|
Activity Number:
|
209
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Monday, July 30, 2007 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Social Statistics Section
|
| Abstract - #310389 |
|
Title:
|
Evaluating a Multilevel Item Response Theory Model
|
|
Author(s):
|
Thomas Schmitt*+ and Cindy M. Walker
|
|
Companies:
|
University of Wisconsin-Milwaukee and University of Wisconsin-Milwaukee
|
|
Address:
|
3215A S New York Ave, Milwaukee, WI, 53207,
|
|
Keywords:
|
item response theory ; multilevel models
|
|
Abstract:
|
Embedding item response theory (IRT) models within multilevel models tends to improve parameter estimates over an approach in which the latent trait is first estimated and then used as an outcome variable in a common linear model. Though extensions have been proposed, there exists a dearth of research systematically investigating multilevel IRT model's behavior under different conditions and when the model fails to provide robust parameter estimates. A simulation study will be conducted to evaluate parameter estimation error of a three-level IRT model under the following conditions: sample size at level-2 and level-3, the number of predictors at level-2, the number of items at level-1, unequal variances at level-3, and the shape of the latent trait distribution. The results will be used to provide researchers with a set of recommendations regarding the investigated conditions.
|
