The IRT Procedure
-
Overview
- Getting Started
-
Syntax
-
DetailsNotation for the Item Response Theory ModelAssumptionsPROC IRT Contrasted with Other SAS ProceduresResponse ModelsMarginal LikelihoodApproximating the Marginal LikelihoodMaximizing the Marginal LikelihoodFactor Score EstimationModel and Item FitItem and Test informationMissing ValuesOutput Data SetsODS Table NamesODS Graphics
-
Examples
- References
Assumptions
The primary statistical assumptions that underlie the analyses that PROC IRT performs are as follows:
-
The number of latent factors is known.
-
Latent factors are normally distributed.
-
Conditional on latent factors, observed responses (items) are independent. This is the so-called local independence assumption.
These assumptions are necessary if you want to make statistical inferences. In exploratory analysis, these assumptions do not apply.