2001 Fiscal Year Final Research Report Summary
Research on analysis for quality information by new statistical approach
| Project/Area Number |
11680460
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (C)
|
| Allocation Type | Single-year Grants |
|---|
| Section | 一般 |
|---|
| Research Field |
社会システム工学
|
|---|
| Research Institution | Osaka Electro-Communication University |
|---|
Principal Investigator |
IHARA Masamori Dept. of Information Scinence and Arts, Osaka Electro-Communication University, Prof., 総合情報学部, 教授 (40184791)
|
|---|
| Project Period (FY) |
1999 – 2001
|
| Keywords | Post-stratification / Linear Structural Equation Modeling / Factor analysis model / Identifiabkity |
|---|
| Research Abstract |
We were able to clarify causes where some subpopulation occurred from a population., Meredith(1964) shows results of research in The Psyckometrika, in the "selection" case and which Little (1972, Biometrika) shows some basic resuluts by using ideas of missing data and Dempster et.al (1984, JRSS) also gives EM algorithm for the missing data
In this research, in a Linear Structure Equation Model including a factor analysis or a covariance structure analysis model, We did a study about extraction of quality information by classifying by stratum after a thing
We already announced one part of results of research in No. 13 and No. 14 Asia quality control symposium and announced it in last year held international psychometrics party (2001 IMP S)
We presented some conditions that a factor analysis model could hold in a subpopulation drawn by post-stratification if a factor analysis model held in whole population when we classified population by a value of one part of observation vector according to approach of Muthen et al (1989:Psychometrika) and Mereidth (1964). Moreover, if a suitable condition was concluded, the factor analysis model clarified that it was not it with a model with an impropriety solution
Furthermore, we applied this way of thinking to a linear structure equation model and suggested experimental planning to do optimum design of a product package and a method of analysis that fused (now I am contributing an article.)
Now we study heterogeneity of approach by, missing data and aproach by selection and a homogeneity
|
