Introduction to analysis of large scale assessment data using IRT and generalized latent variables
Pre-conference workshop with Matthias von Davier, Educational Testing Service, Princeton, USA
20th July, 14.30 - 18.30, Boys Smith Room, Fisher Building
This workshop will focus on large scale analysis using extensions of item response theory (IRT). The datasets used in the examples are from assessment designs with multiple test forms and matrix samples of student responses as found in survey assessments such as NAEP, PISA, TIMSS and PIRLS.
The workshop will give an overview of item response theory, including advantages and disadvantages of different models currently in operational use. Extensions of IRT used for linking proficiency distributions to background variables, extensions of current models to multidimensional approaches, and models for assessing differential item functioning or country-by-item interactions will be covered.
An overview of principles of population modeling and proficiency estimation using current operational models will be given. Central concepts will also be covered by hands-on exercises using software provided by the organizer for workshop participants. Hands-on exercises will demonstrate how to carry out item calibration for matrix samples of item responses as well as group-level estimation of proficiency distributions.