Journal of School Psychology 1980 • Vol. 18, No. 1
0022-4405/80/1300-0067500.95 (~) 1980 The Journal of School Psychology, Inc.
EMOTIONAL AND DEVELOPMENTAL ASPECTS OF HUMAN FIGURE DRAWINGS IN PREDICTING SCHOOL READINESS CHARLES W. SZASZ
LYLE E. BAADE
Kanawha County Schools Charleston, West Virginia
University of Kansas School of Medicine Wichita, Kansas
CHARLES W. PASKEWlCZ West Virginia College of Graduate Studies Institute, West Virginia
Summary: It was hypothesized that the combined use of Koppitz' developmental and emotional scores can improve the prediction of school readiness from children's Human Figure Drawings (HFD) when compared with each measure separately. The relationship of SES to HFD scores was also investigated. Multiple regression analysis was used with four factors----developmental score, emotional score, sex of the child, and SES of the school-to derive comparative correlations to test the hypotheses and to create a prediction equation to estimate the Metropolitan Readiness Test (MRT) scores of 141 kindergarten students ages 5 to 6 years representing three SES levels. Hit rate data were also presented. The results showed that the correlation for the combined developmental and emotional scores with the MRT score was not different than their separate correlations with the MRT. The SES factor was found to be significantly correlated (p < .01) with both developmental and emotional scores. The best single variable predictor of MRT scores was SES. The application of Bayes' Theorem of probability to the hit rate data showed that the nonreadiness predictions yielded by the developmental score, emotional score, and their combination are not better than chance predictions. Thus, it was found that Koppitz' developmental or emotional scoring systems are not viable screening instruments. In recent years there has been a growing interest in the screening of kindergarten students both as a means to determine the readiness for learning and as a way to identify children with potential learning problems. Clear advantages would appear to lie in the early identification of those students who would experience learning problems as their academic careers progress. The prediction of learning problems would in many cases permit the early application of appropriate educational programming, thus preventing later emotional and educational difficulties. Two important factors in school readiness are intelligence and social-emotional adjustment (Hammond & Skipper, 1962; Ilg & Ames, 1965; Kohn & Rosman, 1972; Koppitz, Sullivan, Blyth & Shelton, 1959; Sawrey, 1955; Trachtman, 1958; Petrone, Note 1). Several recent studies (Koppitz, 1967; Shipp & Louden, 1964; Strahl, 1972; Vane & Kessler, 1964; Berkowitz, Note 2) have effectively demonstrated the use of developmental scoring systems in evaluating preschool and beginning first-grade students' human figure drawings as measures of intelligence. The Koppitz Developmental Scoring system is a measure of intellectual functioning and is not significantly affected by children's drawing ability nor by school training in kindergarten (Koppitz, 1968). Other studies (Dillard & Landsman, 1968; Koppitz, 1966a, 1966b, 1967, 1968; Vane & Eisen, 1962; Vane & Kessler, 1964) have demonstrated the efficacy of scoring 67
68
Journal of School Psychology
systems to assess the emotional adjustment of young children from human figure drawings. Scores obtained by the Koppitz system for emotionality (emotional indicators) are for the most part unrelated to chronological age and maturation (Koppitz, 1968). Research to date has investigated only one aspect of children's drawings-emotional or intellectual. However, if intelligence and socioemotional adjustment are both important factors in school readiness, then it is essential to consider both these factors together in predicting school readiness. The Koppitz Human Figure Drawing Test lends itself to multiple interpretations since scoring systems are available to assess emotionality and intellectual ability independently. Other important features of the HFD test include convenience and its appropriateness for children. It is convenient since the time required to evaluate a child is brief. Also, the HFD can be used either as a group test or as an individual test. It is appropriate for children since it requires no reading, takes a short time, and is an enjoyable task for most children (Koppitz, 1968). The purpose of the present study was to test the hypothesis that the combined use of Koppitz' developmental and emotional scores can improve the prediction of school readiness from children's HFDs when compared with each measure separately. METHOD
Subjects. The subjects were 141 kindergarten students (62 boys, 79 girls) enrolled in six schools of the Kanawha County Public School System during the 1976-77 school year. Originally 150 children (68 boys, 82 girls) participated in the study. Of the nine students who were deleted from the study, five moved out of the county, three students were absent during the testing phases of the study, and one student was found to be chronologically too young for inclusion in the study. The mean age of the 141 students was 5 years, 5 months. Because a close relationship between SES and intelligence has been suggested by other studies (e.g., Haggard 1954), two schools were selected to represent each of three SES levels based on the percentage of children attending each school from low-income families. The two schools with the lowest percentage (0%) of children from low-income families were designated as schools A and B. The two schools with a moderate percentage (13%, 12%) of children from low-income families were denoted as schools C and D. The two schools with the highest percentage (36%, 57%) were designated as schools E and F. Schools A and B were comprised of 44 students, whereas schools C and D and E and F had 54 and 43 students, respectively. Procedure. The standard administration of the Human Figure Drawing test was employed with groups of kindergarten children. Each group of students was requested to draw "a whole person and not a stick figure or a cartoon figure" on a blank sheet of 8½" x 11" white paper, with a black crayon (Koppitz, 1968). The tests were administered within a one-week period of time during the last week in October, the second month of the school year. The HFD protocols were scored for the presence of Koppitz' 30 emotional indicators and 26 developmental items (Koppitz, 1968). At the end of the school year, the Level II, Form Q of the 1976 edition of the Metropolitan Readiness Test (MRT) was administered to all of the subjects in groups by their classroom teachers according to standard procedures outlined in the Metropolitan manual. Prior to the administration of the MRT, a meeting was held with the teachers to familiarize them with the standardized testing procedures of the MRT. The MRTs were scored by the teacher.
