机构地区: 华南师范大学教育科学学院心理应用研究中心
出 处: 《考试研究》 2009年第3期71-78,共8页
摘 要: 为探讨全测验与锚测验不同的客观题与主观题分值比对等值误差造成的影响,本文设计两种全测验与锚测验题型分值比,以等值标准误为因变量,构建2×2的两因素完全随机化设计进行等值误差的方差分析。结果表明,全测验题型分值比与锚测验题型分值比两因素的主效应显著(p<0.001),交互作用显著(p<0.01),简单效应检验表明两因素在各水平上差异显著(p<0.01)。全测验题型分值比与锚测验题型分值比对等值误差产生一定的影响,在等值过程中应该考虑这两个影响因素,为了减小等值过程的误差,锚测验题型分值比应该尽量与全测验题型分值比相一致。 The paper aims to discuss the effect which objective and subjective score ratio of the entire and anchor test have on equating errors. By designing two different item-kind score ratios of the entire test and anchor test and seeing equating standard error as independent, an analysis of variance is done based on 2 ~ 2 two factors full random design. Main effects of two different item-kind score ratios of the entire test and anchor test are significant ( p 〈 0. 001 ) and their interaction effects also are significant( p 〈 0.01 ). Simple effect test finds that level effects on another factor are significant( p 〈 0.01 ). As for two different item-kind score ratios of the entire test and anchor test, they have some influence of equating errors and are worth considering. Item-kind score ratios of the entire test should be in accordance with item-kind score ratios of the anchor test.