Parscale 3.0
Published on 10. July 2007
Software
CD-ROM
978-1-56321-127-0 (ISBN)
Description
The PARSCALE program extends item response theoretic (IRT) measurement methods to multiple-category rating scale items. It provides efficient scaling of data obtained in:
* Surveys of social attitudes and opinions; * Psychological evaluations based on personality inventories and self reports; * Ratings of the clinical status of patients in therapy; * Individually-administered cognitive tests in which the interviewer rates the quality of each response; and * Educational achievement testing and assessment based on ratings of constructed-response exercises, including essay questions, open-ended problem solving, and performance tasks
In educational testing applications, the program can analyze tests consisting of both rating scale items and multiple choice items subject to guessing effects. It combines the information from these types of items optimally into scale scores for the examinees. Other benefits include:
* The choice of the two most widely accepted models for rating scale data -- Samejima's model or the Masters-Andrich partial credit model.
* Parameter estimation for all forms of these models performed by the consistent and efficient marginal likelihood method, including users' options for the type of population distribution and possibility of stochastic constraints on the item parameters.
* Scale scores for respondents or examinees estimated by Bayes or maximum likelihood methods, along with the corresponding posterior standard deviations or standard errors; multiple subtests analyzed simultaneously and scale scores estimated for each.
* Surveys of social attitudes and opinions; * Psychological evaluations based on personality inventories and self reports; * Ratings of the clinical status of patients in therapy; * Individually-administered cognitive tests in which the interviewer rates the quality of each response; and * Educational achievement testing and assessment based on ratings of constructed-response exercises, including essay questions, open-ended problem solving, and performance tasks
In educational testing applications, the program can analyze tests consisting of both rating scale items and multiple choice items subject to guessing effects. It combines the information from these types of items optimally into scale scores for the examinees. Other benefits include:
* The choice of the two most widely accepted models for rating scale data -- Samejima's model or the Masters-Andrich partial credit model.
* Parameter estimation for all forms of these models performed by the consistent and efficient marginal likelihood method, including users' options for the type of population distribution and possibility of stochastic constraints on the item parameters.
* Scale scores for respondents or examinees estimated by Bayes or maximum likelihood methods, along with the corresponding posterior standard deviations or standard errors; multiple subtests analyzed simultaneously and scale scores estimated for each.