The thresholds in this modeling translate to the Rasch difficulties (the "b" parameters in terms of the logistic IRT model with logit = a(factor - b)) as
b = threshold/loading.
Anonymous posted on Saturday, December 03, 2005 - 3:11 pm
Has the IRT model been extended to polytomous answer choices? That is, has anyone used this model to examine Likert type scales (i.e., strongly disagree, disagree, agree, strongly agree) types of data? Thank you.
bmuthen posted on Saturday, December 03, 2005 - 3:36 pm
Yes. Search for references containing "graded response".