No abstract

PROBLEMIn 1963, Harris in his book on Children's Drawings(2) published a revision and re-evaluation of the Goodenough Draw-a-Man Test. The revision consisted mainly of adding 22 new scoring categories to the original 51 categories developed by Goodenough and in devising a Draw-a-Woman Scale which paralleled the Drawa-Man Scale. He presented two separate scoring systems, one for the Man and one for the Woman. I n addition, Harris developed a Quality Scale by which an I& for the Draw-a-Man and Draw-a-Woman Tests might be derived froin comparison with 12 quality plates. Harris(2) admitted, however, that "the saving in time by using the Quality scale is accompanied by an appreciable reduction in accuracy of measurement."This study investigated whether the method of scoring the Draw-a-Man Test, including the use of the additional categories devised by Harris, was significantly better than the Goodenough method of evaluating the same test. The directions used by Harris for the Draw-a-Man Test are the same as those used by Goodenough so it was felt a meaningful comparison could be made. PROCEDURE836 Draw-a-Man Tests drawn by white and Negro girls and boys ranging in age from 5 through 7 years were taken from the files of several New York school districts. These drawings had been administered as part of routine classroom testing situations and had been scored by the Goodenough scoring method by the various psychologists who had administered the tests. All of the children whose drawings were selected for study had been given individual Stanford Binet intelligence tests by school psychologists as part of the school program. Although a number of these children had been tested because they were school problems, the majority included in the sample came froni classrooms where the Binets had been routinely administered as part of an educat,ional and intellectual assessment project being conducted in the various schools.All 336 drawings were scored by the Harris method by a single psychologist. The first 30 drawings were rechecked by another psychologist to make sure no deviation from the Harris method occurred. The original Goodenough scoring done by the school psychologists was not changed.

Statistical analyses of the data were made possible through the Hofstra University Computer Center. 84 89 98108 113 ll7 l27136ljo 83 88 9 7 1 0 6~l l 5~~1~1 4 8 82 86 95 105 109 ll4 123 132 145 81 85 94 103 107 l l 2 121 130 143 79 84 93 101 106 ll0 U9 128 141 78 83 91 100 104 109 116 126 139 77 81 90 99 103 107 115 124 137 7 6 8 0 8 9 9 7 101 106 114 123 135 75 79 88 96 100 104 U 3 121 133 74 78 86 95 99 103 Ul 119 132 73 77 85 93 97 101 109 ll8 lW 72 76 84 92 961001~ll6l.28 71 75 83 91 95 99 107 114 126 70 74 82 90 97 105 113 125 69 73 8 l 88 92 96 104 112 123 68 72 80 87 9l 95 103 110 122 68 71 79 86 90 94 101 109 120 67 70 78 85 89 93 100 107 119 66 70 n 84 88 91 99106117 65 69 76 83 87 90 96 105 116 5-2 5 3 5k 5 5 5-6 5 7 5-8 5-9 5 1 0 551 6-0 6-1 6-2 6-167 171 84 98 108 117 131 141 155 159 164 169 83 97 106 115 129 138 152 157 l62 166 82 95 105 114 127 136 1 9 155 159 164 81 94 103 112 125 IN 148 1 s l m 161 79 93 101 ll0 1% 132 146 150 1% 159 78 91 100 109 122 130 143 148 1% 1% 77 90 99 107 120 129 141 146 1% 1 9 76 a9 97 106 ll8 127 139 3 148 15 2 75 88 % l a 117 125 138 142 146 1% 74 86 95 103 U 5 123 136 140 144 148 73 85 93 101 ll4 122 1 9 138 142 146 72 84 92 100 ll2 120 132 l36 140 144 71 83 91 99 11l 118 130 138 142 70 82 90 W 109 117 I29 132 136 140 69 81 88 96108 115127131135138 68 80 87 95 106 114 125 129 133 137 68 79 86 94 105 112 124 128 13l 135 67 78 85 93 104 U 122 126 130 133 66 ?? 84 91 102 ll0 121 I24 128 132 65 76 83 90 101 108 ll9 123 127 130 c w Qo 2 E w m r! cj M s e

No abstract

PROBLEM AND METHOD The widespread use of the Goodenough Draw-a-Man Test and the variable results obtained with it provided the impetus for a long term study of reliability and validity.Reliability was determined by administering the Draw-a-Man Test to a11 kindergarten children in five New York State public schools and retesting the same group in the first, second and third grades. Twelve months intervened each time. The original group was composed of 280 white and Negro children from all aocioeconomic levels. Four years later, because of transfers and absences, only 112 children remained.Validity was measured by comparing the results of the Draw-a-Man Test with the results of the Stanford Binet Scale administered to 98 of the children, and with the results of the American School Achievement Test given in third grade. In addition the third grade teachers rated all the children on a nine point behavior adjust ment scale. Table 1 shows that although the kindergarten, first, second and third grade Draw-a-Man Test results all correlated highly with one another, none were as high aa the .94 obtained by Goodenough(*). The fact that the interval between Good-TABLE 1. MEANS AND STANDARD DEVIATIONS OF DRAW-A-MAN TEST, STANFORD BINET, AND AMERICAN SCHOOL ACHWVEMJZNT TESTS WITH CORRELATIONS BETWEEN TEE TESTS K DAMT let DAMT (N-112) 2nd DAMT (N-112) 3rd DAMT (N-112) S. Binet (N-98) Achvt.* (N 112) ($-112) Mean SD RESULTS104.1 20.5 102.6 17.9 98.4 17.0 99.0 18.4 104.0 17.6 + .03 .98 Kgt 1st 2nd 3rd Stanford DAMT DAMT DAMT DAMT Binet .75 .69 .80 .65 .69 .74 .53 .57 .58 .54 .52 .47 .40 .48 .65 *Reported in deviation from grade level.enough's test-retest wm only one day may have resulted in the higher correlation obtained by her. Studies by McCarthyca) and McCurdy'') who used longer intervals, one week and three months respectively, obtained lower correlations, .68 and .69. Our results indicate that the tests with the greatest time interval between them, namely the kindergarten and third grade, showed the lowest reliability, whereas those closest in time, namely the kindergarten and first, first and second, and second and third, showed the highest reliability.In terms of validity, all the Draw-a-Man Test results show a substantial correlation with the Binet and indicate that to a large extent both scales are meaauring the same function, namely intelligence. As expected, the Binet showed the highest