Please use this identifier to cite or link to this item:
Title: Applying Item Response Theory to Science Educational Measurement:Galt and Tips
Other Titles: 統整科學過程技能與邏輯推理能力測驗試題潛在特質分析研究
Authors: 許榮富
Issue Date: Jun-1991
Publisher: 國立臺灣師範大學研究發展處
Office of Research and Development
Abstract: 本文探討一參數試題反應模型(即拉覯句模型,Rasch Model,RM)的基本假設及其估計理論,並實際應用於科學教育測驗的分析。單維性測驗、局部推論獨立性與邏輯型式(logistic form)的試題反應函數為RM的三個基本假設,據此可得與樣本無關、與試題無關的測量,此特性稱為「特定客觀性(specific objectity)」的測量。於本文中,嘗試賦予單維性假設經驗意義。RM的參數估計可有幾種途徑,本文討論非條件與條件最大概度(Unconditional/Conditional Maximum Likelihood)估計的優缺點。除了RM之外,也討論單維的線性邏輯特質模型與多維的成份潛在特質模型等兩個RM的類化模型在科學教育測量上的可應用性。於此研究中,實際以RM對兩種測驗(TIPS (II)與GALT,前者為統整科學過程技能測驗,後者為邏輯推理能力測驗)進行試題分析,結果顯示理論預測值與觀察值確受試題潛在特質影響,這些影響均以試題特徵曲線分析顯現之。另一方面,結果亦顯示RM在試題特性的偵測上,深具極高鑑別精緻結構及試題偏失的優越性。
This study developed the techniques for representing latent traits in Science Process Skillsand Reasoning Ability measurement based on item response theory. The three basic assumptions and specific objectivity in Rasch Model were examined inview of science education. Both linear logistic trait models and component latent trait modelsfor accounting the nature of the abilities measured in test situations were also proposed. Results of the study revealed that the predicted-observed patterns were affected by latenttraits underlying each item and the stems for each item. The analysis might demonstrate thehigh sensitivity and resolution in detecting bias of test items of Rasch model in this study.
Other Identifiers: 6FF4E29B-5BC1-EEC9-3740-5C8833997F5D
Appears in Collections:師大學報

Files in This Item:
File SizeFormat 
ntnulib_ja_L0801_0036_191.pdf500.89 kBAdobe PDFView/Open

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.