摘要
目的 研制一种高效的计算机化智力测验方法 ,初步探讨其信度和效度。方法 把瑞文标准推理测验的 6 0个项目按照由易到难排序 ,定义每 5个相邻的项目为一个级别 ,分为 12个级别。测验时 ,软件根据被试的回答估计其智力水平 ,并据此呈现合适的测验项目。用 179名潜水人员的测验结果进行 5次模拟测验 ,考察计算机化测验方法的效率及其信度和效度。结果 5次模拟测验平均仅用 2 5 %的测验项目 ,测验结果之间的相关系数大于 0 .72 (P =0 .0 0 0 1) ,与原测验的相关系数大于 0 .79(P =0 .0 0 0 1)。结论 初步认为计算机化智力测验的效率及信度和效度较高 。
Objectives To develop a computerized intelligence test with higher efficiency,and to study its reliability and validity.Methods 60 items of Raven's Standard Progressive Matrices (SPM) were sorted from easy to difficult,each 5 items being defined as a rank and 12 ranks in all.The intelligence rank of a test was estimated by the system based on his response,the following items was selected and presented by the system according to the expected intelligence rank.In order to test its efficiency as well as reliability and validity, 5 simulation tests were conducted using the test results of 179 divers.Results The computerized intelligence test only used 25% items of the original test,the correlation coefficients among the ranks of 5 simulation tests were greater than 0.72(P= 0.0001) and the correlation coefficients between the ranks of 5 simulation tests and the original test were greater than 0.79(P= 0.0001).Conclusion The preliminary conclusion is that the computerized intelligence test have high efficiency,reliability and validity,and can be used in practice.
出处
《中国行为医学科学》
CSCD
2002年第1期100-101,共2页
Chinese Journal of Behavioral Medical Science
