Rasch measurement model or one parameter model is the simplest IRT model and it has strong
measurement properties. The probability of people getting the same item correctly resort to two parameters in Rasch model, which are item difficulty and person ability . According to Wright (1977), there are numerous benefits using Rasch model in the test measurement. Firstly, Rasch model can evaluate whether the item is fit and identify if item bias exists. Secondly, its item calibration is not influenced by the ability of sample, which means it is sample free. Thirdly, standard error of calibration can be exploited to examine the precision of each item. Fourthly, Rasch model can estimate the item difficulties from various samples and convert them into a common scale. Hence, the item banks can be equated automatically as a common calibration to be shared by all items. Fifthly, the ability of two people can be compared although they do not have any item in common by transforming the ability estimates into a common scale. This is called test-free person measurement. Sixthly, Chi-square of person fit can be utilized to assess measurement quality. Lastly, by using Rasch model, it makes the construction and design of the best test as well as tailored testing easier to be governed.
Rasch measurement model or one parameter model is the simplest IRT model and it has strongmeasurement properties. The probability of people getting the same item correctly resort to two parameters in Rasch model, which are item difficulty and person ability . According to Wright (1977), there are numerous benefits using Rasch model in the test measurement. Firstly, Rasch model can evaluate whether the item is fit and identify if item bias exists. Secondly, its item calibration is not influenced by the ability of sample, which means it is sample free. Thirdly, standard error of calibration can be exploited to examine the precision of each item. Fourthly, Rasch model can estimate the item difficulties from various samples and convert them into a common scale. Hence, the item banks can be equated automatically as a common calibration to be shared by all items. Fifthly, the ability of two people can be compared although they do not have any item in common by transforming the ability estimates into a common scale. This is called test-free person measurement. Sixthly, Chi-square of person fit can be utilized to assess measurement quality. Lastly, by using Rasch model, it makes the construction and design of the best test as well as tailored testing easier to be governed.
การแปล กรุณารอสักครู่..
