MEASUREMENT ERRORS IN THE BEHAVIORAL-SCIENCES - THE CASE OF PERSONALITY ATTITUDE RESEARCH

被引：102

作者：

BUCKLEY, MR ^{[1
]}

COTE, JA ^{[1
]}

COMSTOCK, SM ^{[1
]}

机构：

[1] WASHINGTON STATE UNIV,PULLMAN,WA 99164

来源：

EDUCATIONAL AND PSYCHOLOGICAL MEASUREMENT | 1990年 / 50卷 / 03期

关键词：

D O I：

10.1177/0013164490503001

中图分类号：

G44 [教育心理学];

学科分类号：

0402 ; 040202 ;

摘要：

It has been established that construct measures often contain measurement error and method effects. However, the extent of these measurement problems and their effects on personality constructs has not been fully examined. In this study, an analysis of published data sets affords estimates of the amount of trait, method, and error variance of measures used in the social/behavioral sciences. It was found that traits accounted for less than 50 percent of the variance in construct measures among the entire group of data sets included in the study. Measures of attitudes and personality variables contain considerably less trait variance. Implications for theory testing are discussed. © 1990, Sage Publications. All rights reserved.

引用

页码：447 / 474

页数：28

共 79 条

[61] MULTITRAIT-MULTIMETHOD VALIDATION OF MEASURES OF STUDENT ATTITUDES TOWARD SCHOOL, TOWARD LEARNING, AND TOWARD TECHNOLOGY IN SIXTH GRADE CHILDREN [J].