|
Activity Number:
|
209
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Monday, July 30, 2007 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Section on Survey Research Methods
|
| Abstract - #309534 |
|
Title:
|
Standard Error Estimation in Latent Variable Model
|
|
Author(s):
|
Yan Wang*+ and S. Lynne Stokes and Andreas Oranje
|
|
Companies:
|
Southern Methodist University and Southern Methodist University and Educational Testing Service
|
|
Address:
|
Dept. of Statistical Science, 144 Heroy Science Hall, Dallas, TX, 75275-0332,
|
|
Keywords:
|
Latent Variable Model ; Standard Error ; NAEP
|
|
Abstract:
|
In the large-scale educational assessment, the observations are selected from population by the complex sampling design, and the quantity of interest, such as student ability, is not observed directly. The data selection method causes clustering in the observations which are often positively correlated. The manifest variables of the student ability in the educational measurement are commonly the student's responses to a group of questions. The questions are often related to the same stimulus which causes positive correlations among the student's responses. Ignoring the positive correlation and making independence assumption in the standard statistical analysis underestimate the standard errors of the estimates. The new models relax two sources of the independences mentioned above and are used to obtain the proper standard errors. We investigate the standard error by simulation studies.
|
