| Summary: | 博士 === 國立臺南大學 === 教育學系測驗統計博士班 === 101 === The purpose of this study is to compare the differences of ability estimates in the
application of subtests, such as Bi-Factor Model (DeMars, 2006b) , Rasch subdimensions
model (Brandt, 2008), and Augmented scores method (Wainer & Thissen, 2001). Meanwhile,
it examines the accurancy of utilizing three models in scoring.
With the manipulation of numbers of dimensions, numbers of each dimensions’ items,
correlated between different dimensions’ abilities, and 1,000 examinee, simulated responses
were generated under different conditions. Aside from, under different simulated conditions,
the empirical data was obtained from the science assessment of BCTest 2009 which was
administered to 15-year-old students. The sample consisted of 5,000 students. The main
findings were the following: Firstly, in the simulated data, with the difference of simulated
conditions, such as dimensions, items, and correlations, the accurancy of different scoring
model would be affected. Secondly, in each simulated conditions, the estimated examinee’s
abilities of Bi-Factor Model was higher than the other two scoring models. Thirdly, the
empirical data showed that the results of the estimated examinee’s abilities with Bi-Factor
Model and Rasch subdimensions model were more similar. Finally, comparing with the
empirical data and one of the simulated data, it found that these two resule were similar.
|
|---|
