THE RELATIONSHIP BETWEEN GSR AND ANXIETY INDEXES IN PROJECTIVE DRAWINGS Thesis for the Degree of Ph. D. MICHIGAN STATE UNIVERSITY Leonard Handler 1964 ll mum; Illzlllsljlfllllll L"! g; MI 11" @ IQ Mill; II 1, This is to certify that the thesis entitled The Relationship Between GSR and Anxiety Indexes in Projective Drawings presented by Leonard Handler has been accepted towards fulfillment of the requirements for 4911.11.— degree in My QW/ W / /Major prglessor 0-169 LIBRARY Michigan State University PLACE ll RETURN BOX to remove this checkout from your record. TO AVOID FINES return on or betore date due. DATE DUE DATE DUE DATE DUE MSU le An Affirmdtve AetiorVEqueI Opportunlty Infiltwon . 6mm.- .——————— ‘ ’ “‘-".“ 7 —-— ~r — * * -'-— - M __, A, ABSTRACT THE RELATIONSHIP BETWEEN GSR AND ANXIETY INDEXES IN PROJECTIVE DRAWINGS by Leonard Handler: Ninety-six male college students drew a man, woman. and an automobile while continuous GSR's were obtained. A counterbalanced design was employed. Thus, there were six groups of 16 subjects each. The following hypotheses were investigated: I. Drawing a man and a woman is more anxiety producing than drawing an automobile, because of the greater reflection of symbolic material. II. Drawing a woman is more anxiety producing. followed by drawing a man, and the automobile. It was thought that the relatively well adjusted male college student would have more active problems concerning heterosexual relationships than with homosexual or homoaffectional relationships. Leonard Handler III. a) There is a progressive diminution of anxiety, due to the desensitization of the anxiety- producing properties of the testing session. There is less anxiety associated with drawings done second. and third. compared with drawings done first (adapta- tion). b) The anxiety-producing properties of a drawing vary as a function of the six possible orders of presentation. IV. Primary sexual characteristics are more anxiety producing than secondary sexual characteristics. which. in turn, are more anxiety producing than non- sexual body areas. V. There is a direct relationship between the number and degree of graphic indexes of anxiety and the GSR. VI. The various indexes of anxiety differ in terms of the magnitude of the anxiety which they reflect. from most to least, in the following order: 1. heavy pressure 2. heavy line 3. delineation line absence Leonard Handler 4. erasure 5. light pressure 6. light line 7. small size 8. distortion 9. body simplification 10. head simplification ll. omission 12. shading 13. vertical imbalance 14. emphasis line 15. re- inforcement l6. placement (upper left) 17. line discontinuity 18. large size 19. small head 20. large head 21. transparency 22. hair shading 23. head:body ratio. Five of the six hypotheses were supported. A significant difference was found in the degree of an- xiety for the man. woman. and automdbile drawings for the graphic anxiety indexes. and for the two GSR meas- ures (GSR frequency. and mean conductance). Thirteen (out of 18) of the separate indexes significantly dif- ferentiated the man and woman drawings from the auto— mobile drawing. Thus. the automobile. as a more neu- tral figure, may be used as a control drawing in clinical practice. The automdbile yielded the lowest level. and the woman drawing the highest level of anxiety. There were no order differences for the GSR. but there were Leonard Handler order differences for five of the 23 graphic indexes. Adaptation effects were found for the two GSR measures. and for seven of the graphic anxiety indexes. For the man drawing, secondary sexual body parts yielded the greatest anxiety. followed by the sexual area. and the nonsexual body areas. For the woman drawing. the sex- ual body areas yielded the greatest anxiety. followed by the secondary sexual body areas. and the nonsexual body areas. There were no significant correlations between the summed graphic indexes and the two GSR variables. but 10 of the individual indexes correlated signifi- cantly with one or both of the GSR measures. Five of the indexes correlated significantly and negatively with GSR frequency. thereby supporting previous re- search. The attempt to predict the rank order of the 23 indexes in terms of how well they reflected anxiety was unsuccessful. THE RELATIONSHIP BETWEEN GSR AND ANXIETY INDEXES IN PROJECTIVE DRAWINGS BY Leonard Handler A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Psychology 1964 Acknowledgments Although this volume bears but a single name as author. the assistance and cooperation of a number of people deserves grateful acknowledgment. My wife, Barbara. deserves a great