|
Activity Number:
|
12
|
|
Type:
|
Topic Contributed
|
|
Date/Time:
|
Sunday, August 6, 2006 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Social Statistics Section
|
| Abstract - #307306 |
|
Title:
|
Improve Variance Estimation for the Assessments Based on the Plausible Values Approach
|
|
Author(s):
|
Jiahe Qian*+ and Shelby Haberman
|
|
Companies:
|
Educational Testing Service and Educational Testing Service
|
|
Address:
|
Rosedale Road, T198, Princeton, NJ, 08541,
|
|
Keywords:
|
jackknifed variance estimation ; assessments ; NAEP ; conditioning models
|
|
Abstract:
|
The jackknife procedure is used to estimate the variance of plausible values generated by the conditioning process for assessments such as NAEP. The plausible values are generated from a posterior distribution based on conditioning on test scores and demographic and education-associated variables. Then, the plausible values are transformed to reporting scale. Currently, the jackknifed estimation is based on one set of plausible values. This study will estimate jackknifed variance from all sets of the plausible values generated from jackknife replicate samples by conditioning models. Although the process is computationally intensive, it is no longer an obstacle with the growth of modern computing capabilities. Data from the NAEP assessments will be used to compare new results with the old and conduct variance analysis.
|