Szasz, Baade, and Paskewicz
69
The MRT was selected because of its wide use and its established reliability and validity. The MRT manual reports a split-half reliability of.94 and an alternate form reliability of.87. Validity studies reported in the manual indicate correlation of.70 and .69 between the MRT and the total reading scores of the Metropolitan Achievement Test and the Stanford Achievement Test, respectively. These validity studies correlated MRT scores obtained at the beginning of the first grade and achievement scores obtained at the end of the first grade. Because four factors--SES of the school, sex of the child, emotional score, and developmental score--are utilized in this investigation, multiple regression analysis was used in the treatment of the data in deriving multiple correlation coefficients to test the hypothesis and in creating a prediction equation to estimate the criterion score, MRT score (Metropolitan Pre-Reading Skills Composite Score). Hit rate data were presented in the form of percentages of those students correctly identified, as well as false positives and negatives for each scoring system and their combination. The criterion for the hit rate data was the Reading Skills Composite Score of the MRT. All kindergarten students who achieved a Performance Rating of Average (stanines of four, five and six) or High (stanines of seven, eight and nine) on the MRT score were considered academically ready for the first grade. All students who achieved a low Metropolitan Performance Rating (stanines of one, two and three) on the MRT score were considered academically not ready for school. The predictors for the hit rate data were Koppitz' emotional and developmental scores. Children who achieved an emotional score of 2 or more on their HFDs were designated as not being ready for school, whereas students who achieved an emotional score of 1 or less were considered ready for school. All students who achieved a developmental score of 3 or less on their HFDs were arbitrarily designated as not being ready for school. Children with a developmental score of 4 or more were considered ready for the the first grade. Hit rate data were computed separately for the developmental and emotional scores. Utilizing the predictors in combination with each other, hit rate data were computed comparing students who are considered ready for school, as designated by an emotional score of less than 2 or a developmental score of 4 or more on their HFDs. Similarly, hit rate data were computed for students who are not considered ready for school, as designated by both an emotional score of 2 or more and a developmental score of 3 or less. RESULTS The developmental score yielded a multiple correlation coefficient of .43 (p <.0001) and accounted for approximately 19% of the variance. The emotional score yielded a multiple correlation coefficient of - . 2 6 ( p < .01) and accounted for approximately 7% of the variance. When the two variables were used in combination to predict the MRT score, the multiple correlation coefficient (.43, p < .0001) and the amount of variance accounted for (approximately 19%) were identical to those of the developmental score by itself. Thus, the evidence failed to support the hypothesis. Table 1 presents the multiple correlation coefficient matrix for all the variables. A significant intercorrelation exists among the variables: developmental score, emotional score, sex, SES, and MRT score. The relatively high correlation between the developmental score and the emotional score (r = - .64) clearly shows that a relationship between these variables exists. Table 1 also shows that there is a significant relationship between the SES level and the MRT
Journal o f School Psychology
70
Table 1 Multiple Correlation Coefficient Matrix Variable 1. 2. 3. 4. 5.
1
Developmental Score Emotional Score SES Sex MRT Score
2
---. .
3
-.64" --.
.43* -.34" -. .
. .
.
4
5
.05* .29* .06
.43* .26* .46* .02
.
