Scorer's reliability in the quantitative scoring of the H-T-P technique.

The House-Tree-Person Projective Technique (later referred to as H-T-P) includes quantitative and qualitative psychological evaluation of free-hand drawings. The quantitative evaluation which distinguishes this technique essentially from other similar drawing techniques, such as The Drawing of the Human Figure""> or Drawing Completion^*' provides for a scoring system and numerical evaluation of the performance. This quantitative scoring system devised by Buck was standardized on a "carefully selected sample of 120 adults of six intelligence levels (imbecile, moron, borderline, dull average, average, and above average)"'** "• *\ A scoring manual gives detailed directions for scoring and include.s illustrative samples of scorable points in each item (House, Tree, Person); it lists 84 titles for the items to be scored which again include many subtitles, actually increasing the number of scorable points to well over one hundred. Buck's scoring differs from Goodenough's'*' scoring through the provision of flaw scores for missing items along with scores for present items. Four IQ scores are obtained as the final result of the quantitative evaluation which may serve as a rough measure of intelligence'*P̂ '^; but more than that, it is proposed to utilize these IQ scores as an objective basis for qualitative analysis'^'. Comparisons between H-T-P IQ's and other conventional intelligence tests, such as W-B Intelligence Scale, Stanford-Binet, and Otis have been made as either part of the original validation studies'^P " ' or as attempts to cross-validate some aspects of the technique'*"*. The findings in these studies show considerable variations, with correlation coefficients ranging from .165 to .412'** and from .41 to .746'**. Several plausible explanations were suggested to account for such differences. The question, however, as to how much the results may be influenced by individual differences on the part of the scorers appears to be legitimate. While McCarthy"* found interscorer correlations among three scorers of .87, .90, and .92 on the Goodenough Draw-A-Man Test, a review of the literature did not show that an empirical scrutiny of this question has been undertaken on the H-T-P. Therefore, the present study dealing with scorer's reliability in H-T-P scoring has been designed as an attempt to investigate to what extent subjective elements may influence the scoring. It is a part of the H-T-P cross-validation studies proposed by this investigator in 1953 '̂ *. The problem of this study was to determine to what extent different scorers agree in their results on quantitative scoring of the H-T-P drawings, or in other words, to what extent the scorers are reliable.