A Hybrid Approach to Moderation of School-based Assessment in Advanced Supplementary Level Liberal Studies – Using Expert Judgement and Bayesian Hierarchical Statistical Modelling to Enhance Reliability and Comparability

A Hybrid Approach to Moderation of School-based Assessment in Advanced Supplementary Level Liberal Studies – Using Expert Judgement and Bayesian Hierarchical Statistical Modelling to Enhance Reliability and Comparability

  • Version 1.0.0
  • Download 9
  • File Size 80.36 KB
  • File Count 1
  • Create Date August 2, 2018
  • Last Updated August 2, 2018

A Hybrid Approach to Moderation of School-based Assessment in Advanced Supplementary Level Liberal Studies – Using Expert Judgement and Bayesian Hierarchical Statistical Modelling to Enhance Reliability and Comparability

In the Hong Kong Advanced Level Examination, Liberal Studies (LS) requires each student to complete an individual research project. As a mode of school-based assessment (SBA), theprojects are marked by their teachers and counted as part of the public examination. However, ,teachers are not necessarily aware of the standards of performance across all schools. To achieve comparability of assessments across schools, the project marks awarded by teachers will be moderated statistically. In 2010 LS exam, moderation will be conducted on school basis. In principle, 5 projects will be selected from each school from different levels of performance in order to obtain a ,representative sample. Each sampled project will be double marked by two external assessors.Based on external assessors’, marks, the school performance level on SBA and the corresponding variability are estimated. However, the reliability of these statistics may be cast in doubt in view of small sample size. With respect to this problem, the use of Bayesian hierarchical statistical modeling is proposed so as to share information across different schools in order to increase the reliability of statistics concerned. Empirical study shows that the approach is promising in stabilizing the estimations and preventing excessive changes to ,teachers’, marks.

Attached Files

FileAction
paper_2fb222747.pdfDownload 
Menu
X