deal of loving thanks not only for her warm support. encourage- ment. and motivation when mine waned. but for doing a number of difficult and time consuming tasks despite her own very full schedule. I would also like to thank the chairman of my committee. Dr. Joseph Reyher. for his guidance. encour- agement. and for the speed and efficiency with which he worked to help me complete the study. Dr. Bert Karon deserves special thanks for his assistance with the statistical analysis. for the tremendous amount of personal time he gave up in helping to clarify the data analysis. and for additional incisive suggestions. ii Dr. Harry Grater and Dr. Paul Bakan made a number of extremely perceptive and helpful suggestions for improving the presentation of the study. For their efforts I would like to extend a sincere thank you. Phillip Snett spent a good deal of time rating drawings. a not too pleasant and somewhat difficult task which he nevertheless approached with interest and involvement. For this. and for his friendship. I extend my sincere thanks. James Clark deserves thanks for assistance in writing computer programs. and for his patient efforts to teach me some fundamentals of computer programming. Dr. Robert McMichael deserves a special thank— you for his understanding. and for knowing just how to motivate me when I was ready to quit. I'm grateful for this. and for his personal friendship. Dr. William Mueller deserves thanks for his valuable advice and assistance in data interpretation. and for his most appreciated and most welcomed sense of humor when things seemed discouraging. iii. To Jo Morse and Bill Kell I extend a very special thank-you for so many. many things. but most of all. for helping me to grow. iv TABLE OF CONTENTS Page Acknowledgments . . . . . . . . . . . . . . . . . ii List of Tables. . . . . . . . . . . . . . . . . . vii Method. . . . . . . . . . . . . . . . . . . . . . 5 GSR Measures. . . . . . . . . . . . . . . . . 7 Scoring . . . . . . . . . . . . . . . . . . . 8 Results . . . . . . . . . . . . . . . . . . . . . 10 Hypothesis I: Graphic Anxiety Indexes. . . . 10 Hypothesis I: GSR Data . . . . . . . . . . . 12 Hypothesis II: Graphic Anxiety Indexex . . . l3 Hypothesis II: GSR Data. . . . . . . . . . . 13 Hypothesis III: Graphic Anxiety Indexes. . . l4 Hypothesis III: GSR Data . . . . . . . . . . l9 Hypothesis IV . . . . . . . . . . . . . . . . 20 Hypothesis V. . . . . . . . . . . . . . . . . 23 Hypothesis VI . . . . . . . . . . . . . . . . 26 Page Discussion. . . . . . . . . . . . . . . . . . . . 27 Hypothesis I. . . . . . . . . . . . . . . . . 27 Hypothesis II . . . . . . . . . . . . . . . . 33 Hypothesis III. . . . . . . . . . . . . . . . 35 Hypothesis IV . . . . . . . . . . . . . . . . 38 Hypothesis V. . . . . . . . . . . . . . . . . 40 Hypothesis VI . . . . . . . . . . . . . . . . 44 References. . . . . . . . . . . . . . . . . . . . 48 Footnotes . . . . . . . . . . . . . . . . . . . . 51 Tables. . . . . . . . . . . . . . . . . . . . . . 52 vi LIST OF TABLES Table 1 Analysis of Variance for the Summed Anxiety Indexes. . . . . . . . . . . . . . 2 Results for Drawing Differences. Order Differences. and for Adaptation. for Each Of the 23 Indexes O O O O O I O O C O O O O 3 Analysis of Variance for GSR Frequency . . 4 Analysis of Variance for the Mean Conduc- tance scores 0 C O O O O O I O I O O O O O 5 Frequency of Scores Above and Below the Median for Body Areas. and Their Associ- ated Significance. . . . . . . . . . . . . 6 Pearson Product Moment Correlations Be- tween the 23 Indexes. GSR and Mean Con— ductance . O 0 O O O O O O O O O O O O O 0 vii Page 52 53 55 56 57 58 The Relationship Between GSR and Anxiety Indexes in Projective Drawings Leonard Handler Michigan State University Clinical experience continues to yield abundant support for the contention that the drawing of the human figure often symbolizes important aspects of the drawer's personality: however. the rules for determining the sym- bol-referent relations are as yet imperfectly understood. There exists a number of reservations concerning the pro- positions upon which figure drawing analysis rests. Swensen's (1957) review of the literature con- cludes that research does not support the validity of Machover's indexes of anxiety; however. a recent study by Handler and Reyher (1964) supports the validity of a number of anxiety indexes. The authors had 57 male college students draw a male. female. and an automobile 1 under stress and nonstress conditions. A previous study (Handler. 