计算机自适应测验开发的程序研究

摘要

计算机自适应测验（Computerized Adaptive Testing,CAT），是基于项目反应理论（IRT）建立题库，由计算机根据被试能力自动选择试题，最终对被试能力做出估计的一种新的测验形式。CAT因人选题，项目针对性强，用较少的题目较精确地估计被试的能力。CAT虽好，但不能盲目进行。从构思到使用，一个完整的CAT一般需要经历四个阶段：CAT可行性、适用性的考察研究，建设题库，确定CAT的算法及参数，投入使用和维护。每一阶段对CAT的有效性都至关重要。%Computerized Adaptive Testing （CAT） is a new method that item bank is based on response theory （IRT）, computer automatically selects items according to the ability of each examinee, and ultimately estimate precise ability of each examinee . The option of CAT＇ s item differs from person to person with a highly targeted, which estimates the ability of examinees by fewer subjects with a good accuracy. CAT is good, but can＇ t blindly used. From design to publish, generally go through four stages： feasibility and applicability studies, item bank development, CAT algorithm and parameters, publish and maintenance. The effectiveness of each stage of the CAT is essential.