NOTE: N = 141 *p ~< .01
score (r = - . 4 6 ) . The SES factor w a s found to be also significantly correlated with both the developmental and emotional scores (r ~ .43 and r = - . 3 4 ) . Using stepwise multiple regression analysis, the best one and two variable prediction equations were created to predict the Metropolitan Pre-Reading Skills Composite Score (MRT). (a) For One Variable: M R T = 6.85 + ( - 1 . 1 1 ) X1 where X ~ = SES (b) For T w o Variables:
M R T = 2.46 + (0.83) X~ + ( - 0 . 1 8 ) X2 where X1 = Developmental Score, Xz = Developmental Score X SES
Table 2 Hit Rate Data for the Prediction of MRT Scores
Prediction From Developmental Score Emotional Score Developmental Score +Emotional Score Best One Variable Prediction Equation (SES) Best Two Variable Prediction Equation (Developmental Score + Developmental Score x SES) NOTE:
Valid Positives n
False Negatives n
False Positives n
Valid Negatives n
17 29
26 14
11 36
87 62
17
26
11
87
18
25
25
73
25
18
29
69
N = 141 Valid Positives = the number of children correctly identified as ready for school. False Negatives = the number of children identified as not ready for school when they were ready. False Positives = the number of children identified as ready for school when they were not ready. Valid Negatives = the number of children correctly identified as not ready for school.
Szasz, Baade, and Paskewicz
71
The one variable prediction equation (SES) yielded a multiple correlation coefficient of.46 (p < .0001) and accounted for 21% of the variance. The two variable prediction equation (developmental score and developmental score x SES) yielded a multiple correlation coefficient of.53 (p < .0001) and accounted for 28% of the variance. Table 2 presents the hit rate date for the prediction of the MRT score from the developmental score, emotional score, the combined developmental and emotional scores, the best one variable prediction equation (SES), and the best two variable prediction equation (developmental score + developmental score × SES). Hit rate data shows that the developmental score correctly identified 74% (true positives + true negatives) of kindergarten students. This is in contrast to the emotionai score, which correctly identified 64% of the kindergarten students. When both the developmental and emotional scores were combined (readiness = adequate developmental score or emotional score; nonreadiness = poor developmental and emotional score) to predict MRT score, the hit rate data were identical to those of the developmental score alone (74%). The probability associated with the developmental score, emotional score, and their combination in predicting school readiness was determined by applying Bayes' Theorem of probability to the hit rate data. The developmental score and the developmental score + the emotional score both have probabilities of only .61 of predicting a kindergarten student as being not ready for school when he is in fact not ready. The other three predictors----emotional score, SES, and developmental score + developmental score × SES with probabilities of .43, .42, and .58 respectively---are all about equal to chances. Thus, these data provide additional support for the conclusions made on the basis of the multiple correlation coefficients. DISCUSSION Since the multiple correlation coefficient for the combined developmental and emotional scores with the MRT score is not any better than the separate multiple correlation coefficients for these scores with the MRT, the evidence provided no support for the hypothesis. The developmental score and the emotional score are correlated, and the possibility that they are in fact measuring a common factor should be considered. Such a relationship would explain the lack of additional predictability when the developmental score and the emotional score are used in combination to predict the MRT score. Previous literature has dealt with these measures as being separate. Koppitz (1967, 1968) reported findings that the Developmental Items were primarily related to age and maturation and that they could be used to assess a child's general level of mental maturity. In several other studies, Koppitz (1966b, 1968) demonstrated that the emotional indicators were not primarily related to age and maturation and he showed the efficacy of these indicators in assessing the emotionality of children. In a related study, Koppitz (1966a) found the emotional indicators to be diagnostically highly significant in predicting achievement problems. The above studies suggest that the developmental items and the emotional indicators are both different, and both helpful. However, the results of this study show that the two measures are highly correlated and their use in combination adds nothing to prediction.
72
Journal of School Psychology
The relatively high correlation between SES and the developmental score is not unexpected, and it is suggested that the two are related to intelligence. The close relationship of SES with intelligence has been verified by other studies (e.g., Haggard, 1954; Jensen, 1969). The relationship between SES and the developmental score is also supported by the relatively high correlations of the two with the MRT. In fact, SES and the developmental score have about the same correlation with the MRT score. Since the criterion for SES levels was on a group basis by school, the question of what would happen if the criterion for SES levels were on an individual basis for each child remains for further research. The relationship between SES and the emotional score is not a clear one. The modest correlation of SES with the emotional score suggests that they may also be measuring a common factor to some extent. A comparison of the MRT correlations with SES and the emotional score suggests that the common factor is probably intelligence. The emotional scoring system may be only a variation of the developmental scoring system. In fact, 8 of the 30 emotional indicators are included in the 26 developmental items. This conclusion is further supported by the high correlation between the emotional and developmental scores. Although the developmental score, emotional score, their combination, SES, and developmental + developmental x SES score all correlated significantly with MRT, clinical evidence suggests that their use yields prediction no better than chance. If the 141 students in the present study were all classified as ready for school, only 43 students, or 30%, would be misidentified. Using the developmental score, 37, or 26%, of the students would be misidentified. Even considering the ease with which the HFD can be administered and scored, this small gain in prediction accuracy must be viewed as impractical. Of course hit rates will vary depending upon the base rate of children not ready for school. Assuming 50% of a population were not ready for school, the probability that a child identified by his developmental score as not ready for school was in fact not ready would be .78. The probability that a child identified as ready for school was in fact ready would be .59. In a population of 141 students, the use of the developmental score would thus misclassify 44, or 31%, of the students. The increase in accuracy of predictions over random classification of students would be 19%. Thus it can be seen that, depending upon the population parameters, the developmental score might have some limited utility. However, as a general screening device with populations where 70% or more of the children are ready for school, the efficacy of the developmental score must be challenged along with the other HFD predictor scores examined in this study. Therefore, school psychologists and allied personnel who use these scoring systems thinking that they predict readiness should reconsider using them for these purposes. Finally, and perhaps most importantly, the disparity between statistical and practical significance suggests that other screening devices should also be re-evaluated through the use of hit rate data and probability theory.
