Diagnostic Testing in Educational Settings

Introduction

Diagnostic testing in education refers to an in-depth assessment of pupils' learning difficulties, whatever their causes. The way to conduct such an assessment is not obvious. An essential distinction should be made between an assessment that focuses on performance and an assessment that focuses on competence. The first section clarifies this distinction. The second and the third sections analyse the usefulness and the limits of each of these two levels of assessment.

Two Levels of Diagnostic Testing: Performance and Competence

In the field of linguistics, Chomsky (1965) made a fundamental distinction between performance and competence. Performance refers to the use of language in functional situations. Competence refers the underlying system of rules mastered by the speaker. The speaker's competence cannot be observed directly. This is only inferred from the speaker's performance, which imperfectly reflects [Page 335]his competence. This distinction between performance and competence is essential for the diagnosis of learning difficulties. The diagnostic testing can focus on the quality of the behaviours produced by a subject – the performance – (e.g. the correctness of a text reading) or on the cognitive abilities underlying these behaviours the competence – (e.g. the mental processes used when reading a word). Diagnostic tests are very different if they target the assessment of performance or competence.

Performance tests are basically a-theoretical, in the sense they do not refer to any model of the mental activity being at work in the items. The test items are usually selected on the basis of an accurate definition of the knowledge domain to be assessed and on a specific description of the behaviours corresponding to the mastery of a particular knowledge. The validation of the performance test items relies on experts of the domain who judge the pertinence of the selected items with regard to the learning goals.

On the other hand, competence tests rely on models of cognitive processes involved in the items. To validate these tests, the pertinence of the items chosen as indicators of the mental processes to be measured should be proved. Are the selected tasks involving the intended processes, and only these? Validation of the competence tests is often difficult because the cognitive activities involved in the items are never straightforward. Even apparently very simple tasks involve rather complex cognitive processes (e.g. Longstreth, 1984). Since a pure measure of the intended processes does not exist, the interpretation of the scores in competence tests is often difficult. Moreover, the models underlying the competence tests are only partial and temporary representations of the mental reality. Therefore, the difficulty of building and interpreting such tests could stem from model shortcomings.

Diagnostic Testing That Focuses on Performance

There are two main arguments for using performance tests to diagnose learning difficulties. The first is that the performance measurement guarantees the ecological validity of the diagnostic testing. Performances are useful behaviours allowing subjects to cope with everyday problems. When the diagnostic assessment only focuses on mental processes, there is a risk of unduly emphasizing disabilities without any real consequences for the subject's environmental adaptation. The second argument for using performance tests is that their items match the teaching goals. Performances are essential information for appraising the child's adjustment to the school demands. Referring to school demands avoids describing an imperfect performance, but corresponding to a sufficient performance at a given school level, as an indication of learning disability. Learning is a step-by-step process. It is essential to distinguish between what is related to the normal learning process and what is related to disability.

...