Design and Evaluation of an XMLBased PlatformIndependent Computerized Adaptive Testing System - PowerPoint PPT Presentation

1 / 20
About This Presentation
Title:

Design and Evaluation of an XMLBased PlatformIndependent Computerized Adaptive Testing System

Description:

... factor: 0.526. Reporter: Tsung-Yao Lee (???) Date: 2005/9/10 ... Precision and Efficiency. CAT platform (PC or PDA) does not cause significant difference on ... – PowerPoint PPT presentation

Number of Views:37
Avg rating:3.0/5.0
Slides: 21
Provided by: cda64
Category:

less

Transcript and Presenter's Notes

Title: Design and Evaluation of an XMLBased PlatformIndependent Computerized Adaptive Testing System


1
Design and Evaluation of an XML-Based
Platform-Independent Computerized Adaptive
Testing System
  • Author Rong-Guey Ho and Yung-Chin Yen
  • Source IEEE Transaction on Education, VOL. 48,
    NO. 2, May 2005
  • SCI 10/20 50 Impact factor 0.526
  • Reporter Tsung-Yao Lee (???)
  • Date 2005/9/10

2
Outline
  • Introduction
  • IRT-based CAT
  • XML-based CAT system
  • Experiment
  • Result
  • Conclusion
  • Comment

3
Introduction
  • CAT can be tailored to the ability of each
    examinee
  • Based on item response theory (IRT)
  • Benefit
  • Improve the accuracy of testing
  • Save testing time
  • Require fewer items per examinee
  • Using XML to describe the item bank, one is able
    to solve the problem while implementing a CAT
    system on a different platform (PC, PDA, Handheld
    device)

4
Item Response Theory (IRT)
ai item discrimination bi item difficulty ci
pseuso-guessing
ai 2.0, bi 0.0, ci 0.25 ? -3.0 to 3.0
5
Adaptive strategy
  • 1) An initial ability is estimated in some way
  • 2) The initial ability estimate is used to select
    an appropriate item from the item bank.
  • 3) The item is scored as either correct or
    incorrect, and the examinees ability level is
    reestimated
  • 4) If the estimate is sufficiently precise or a
    prespecified criterion has been reached, the
    procedure is terminated. Otherwise, the estimate
    is further refined by the return to Step 2

6
XML-Based CAT System
7
(No Transcript)
8
Workflow 1/3
  • Step 1) The item parameters of each item in the
    item bank will be sent from the item retrieval
    subsystem to the CAT subsystem as soon as the
    system is activated.
  • Step 2) The system first detects the type of
    platform
  • Step 3) Assuming the examinee is of median
    ability, the system will pick an initial item of
    medium difficulty and send its item number to the
    item retrieval subsystem.

9
Workflow 2/3
  • Step 4) CAT subsystem extracts the XML-format
    item from the bank and sends it to the XSL
    implementation procedure.
  • Step 5) XSL implementation procedure generates
    the HTML-format item by applying the
    corresponding XSL template
  • Step 6) displays HTML-format item on the screen

10
Workflow 3/3
  • Step 7) After the answer has sent, the system
    will check the answer and reevaluate current
    ability
  • Step 8) The system will then check whether the
    stopping condition is met. If the stopping
    condition is not met, the next step will return
    to Step 3).
  • Step 9) If the stopping criterion is met, the
    system shows his or her final estimation of
    ability on the screen and sends the response data
    to response database for further analysis.

11
Experiment design
  • Participants
  • Fifty senior high school students, 95.92 had at
    least one year of computer experience.
  • Only 28 had ever taken tests on the computer,
    and only 22 of them had experience with a PDA.
  • Item bank
  • 84 items from the vocabulary sections of CEEC
  • The value of IRT parameters of each item was
    calibrated using the BILOG program following a
    3-p logistic model based on 5000 high school
    students
  • Properties in item bank

12
Procedure
  • Group A (n25) PC-version CAT taken first
    PDA-version CAT taken second.
  • Group B (n25) PDA-version CAT taken first
    PC-version CAT CAT taken second.
  • PC-CAT was administered on Intel Pentium III PC
    with 128M RAM
  • PDA-CAT was on HP Jornada 565 and COMPAQ
    iPAQH3850.

13
Precision and Efficiency
Ability
Time
Item no.
  • CAT platform (PC or PDA) does not cause
    significant difference on
  • accuracy (ability)
  • efficiency (time, item no.)

14
Participants Response Toward Testing Environment
15
Conclusion
  • CAT on PDA is practical
  • The results may differ when experiment is of a
    different age or other domains
  • Probably is only a result of the examinees
    curiosity regarding PDA

16
Future work
  • Revising this system to a platform-independent,
    on-line practice system
  • Implementing this system on other domains
  • Expanding the diversity of the CAT administration
    platform

17
Comment
  • ??????????,??XML????????????
  • ???????????,??????SCORM????
  • ???XML???????,??????SCORM????

18
IRT
  • IRT?Load?1980???????????(item response
    theory,??IRT),????????????????,????????,??????????
    ?????????
  • 1.???(Difficulty index)???????????????????????,?
    ???????
  • 2.???(Pseudoguessing parameter)?????????????????
    ,??????????????
  • 3.???(item-descriminating power)?????????,??????
    ?????????????

19
(No Transcript)
20
(No Transcript)
Write a Comment
User Comments (0)
About PowerShow.com