跳到主要內容區塊 :::
   
:::
首頁/研究主軸/績效責任/研究計畫/以自動化組卷法編製臺灣學生學習成就評量測驗之研究

以自動化組卷法編製臺灣學生學習成就評量測驗之研究

Automated test assembly for Taiwan assessment of student achievement
  • 資料類型

    研究計畫

  • 計畫編號

    NAER-100-24-B-2-02-00-2-02

  • GRB編號

    PG10201-0219

  • 計畫名稱

    以自動化組卷法編製臺灣學生學習成就評量測驗之研究 Automated test assembly for Taiwan assessment of student achievement

  • 計畫主持人

    陳姵樺

  • 共同主持人

    謝名娟

  • 經費來源

    國家教育研究院

  • 執行方式

    合作研究(本院經費-各方人員)

  • 執行機構

    國家教育研究院

  • 執行單位

    測驗及評量研究中心

  • 年度

    2011

  • 期程(起)

    2011-01-01

  • 期程(迄)

    2012-12-31

  • 執行狀態

    已結案

  • 關鍵詞

    自動組卷法,抽樣,最佳化,臺灣學生學習成就評量測驗

  • Keywords

    automated test assembly,sampling,optimization,TASA

  • 研究主軸

    中小學生成就長期追蹤

  •   臺灣學生學習成就評量測驗乃是評估臺灣學生從國小、國中乃至高中的學習發展狀況的一項重要參考指標。而此重要參考指標本身所編的測驗是否能保持一貫高品質,乃是一件十分重要的事。近年來關於「臺灣學生學習成就評量測驗」的編製,仍採用人工組卷的方式。本研究旨在發展並導入自動化選題法以編製「臺灣學生學習成就評量測驗」,期透過本研究,將在組卷程序中大量重覆性的工作由電腦運算取代,除了能確保品質、節省時間組卷外,更能將人力作更有效的運用。本研究初步分析了從「國家教育院」取得之TASA 題庫和相關參數,以及「2010年國二數學科之正式施測範本」。運用自動組卷目前兩大主流方法之一:最佳化方法組卷。首先,第一步持續與學者專家商討測驗組卷標準。第二步是組出符合與參考試卷訊息量相當之試卷。第三步乃是依照期中審查建議,採用模擬資料及數種不平衡區塊設計方式,提供TASA 未來組卷參考依據。

      根據上述內容,本研究最後利用「整數型規劃」的軟體CPLEX(IBM, 2009)組卷語法撰寫並建構一與模擬題庫之試卷平行之題本,完成「臺灣學生學習成就評量測驗」國中二年級數學科之自動組卷系統,並可依據最後組卷結果繪製相關結果圖表,包含:測驗訊息函數圖形(TINF)、測驗特徵曲線(TCC)。

      本研究計畫的研究成果預期可以達成的具體貢獻和效益如下:在學術研究的貢獻上,是將「整數型線性規劃」運用在平衡不完全區塊設計之自動化選題的設計上;在國家發展的貢獻上,因建立自動組卷運用於TASA 測驗,可以協助提供一兼顧測驗品質及效率的自動組卷工具,簡化現行組卷人力及時間。

      以對教育政策的立即可行性建議來說,本研究的自動組卷方法中加入了定錨試題,定錨試題的優點是使用者可在組題前指定試題,如此一來較能夠符合現行組卷的實務狀況。以對教育政策的中長期建議來說,本研究也根據不同規模的題庫做出模擬組卷,嘗試修改平衡不完全區塊設計之自動組卷方式中的總區塊數、區塊中題數和試卷的區塊數,提供未來TASA 組卷BIB 標準修正的方向的建議。
     

  •   Taiwan Assessment of Student Achievement(TASA)is an important index to evaluate the learning development of elementary school, middle school and high school students in Taiwan. Keeping a high quality of TASA test becomes an important issue. Recently, the construction of TASA test still uses manual assembly by manpower. The purpose of this study is to develop and implement an automated test assembly(ATA)procedure for TASA test and hope to replace the repetitive work by computer computation in test assembly. So far, we investigated the item pool parameters and test specifications based on the TASA mathematics exam of eighth grade in 2010 provided by National Academy for Educational Research. This study will continue using one of the two common approaches of ATA, optimization, to assemble TASA forms. First, we continued discussing the test assembly test specifications with content experts from now on. Second, current item pool data was used to build objective functions and constraints that satisfy the reference test of TASA. Third, we used simulated data and try different BIB designs based on the recommendations of midterm review to provide guidelines for future TASA BIB test assembly.

      According to the content above, this research used integer programming software CPLEX (IBM, 2009) to construct a form paralleled with test of simulation item pool, and completed a test assembly system for the eighth grade TASA test for mathematics subjects. Related graphics can be drawn based on the results of the final assembling forms, including: test information function (TINF), test characteristic curve (TCC).

      The result of this research expects to achieve the following contributions: In academic research, performing Integer Linear Programing in the design of BIB automated test assembly. In National development, constructing automated test assembly in TASA test can provide an automated test assembling tool with both test quality and efficiency. Besides, we can simplify the current assembling test process and assign human resources appropriately based on this research.

      Immediately feasible suggestions on education policy, our study of automated test assembly added anchor item into our test. The anchor item's advantage is that users can specify the items before test assembling. This anchor item design makes this system satisfy the practical test assembly requirement.

      To medium-and long-term recommendations for education policy, our study also conducted different simulation conditions for test assembly. We attempt to modify the total number of blocks、number of items per block and number of blocks per test in the BIB test assembly approach. The results provide useful suggestions for future TASA BIB test design.

top
回首頁 網站導覽 FAQ 意見信箱 EN