1963) indicated that there was no significant difference in task difficulty between drawing an auto- mobile and drawing a male. External stress was shown to increase manifestations of anxiety on the DAP; fif- teen of the 21 hypothesized indexes of anxiety signifi- cantly differentiated the stress from the nonstress drawings for the male figure. while eleven indexes sig- nificantly differentiated between the female stress and nonstress drawings. Only five of 17 indexes signifi- cantly differentiated between the stress and nonstress drawings for the automobile. thereby supporting the hypothesis of the relative neutrality of the automo— bile drawing. The decreasing frequency in the number of significant indexes for the male. female, and auto- mobile drawings was taken as evidence for a symbolic intrapsychic source of anxiety in addition to the ex- ternal stress situation. A number of indexes discussed by Swensen (1957) were significant. but in the opposite direction. That is. there was more shading. hair shad— ing. erasure. reinforcement. emphasis line. and placement in the upper left hand corner of the page in the non- stress situation. These opposite findings were explained in terms of adaptiveness. flexibility. and an appropriate reaction to a reality situation. The present study is an extension of the study by Handler and Reyher (1964); its purpose is to relate presumed graphic indexes of anxiety to a physiological measure of arousal and anxiety. The following hypothe- ses were investigated: I. Drawing a man and a woman is more anxiety producing than drawing an automdbile. because of the greater reflection of symbolic material. II. Drawing a woman is more anxiety producing. followed by drawing a man. and an automobile. It was thought that the relatively well adjusted male college student would have more active problems concerning heterosexual relationships than with homosexual or homoaffectional relationShips. III. a) There is a progressive diminution of anxiety. due to the desensitization of anxiety—pro- ducing preperties of the testing session. There is less anxiety associated with drawings done second. and third. compared with drawings done first. b) The anxiety-producing properties of a drawing vary as a function of the six possible orders of presentation. IV. Primary sexual characteristics are more anxiety producing than secondary sexual characteristics. which. in turn. are more anxiety producing than nonsexual body areas. V. There is a direct relationship between the number and degree of graphic indexes of anxiety. and the GSR. VI. The various indexes of anxiety differ in terms of the magnitude of the anxiety which they reflect. fromeost. to least effective. in the following order: 1. heavy pressure 2. heavy line 3. delineation line absence 4. erasure 5. light pressure 6. light line 7. small size 8. distortion 9. body simplification 10. head simplification ll. omission 12. shading 13. vertical imbalance 14. emphasis line 15. re- inforcement 16. placement (upper left) 17. line discontinuity 18. large size 19. small head 20. large head 21. transparency 22. hair shading 23. head:body ratio. This rank ordering is based in part on previous research (Handler & Reyher. 1964). and. in part. on a review of the literature. Method The sample consisted of 96 male college students. selected from introductory psychology classes. A Grass Model #5 Polygraph was employed to record the GSR data. The drawings were made on sheets of 8-1/2 x 11 unlined paper attached to a clipboard. which in turn was fastened to one of the arms of a chair. The subject was seated in a chair with double armrests. with his back to the polygraph. The electrodes were attached to the fingers of the nonpreferred hand. and at least fifteen minutes were then allowed for skin hydration under the electrodes. and for general adapta- tion. Yellow Springs Instrument Company electrodes and electrode paste were employed. During the adaptation period the subject was told that he would be asked to make a series of drawings on the first sheet of paper. and thereafter. he would be asked to make Separate drawings on separate sheets of paper. with one drawing per page. He was also told that a freshly sharpened pencil would be used for each drawing. and that the experimenter would supply a new pencil before each drawing was to be done. The polygraph was started as soon as the elec- trodes were attached. and operated continuously through- out the testing session. A GSR baseline was establiShed for each subject. The experimenter then said. "Let's begin!” The result was invariably a large reaction. and additional time was required for the subject's skin resistance to return as closely as possible to the orig- inal level. The subject was then told, "In