OscatsModelHetlgr

OscatsModelHetlgr — Heterogeneous Logistic Graded Response Model

Synopsis

struct              OscatsModelHetlgr;

Object Hierarchy

  GObject
   +----OscatsModel
         +----OscatsModelHetlgr

Properties

  "Ncat"                     guint                 : Read / Write / Construct Only

Description

Details

struct OscatsModelHetlgr

struct OscatsModelHetlgr;

The Heterogeneous Logistic Graded Response IRT model is a generalization of the Homogeneous Logistic Graded Response model (OscatsModelGr) that allows the discrimination of each response category to be different: P(X=k|theta) = P*_k(theta) - P*_{k+1}(theta), P*_k(theta) = 1/1+exp[-(sum_i a_ki theta_i - b_k + sum_j d_j covariate_j)] where P*_k(theta) = P(X>=k|theta) is the probability of scoring at least k, P*_0(theta) = 1, P*_{Ncat} = 0, a_ki is the discrimination on dimension i for response k, and b_k is the scaled response boundary location. Responses are numbered 0, 1, ..., Ncat-1. Note that all respose options share the same parameters for the covariates. No distance metric is currently provided, but see Ali (2011) for possibilities.

Parameter names: Diff.k, Discr.k.DimName

References:

F. Samejima (1976). "The graded response model of latent trait theory and tailored testing." Proceedings of the First Conference on Computerized Adaptive Testing, 5-15.

F. Samejima (1969). "Estimation of latent ability using a response pattern of graded scores." Psychometrika Monograph Supplement, 17.

Usama Ali (2011). Item Selection Methods in Polytomous Computerized Adaptive Testing. Dissertation, University of Illinois.

Property Details

The "Ncat" property

  "Ncat"                     guint                 : Read / Write / Construct Only

The number of response categories in the IRT model, numered 0, 1, ..., Ncat-1. The default number of categories is 3, which is also the minimum.

Allowed values: >= 3

Default value: 3