计算机化分类测验的特点与发展述评

The Comments and Review of Variable-length Computerized Classification Testing

计算机化分类测验(VL-CCT)的基本原理是在CAT形式下按照某一能力估计值分界点对被试进行分类。VL-CCT能用较少的试题实现准确分类,分类准确性要优于CAT。本文概述VL-CCT的几个技术组成部分及发展趋势,主要论述特别适用于VL-CCT的两种选题策略、VL-CCT选题策略的两种应用模式、四种终止规则,以及被试分类准确性、测验效率方面的评价指标。在VL-CCT下进行选题策略比较研究、终止规则的比较研究、VL-CCT在多维能力测量、人格测量、学校教育测验评估方面的应用等是近年来的主要研究方向。

Variable-length Computerized Classification Testing(VL-CCT)is a modification of computerized adaptive testing(CAT)with the goal of classifying examinees into pre-specified categories.Variable-length computerized classification testing is widely used in educational test,occupation qualification tests,measurement of personality test,and cognitive diagnosis tests.CAT researches in the literatures could be divided into four kinds.Computer adaptive classification test is essentially a standard-reference test.In this paper,the authors summarize two categories of VL-CCT research paradigm,summed up the VL-CCT simulation method and it comprises of seven components,including Psychometric model,item pool,starting point,item selection strategies,termination criterion(classification/scoring procedure),ability estimations,etc.This paper also summarizes several evaluation angle and the evaluation indicators of VL-CCT simulation results.The measurement accuracy,security,test question library utilization rate,and test the validity of constraint object control are usual evaluation indicators.Some new developments and future research direction of VL-CCT simulation is also summarized.