ThurstonianThresh: Thurstonian thresholds

In CambridgeAssessmentResearch/unimirt: Applications of Unidimensional IRT using the R Package 'mirt'

Thurstonian thresholds

Description

The Thurstonian threshold for a score category is defined as the ability at which the probability of achieving that score or higher reaches the user-defined "prob" (0.5 by default). If prob=0.5, for items analysed using the graded response model or for dichotomous items analysed as part of a model using the (generalised) partial credit model (or Rasch) these will be equal to the usual difficulty parameters. However, Thurstonian thresholds may be a useful way of understanding the difficulty of marks within polytomous items under the (generalised) partial credit model. This function calculates these.

Usage

ThurstonianThresh(mirtobj, which.items = NULL, prob = 0.5)

Arguments

mirtobj	An estimated IRT model (of class SingleGroupClass) estimated using mirt or unimirt.
which.items	an integer vector indicating which items to include. By default all items are included.
prob	Probability of success for which thresholds are required.

Value

A data frame of Thurstonian thresholds.

Examples

## Not run: 
#model using "Science" data from mirt package
#for graded response model Thurstonian thresholds (for prob=0.5)
#and item difficulty parameters are the same
scimod=unimirt(Science[,1:4]-1)
unimirt::MirtTidyCoef(scimod)
ThurstonianThresh(scimod)

#will vary for other probs
ThurstonianThresh(scimod,prob=0.1)

#for gpcm (or Rasch) model they needn't be
mirtobj=unimirt(Science[,1:4]-1,"gpcmfixed")
unimirt::MirtTidyCoef(mirtobj)
ThurstonianThresh(mirtobj)

## End(Not run)

CambridgeAssessmentResearch/unimirt documentation built on Jan. 30, 2025, 11:05 p.m.