skip to main content skip to footer

Constructs and Their Vicissitudes in Educational and Psychological Measurement

Author(s):
Messick, Samuel J.
Publication Year:
1979
Report Number:
RR-79-11
Source:
ETS Research Report
Document Type:
Report
Page Count:
40
Subject/Key Words:
Behavioral Science Research, Educational Research, Individual Differences, Models, Psychometrics, Test Validity

Abstract

This paper examines some implications for educational and psychological measurement of three existential perspectives on the nature of mediating variables underlying and relating consistencies in test and non-test behaviors. These three perspectives, respectively, treat mediators as (a) interven ing variables operationally tied to real causal entities or source variables, such as personality traits or environmental contingencies; (b) hypothetical constructs that organize and summarize behavioral consistencies but have no reality outside the theoretical system; and (c) manisfestations of real entities that are understood only in terms of constructs that summarize their empirical properties in relation to other constructs in a theoretical network. All three existential perspectives are applicable to personality traits or situational forces or their interactions, so implications for measurement hold whether for personal or environmental sources of behavioral consistency. In particular, the summary power of constructs in both theory and measurement that led to the predominance of construct validity principles in trait measurement implies that these principles should hold with equal cogency for situational and interactive measurement. Some slippery methodological problems arise in applying these principles, however, when clear distinctions are not articulated between the assessment model for measuring a construct and the substantive theory relating the construct to other constructs, each of which should have its own possibly different assessment model coordinated to its salient structural features. (40pp.)

Read More