This study evaluated the scoring systems of Goodenough and Harris (1963) and Koppitz (1968) for using human figure drawings to assess the intellectual abilities of children. Drawing scores of 125 children, aged 5 to 15, were compared to their performance on the Wechsler Intelligence Scale for Children-Revised (WISC-R), while the drawing scores of a separate group of 74 children, aged 5 to 12, were compared to their performance on the Stanford-Binet Intelligence Scale, Form L-M. While both drawing systems correlated significantly with the WISC-R as well as the Stanford-Binet, the longer and more detailed Goodenough-Harris had a significantly higher correlation with Performance IQ on the WISC-R than did the Koppitz. Neither drawing system had a pattern of significantly different validity coefficients for children of varying ages or IQ levels.
Clinicians who work with a diverse child population are likely to encounter some children who need intellectual evaluations, but who are either unwilling or unable to take one of the standard, individually administered IQ tests. Children who are autistic, severely hyperactive, or non-English speaking all can present significant problems to the evaluator. Situations in which a large number of non-reading children must be assessed can pose similar difficulties (Jegede & Bangboye, 1981). While individually administered IQ tests, such as the various Wechsler scales or the Stanford Binet Intelligence Scale (Stanford-Binet), remain the preferred method of assessment, human figure drawings may offer a feasible alternative for children who cannot tolerate a standard IQ test.
While educators and psychologists have looked to children's drawings as a measure of cognitive ability for more than one hundred years (e.g., Burt, 1921;Cooke, 1885; Ricci, 1887), it was not until 1926 that Florence Goodenough empirically demonstrated how children's drawings reflect their intellectual skills and development. Goodenough believed that the more developed intellectually a child was, the more realistic details he or she would include on a human figure drawing. Based on this belief, Goodenough developed the Draw-A-Man Test (later modified by Karen Machover I19491 as the Draw-A-Person Test) and a cognitive scoring system for that test.
Studies that examined this scoring system generally have shown good reliability, but were variable in their demonstration of validity. Harris (1%3), in a review of this research, reports reliability coefficients commonly above .90. However, concurrent validity coefficients when correlated with Stanford-Binet scores ranged from .26 to .92 and when correlated with Wechsler scores ranged from .38 to .77. The majority