廣義隱藏式馬可夫模型應用於時間序列題組型之二元計分測驗程式設計與應用

• Main Authors: Lin, Kuei-kuang, 林奎光
• Other Authors: 劉湘川
• Format: Others
• Language: zh-TW
• Published: 2007
• Online Access: http://ndltd.ncl.edu.tw/handle/02997437670498472024

碩士 === 國立臺中教育大學 === 教育測驗統計研究所 === 95 === The main purpose of this study aims to combine an integrated model of Generalized Hidden Markov Model（GHMM）and Kernel Smoothing Non parametric IRT（KN-IRT）with Item Relational Structure（IRS）for researching and developing of application programs as well as processing an empirical study by testlet based on every unit of mathematics for sixth grade in elementary school. The program can estimate experimenter’s ability, guessing degree and unreached parameters in response to the testlet effectively without matlab program and analyze the correlation between testlet and experimenter. The integrated model of Generalized Hidden Markov Model（GHMM）and Kernel Smoothing Non parametric IRT（KN-IRT）sifts the wheat from the chaff and combines characteristics of arametric with non-parametric items. There are three major advantages concurrently as follows: 1. The model is not restricted to exceed 200 above for the experimenters. It applies to large-scale criterion tests for more experimenters and self-edit tests for less ones but with reliabilities and validities. 2. The model can not only estimate experimenter's ability and item’s characteristic curve, but also can analyze and separate guessing parts from right response to the items and unreached parts from wrong responses to the items. 3. The model is not restricted to the local independence of the items. It can be futher applied to integrate with Item Ordering Theory (IOT) or Item Relational Structure (IRS).