National Changhua University of Education Institutional Repository : Item 987654321/15195
English  |  正體中文  |  简体中文  |  Items with full text/Total items : 6507/11669
Visitors : 29732294      Online Users : 453
RC Version 3.2 © Powered By DSPACE, MIT. Enhanced by NTU Library IR team.
Scope Adv. Search
LoginUploadHelpAboutAdminister
NCUEIR > College of Science > math > NSC Projects >  Item 987654321/15195

Please use this identifier to cite or link to this item: http://ir.ncue.edu.tw/ir/handle/987654321/15195

Title: 傳統檢定試題偏差(DIF)方法的改良與分析
A Smoothing Approach for Detecting Differential Item Functioning (DIF)---Refinements of IRT Area Method
Authors: 李信宏
Contributors: 數學系
Keywords: 試題偏差;差異試題功能;試題反應理論;試題特徵曲線;核平滑
Item bias;Differential item functioning (DIF);Item response theory (IRT);Item characteristic curve (ICC);Kernel smoothing
Date: 2000
Issue Date: 2013-01-07T09:03:55Z
Publisher: 行政院國家科學委員會
Abstract: 試題偏差是指試題本身包含了某些對於特定群體的受試者(例如:男性或者女性,原住民與非原住民)有利的因素,使得這些特定群體有較顯著的優勢來答對此種試題。在試題反應理論的架構之下,試題偏差可以用兩組不相等的試題特徵曲線來表示,所以傳統上,這兩組試題特徵曲線之間的差異或者是面積大小,便可以當作是檢定試題偏差的一種指標。然而,這種指標一方面受制於試題反應理論模式的選擇,另一方面缺乏適當的統計檢定理論為後盾,所以並非適用。我們所提出的計畫,是利用kernel smoothing 方法來估計這兩組試題特徵曲線,以避免模式選擇和適用的問題; 然後藉由隨機化檢定(randomization test) 來求導試題特徵曲線之間面積的經驗分配,並得到適當的統計量來檢定試題偏差的存在與否。計畫中,我們並將利用模擬研究來瞭解此新方法的性質,這包括了型一錯誤和檢定力的研究,我們預期能夠改善傳統面積檢查法在檢定試題偏差的缺失。
Item bias or differential item functioning(DIF) means that for two or more individuals, each with the same latent proficiency or latent ability level, the probability of answering a item correctly is not equal from individual to individual or group to group. In item response theory (IRT) terms, DIF can be represented by two non-identical item characteristic curves (ICC). Therefore, the difference or area between the two ICC’s is often used as convenient measures of DIF. However, these measures are descriptive and the importance of significance tests for these
measures is clear. The proposed procedure first utilizes a kernel smoothing technique to estimate the ICC’s. A hypothesis-testing
statistic of the area between two ICC’s then is derived through randomization test method. To investigate the performance of our procedure, a large-scale simulation study will be carried out. The simulation includes Type I error study as well as power study. Also, the comparisons of the behavior of our procedure with other DIF detection procedures are analyzed and carefully discussed. Our new approach is expected to be a good refinement of the IRT-area-based methods for assessing DIF.
Relation: 國科會計畫, 計畫編號: NSC89-2118-M018-003; 研究期間: 8808-8907
Appears in Collections:[math] NSC Projects

Files in This Item:

File SizeFormat
2020101612002.pdf33KbAdobe PDF511View/Open


All items in NCUEIR are protected by copyright, with all rights reserved.

 


DSpace Software Copyright © 2002-2004  MIT &  Hewlett-Packard  /   Enhanced by   NTU Library IR team Copyright ©   - Feedback