REFERENCES NOTES 1. Petrone, F. R. An evaluation o f some intellectual measures in predicting first grade achievement. Unpublished doctoral dissertation, Temple University, 1963. 2. Berkowitz, M. C. A revised Goodenough Draw-A-Man Test as a measure o f school readiness. Unpublished Master's thesis, Kansas State Teachers College, 1969.
Szasz, Baade, and Paskewicz
73
REFERENCES Dillard, H. K., & Landsman, M. The Evanston Early Identification Scale: Prediction of school problems from the human figure drawings of kindergarten children. Journal of Clinical Psychology, 1968, 24, 227-228. Haggard, E. A. Social status and intelligence: An experimental study of certain cultural determinants of measured intelligence. Genetic Psychological Monographs, 1954, 4, 141-186. Hammond, S. L., & Skipper, D. S. Factors involved in the adjustment of children entering first grade. Journal of Educational Research, 1962, 56, 89-95. llg, F. L., & Ames, L. B. School readiness: Behavior tests used at the Gessell Institute. New York: Harper & Row, 1965. Jensen, A. R. How much can we boost IQ and scholastic achievement? Harvard Educational Review, 1969, 29, 1-123. Kohn, M., & Rosman, B. L. Relationship of preschool social-emotional functioning to later intellectual achievement. Developmental Psychology, 1972, 6, 445-452. Koppitz, E. M. Emotional indicators on Human Figure Drawings and school achievement of first and second graders. Journal of Clinical Psychology, 1966, 22, 481-483. (a) Koppitz, E. M. Emotional indicators on Human Figure Drawings of children: A validation study. Journal of Clinical Psychology, 1966, 22, 313-315. (b) Koppitz, E. M. Expected and exceptional items on Human Figure Drawings and IQ scores of children ages 5 to 12. Journal of Clinical Psychology, 1967, 23, 81-83. Koppitz, E. M. Psychological evaluation of children's Human Figure Drawings. New York: Grune & Stratton, 1968. Koppitz, E. M., Sullivan, J., Biyth, D. D., & Sbelton, J. Prediction of first grade school achievement with the Bender Gestalt Test and Human Figure Drawings. Journal of Clinical Psychology, 1959, 15, 164-167. Sawrey, J. M. The predictive effectiveness of two non-verbal tests of intelligence in first grade. California Journal of Educational Research, 1955, 5, 133. Shipp, D. E., & Loudon, M. L. The Draw-A-Man Test and achievement in the first grade. Journal of Educational Research, 1964, 57, 518-519. Strahl, R. R. The use of human figure drawings collected in kindergarten for prediction of learning performance in the first grade (Doctoral dissertation, Northern Illinois University, 1972). Dissertation Abstracts International, 1973, 33, 4191A. (University Microfilms No. 73-4178, 117). Trachtman, G. M. Personality and developmental characteristics of children rated most and least ready for first grade by their kindergarten teachers (Doctoral dissertation, New York University, 1958). Dissertation Abstracts International, 1959, 20, 1002A. (University Microfilms No. 59-1042, 191). Vane, J. R., & Eisen, V. W. The Goodenough Draw-A-Man Test and signs of maladjustment in kindergarten children. Journal of Clinical Psychology, 1962, 18, 276-279. Vane, J. R., & Kessler, R. The Goodenough Draw-A-Man Test: Long term reliability and validity. Journal of Clinical Psychology, 1964, 20, 487-488. Charles W. Szasz School Psychologist Kanawha County Schools 200 Elizabeth Street Charleston, West Virginia 25311 Lyle E. Baade Depa~ment of Psychiatry University of Kansas School of Medicine 1001 N. Minneapolis Wichita, Kansas 67214 Manuscript received: June 2, 1978 Revision received: January 9, 1979
Charles W. Paskewicz Assistant Professor of Psychology West Virginia College of Graduate Studies Institute, West Virginia 25112