Comparison of reproductions of the Bender-Gestalt and memory-for-designs by delinquents and non-delinquents.

Abstract

Zolik (1958) compared the Bender-Gestalt (1) reproductions of delinquents and non-delinquents using the Pascal and Suttell (3) scoring system. His trichotomized cutting score placed 5% of the non-delinquents in the critical range labeled as intellectual impairment or emotional maladjustment (5, p. 247). Of his delinquents, 77% fell into this critical area. The purpose of the present study was to replicate these results using both the Bender-Gestalt and the Memory-for-Designs (2). Thirty adolescent institutionalized male delinquents and 30 non-delinquents matched for age and IQ volunteered as subjects. The Wechsler Intelligence Scale for ChildrenRevised (4) was used to estimate the mean IQ for delinquents (M = 88.33) and nondelinquents (M = 91.56). Scores ranged from 67 to 100 for the delinquents and from 73 to 100 for the non-delinquents. All subjects resided in a rural area. The non-delinquents came from various economic backgrounds, whereas the delinquents were of somewhat lower economic status. No one in either group had documented brain impairment, difficulty with vision, or physical handicap. A t test showed a significant difference between delinquents' (M = 53.90, SD = 22.56) and non-delinquents' (M = 33.18, SD = 12.47) errors on the Bender-Gestalt (t = 4.37, p < .001). Differences in errors between the delinquents (M = 2.66, SD = 3.57) and non-delinquents (M = 1.65, SD = 2.35) on the Memoq-for-Designs were not significant (t = 1.29, p < .2). Using Zolik's cut-off criteria, 30% of the delinquent group scored in the critical area on the Bender-Gestalt as opposed to 3% of the non-delinquent group. The value of 35 false-positives for the non-delinquents is satisfying and similar to Zolik's findings. Zolik's cut-off criteria, however, also yielded 47% false-negatives for the delinquents. An improved trichotomized cutting score was implemented for the Bender-Gestalt. This involved lowering Zolik's (5) raw score cut-off value 10 points. Consequently, errors below 40 were considered normal; scores between 40 and 50 represented the borderline range; and those scores above 50 represented the critical range. The new cutting score marginally raised (7%) the false-positive rate for the non-delinquents but substantially lowered (23%) the false-negative rate for the delinquents. Further research is needed.