Title

Effects of rater knowledge of children's and adolescents' presenting problem on scores for the Draw-a-Person: Screening Procedure for Emotional Disturbance (DAP:SPED)

Date of Award

1994

Degree Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Psychology

First Advisor

William G. Wagner

Advisor Department

Psychology

Abstract

Human figure drawings (i.e., man, woman, self) were collected from 20 sexually abused (CSA) and 20 non-sexually abused (NCSA) girls. The drawings by the CSA girls were randomly assigned to one of two conditions: CSA girls (n = 10) who were identified as "sexually abused girls" (i.e., Actual Sexually Abused (ASA)), and CSA girls (n = 10) who were identified as "non-sexually abused girls" (i.e., Pretend Non-sexually Abused (PNSA)). Likewise, the 20 protocols for NCSA girls were randomly assigned to one of two conditions: NCSA girls (n = 10) who were identified as "sexually abused girls" (i.e., Pretend Sexually Abused (PSA)), and NCSA girls (n = 10) who were identified as "non-sexually abused girls" (i.e., Actual Non-sexually Abused (ANSA)). Raters (n = 3) were trained to score the human figure drawings according to the DAP:SPED scoring system. The results revealed no significant difference on Total DAP:SPED standard scores for the ASA and ANSA samples or for the PSA and PNSA samples; no significant difference on Total DAP:SPED standard scores given to the CSA and NCSA samples; no significant differences in the raw scores given to the drawings of the man, woman, and self; and no significant difference in the raw scores given to the man, woman, and self drawings for the ASA and ANSA samples or the PSA and PNSA samples. Post hoc analysis of a significant three-way interaction (i.e., Actual samples x Pretend samples x Drawing Topic) revealed that raw scores given to the drawings of the man for the Actual samples were significantly lower than those obtained for the Pretend samples. Because additional analyses revealed a main effect for Rater, data were re-analyzed with rater examined as a repeated measure. Results revealed a significant three-way interaction (i.e., Actual samples x Pretend samples x Drawing Topic); however, post hoc analysis using paired t-tests failed to localize the source of the significant difference.