传统效度理论与统一构件效度的比较研究  

A Comparative Study of the Traditional Conception of Validity and the Unified Construct Validity

作　　者：张宝钧[1] 

机构地区：[1]北京语言大学外语学院

出　　处：《北京林业大学学报（社会科学版）》2003年第4期69-72,共4页Journal of Beijing Forestry University ：Social Sciences

基　　金：国家基金委资助项目

摘　　要：在语言测试理论中 ,效度一直被认为是测试本身的一个质量指标 ,效度验证主要考察内容效度、参照某一标准的共时效度或是构件效度中的一项。 2 0世纪 90年代 ,莫希克提出了统一的构件效度概念。它共有六个层面 ,不但包括了传统意义上的三种不同的效度 ,同时还包含了伦理价值、社会后果等游离于传统效度概念之外的内容。效度验证变成提出构件效度假设 ,然后收集六个层面的数据对假设进行论证的过程。该文通过对旧效度理论的批判和新效度理论的分析 ,发现传统的、只验证某一种效度的做法是片面的 ,也是不科学的。Traditionally,the various means of accumulating validity evidence are grouped into three separate and substitutable categoriescontent-related,criterion-related,and construct-related evidence of validity. Messick views validity as a unified concept with six distinguishable aspects,which are interdependent and complementary forms of validity evidence,conjointly functioning as general validity criteria or standards for all educational and psychological measurement. Comparing the traditional conception of validity with Messic's,we are able to find that the traditional conception of validity is fragmented and incomplete,because it fails to take into account both evidence of the value implications of score meaning as a basis for action and the social consequences of score use.

关 键 词：语言测试理论 传统效度理论 构件效度 比较研究 效度验证 

分 类 号：H0[语言文字—语言学]