Statistical Distribution and Testing of Variation Indicies at the Small Area Variation Analysis
|
|
Nam Chung-Mo
Lee Sun-Hee
Cho Woo-Hyun
|
|
|
Abstract
|
|
|
Objectives:The Study of Small Area Variation(SAV) is most interesting issue in the health care researches. Most studies of SAV have been concluded the existences of variation on the basis of the magnitude of variation on the basis of the magnitude of variation without statistical testing. But it is difficult to explain the existence of variation with this way because variation indices are easily influenced by several parameters and also their distribution are skewed. So, it needs for the study to investigate the distribution of these indices and develop the statistical testing model..
Methods:This study was planned to analyze on the distribution of variation indices such as External Quotient(EQ), Coefficient of Variation(CV), Systematic Component of Variation(SCV) and compare the statistical power among indices. The simulations was performed on the basis of several assumptions and compared to the empirical data.
Results:Main findings can be summarized as follows.
1. If other conditions are constant, the more number of regions, the larger 95 percentile of EQ. But under same situation, 95 percentile of CV and SCV were slightly decrease.
2. If the size of regional population or utilization rate were increase, 95 percentile of all statistics were decreased. Also in the cases of small population size and low utilization rate, 95 percentiles of EQ showed various change contrast to the little change of CV.
3. If the difference at the size of regional population were increased, 95 percentiles of EQ and SCV were increased contrast to the little difference of CV.
4. If the utilization rate were increased, 95 percentiles of all indices were increased. But under the same difference of utilization rate, the power of CV and SCV were increased comparing to no change of the power of EQ.
5. Usually the power of EQ were lowe than that of CV or SCv and it is similar between CV and SCV.
Conclusions:Therefore, we suggest that in selecting the variation indices at the SAVm CV or SCV are superior than EQ in terms of significance level and power.
|
|
|
KEYWORD
|
|
|
small area variation, EQ, CV, SCV, statistical testing
|
|
|
FullTexts / Linksout information
|
|
 
|
|
|
Listed journal information
|
|
  
|
|