Academic journal article Canadian Journal of Behavioural Science

Evaluation of a Checking Procedure for Reducing Clerical and Computational Errors on the WAIS-R

Academic journal article Canadian Journal of Behavioural Science

Evaluation of a Checking Procedure for Reducing Clerical and Computational Errors on the WAIS-R

Article excerpt

Abstract

Examiners make mistakes administering and scoring Wechsler intelligence tests. Such errors can have a substantial impact on IQ. We produced a Compu - Check Form (CCF) of verification algorithms for checking WAIS - R clerical and computational procedures. Senior undergraduates and M.A. graduate students were trained to score fictitious WAIS - R protocols. Sixty percent of the students made errors and approximately 30% of the protocols contained errors. These errors frequently resulted in IQ discrepancies. Most IQ inaccuracies were small, although 10% of the summary IQS on protocols with errors deviated between 4 and 12 points. A subsample of the students were also trained to use the CCF. Fewer subjects made errors after applying the CCF. Changes in error rates and corrections to summary IQS also supported the utility of the CCF. In a field trial, 6 of 7 practitioners who used the CCF detected errors on 15 of 47 WAIS - R protocols selected from their clinical files. Only one of these clinician errors resulted in a substantial IQ error. Methodological issues and implications of the results are discussed.

Resume

Les examinateurs font des erreurs en administrant et en notant les tests d'intelligence. Ces erreurs peuvent avoir des effets considerables sur l'evaluation du QI. Nous avons cree une formule de contro@le informatique (Compu - Check Form, CCF) d'algorithmes de verification pour contro@ler les procedures d'ecriture et d'informatique dans le cas du test WAIS - R. Des etudiants en terminale du premier cycle et des etudiants du deuxieme cycle ont recu la formation necessaire pour noter des tests WAIS - R fictifs. Dans le groupe d'etudiants, 60 p. 100 ont fait des erreurs et environ 30 p. 100 des protocoles de test contenaient des erreurs, et ces erreurs entrai@naient souvent des inexactitudes relativement au QI. La plupart des inexactitudes de QI etaient minimes, mais 10 p. 100 des QI sommaires figurant sur les protocoles errones deviaient de 4 a 12 points. Un sous - echantillon d'etudiants a aussi ete forme a l'utilisation de la formule de contro@le informatique. Un plus petit nombre de sujets ont commis des erreurs apres avoir applique la formule. Les changements dans le taux d'erreurs et les corrections apportees aux QI sommaires sont aussi venus appuyer l'utilite de la formule de contro@le informatique. Dans un essai sur le terrain, six praticiens sur les sept ayant utilise la formule ont decele des erreurs sur 15 des 47 tests WAIS - R choisis dans leurs dossiers cliniques. Un seule de ces erreurs de clinicien a entrai@ne une deviation appreciable du QI. L'article traite aussi des questions methodologiques et des implications des resultats.

It is well established that examiners make a variety of administration, scoring, clerical and computational errors with the Wechsler scales of intelligence (Bradley, Hanna, & Lucas, 1980; Moon, Blakey, Gorsuch, & Fantuzzo, 1991; Ryan, Prifitera, & Powers, 1983; Slate, Jones, & Murray, 1991). Graduate students and practitioners alike frequently make such errors (Slate & Jones 1990; Slate, Jones, Murray & Coulter, 1993). Examiner errors detract from the otherwise sound psychometric properties of the Wechsler intelligence scales. Estimates of error impact vary. Examiner errors often result in relatively small mean IQ discrepancies, but the range of IQ discrepancies can be considerable (Beasley, Lobasher, Henley, & Smith, 1988; Slate et al., 1991; Slate et al., 1993). Some have argued that administration and scoring errors double the error estimates provided in test manuals based upon internal consistency (Hanna, Bradley, & Holen, 1981).

Clerical and computational errors on the Wechsler scales arise in many forms. Simple mistakes in addition of raw and scaled scores have been found to occur frequently (Sherrets, Gard, & Langner, 1979; Slate & Chick, 1989). Arithmetic errors also occur in the calculation of chronological age (Hajzler, 1987; Slate et al, 1993). …

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

Oops!

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.