Academic journal article Cognitie, Creier, Comportament

Inquiry into a Domain of Worry Cognitions in Test Anxiety Research

Academic journal article Cognitie, Creier, Comportament

Inquiry into a Domain of Worry Cognitions in Test Anxiety Research

Article excerpt


This study reports consistent findings from a series of construct domain studies of self-related worry cognitions in terms of failure outcome expectancies in 18 samples of adolescent girls and boys from different ethnic groups in five countries on three continents. A multifacet measurement design was applied to inquire into the domain by means of a combination of generalizability theory and confirmatory factor modelling. The series of results provided consistent support for the distinction between self-referenced and other-referenced worry cognitions. Among other factors the students were shown to be more differentiated by other-referenced cognitions than self-referenced cognitions. The findings may suggest a reconceptualization of the domain of worry cognitions.

KEYWORDS: multifacet measurement design, domain of worry cognitions, self- and other-referenced worry cognitions.

(ProQuest: ... denotes formulae omitted.)

Measurement studies of the test anxiety construct have been mostly concerned with assessing its dimensionality and stability over time. Less interest has been taken in how far we can generalize test scores from test anxiety scales. Ideally, these three issues assume that a domain of test anxiety indicators has been defined before any meaningful empirical assessment can be made. In particular, without any delineation of the construct domain, no meaningful generalization can be made to the domain. Even though it is well accepted that "The very definition of a construct implies a domain of content" (Pedhazur & Schmelkin, 1991, p. 179), the description of a domain of content in test anxiety measurement has not gained much of attention. The typical procedure, however, for creating scales in test anxiety research has rather been to administer a large pool of items mostly derived from prior test anxiety scales. The final criterion for including or excluding items was exploratory factor analysis. Thus the latent factors were identified by factor analysis. A potential result is that the final dimensions may be related to the latent constructs somewhat loosely and post hoc. In some cases a large number of nonduplicable items that belonged to the original pool of construct relevant indicators, have been excluded by the researcher due to the data analytic model that was imposed on the data. It is ample reason to believe, however, that all the items constituting the original pool represent the researcher's conception of test anxiety. In this sense the item indicators presumptively possess content validity (Nunnally & Bernstein, 1994) as indicators of test anxiety. An interesting question remains: what information about test anxiety has been left out during the process of fitting the factor model to the data? Could one or more aspects of test anxiety have been misrepresented or underrepresented (Messick, 1995) in the final set of indicators due to lack of an explicit domain definition from which the item indicators were derived?

Defining a domain of content for a construct may be a rather demanding task, and certainly it is not a task that is made once and for all. Rather a construct definition should over time be reconsidered to sharpen the domain definition as new knowledge about the construct is gained. It appears that no precise method can be stated to outline the domain of indicators for a construct properly. Even though deciding the conceptual relevance of indicators is a challenge, Nunnally and Bernstein (1994) stated that "... there is no way to know how to test the adequacy with which a construct is measured without a well-specified domain" (p. 88). When a domain description is lacking no frame of reference exists to decide the conceptual relevance of the indicators. A vague domain definition may be better than none. A domain description can be improved in different ways. A need for aiming at a complete and systematic domain description is a general methodological requirement for adequately measuring a construct (Nunnally & Bernstein, 1994). …

Search by... Author
Show... All Results Primary Sources Peer-reviewed


An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.