Open AccessVALIDATION BY THE MULTIGROUP‐MULTISCALE MATRIX: AN ADAPTATION OF CAMPBELL AND FISKE'S CONVERGENT AND DISCRIMINANT VALIDATIONAL PROCEDURE
Author(s) -
Centra John A.
Publication year - 1969
Publication title -
ets research bulletin series
Language(s) - English
Resource type - Journals
eISSN - 2333-8504
pISSN - 0424-6144
DOI - 10.1002/j.2333-8504.1969.tb00763.x
Subject(s) - psychology , discriminant validity , adaptation (eye) , perception , scale (ratio) , matrix (chemical analysis) , set (abstract data type) , convergent validity , cognitive psychology , trait , similarity (geometry) , discriminant , social psychology , psychometrics , artificial intelligence , computer science , developmental psychology , physics , materials science , image (mathematics) , quantum mechanics , neuroscience , internal consistency , composite material , programming language
Campbell and Fiske advocate an approach to investigating validity that employs a matrix of intercorrelations among tests representing more than one trait, each measured by more than one method. This paper presents an application of the Campbell‐Fiske multitrait‐multimethod procedure. Referred to as validation by the multigroup‐multiscale matrix, this adaptation compares discrete groups of individuals and the scale scores that represent their responses to a multidimensional perceptual space. As used here, therefore, intergroup agreement on a set of perceptual scales is being assessed, and validity may be defined as the degree of similarity in responses by different constituent groups. The application is illustrated by analyzing group perceptions of their college environment as indicated by the Institutional Functioning Inventory . Multidimensional scales assessing the viewpoints of faculty, administrator, and student groups at 17 colleges and universities comprised the multigroup‐multiscale matrix. Lack of agreement between the groups indicated scales that more likely reflect subjective interpretations of the environment rather than relatively objective environmental constructs. Convergent and discriminant information through the multigroup‐multiscale matrix, while not confirming the theory underlying the measurement, does nevertheless appear to provide useful insights into how a current instrument is operating and how it might be improved.