a little while I'm going to ask you to write your first name." A short period of time was allowed for GSR activity to subside. and the subject was then told. "Write your first name." This same procedure was followed for the subject's middle and last names. In a similar manner the subject was requested to draw a circle. square. rectangle. diamond. star. pentagon. hexagon. and an octagon. After this. time was allowed for the subject's resistance level to.retUrn to the original prestimulus level. He was then told. ”In a little while I'm going to ask you to draw a man (or woman. or automobile)." The order of the drawing requests was counterbalanced. Thus. there were six possible orders. with 16 subjects in each of the six groups. As the subject drew. the experimenter recorded the areas drawn. directly on the polygraph paper. The result was a record of the reaction whidh corresponded to the drawing of a particular area of the body. or of the automobile. After each of the three drawings was completed. the subject's resistance level was again allowed to return as closely as possible to the original prestimulus level. GSR Measures There has been a great deal of controversy over the years concerning which of the many GSR measures is the most appropriate. and most reflects anxiety and tension (Block. 1962; Speisman. Osborn & Lazarus. 1961). Therefore. the measures below were chosen be- cause: 1. They are the measures that have been em- ployed frequently and successfully in the past. 2. The measures supplement each other in terms of the kinds of information they yield. The measures em- ployed were: 1. GSR Frequency (number of GSR's div- ided by the time taken for each drawing). 2. Mean Conductance (the mean of readings taken at the begin- ning. middle. and at the end of a drawing). The GSR frequency provides information about the anxiety- producing properties of specific body parts. The mean conductance measure yields information concerning the overall anxiety-producing pr0perties of each drawing. Scoring All drawings were first coded and were then scored for the 23 anxiety indexes. one index at a time. The drawings were scored for degree of presence or ab- Sence of each index, using a modified and extended version of the Hoyt-Baron scoring manual (Hoyt & Baron. 1959). This is essentially the same scoring manual used in a previous study (Handler & Reyher. 1964). Score values usually ranging from zero to three were assigned. Handler and Reyher (1964) reported that the percentage of agreement between two raters. scoring 90 drawings for degree of presence ofeach index. ranged from .67 to 1.0. with a model score of .87. As a further reliability check. the experimenter had an additional ‘rater. one who was unacquainted with the hypotheses of the study. score 33 drawings for degree of presence of each index. The rater scored 15 of the indexes. The remainder of the indexes were scored with mechanical devices.--e.g. a ruler was used for size and head size. and a grid was used to score placement. Reliabilities were therefore not computed for these measures. The percentage of agreement between therater and the ex- perimenter ranged from .67 to .97. with a median score of .87. The reliabilities. comparable to those reported by Hoyt and Baron (1959L Mogar (1962). and by Handler and Reyher (1964). demonstrate that the indexes can be scored with a high degree of reliability. 10 Results Hypothesis I: Graphic Anxiety Indexes The scores for the 23 indexes were summed for each subject. for each drawing. Hypothesis I was tested by comparing the summed scores for each of the three draw- ings. using an analysis of variance test for repeated measures on independent groups (Edwards. 1950). In this analysis. the test of significance for orders one to six is based upon independent. randomly assigned subjects. whereas the test of significance for drawings. and for the interaction between drawings and orders is based upon the same subjects. Thus. two separate error terms for the tests of significance are used: one for the possibly cor- related data obtained from the same subjects. and one for the data Obtained from the independent groups of subjects. Table I presents the results for the drawings. AnIE of 131.80 was Obtained. which is significant far beyond the .01 11 level of confidence. The results indicate that there is definitely a significant difference in the degree of anxiety reflected in the three drawings. Hypothesis I was thereby supported for the graphic indexes. Significance tests were also done for each of the 23 separate indexes. using the analysis of variance method of Link and Wallace (Mosteller & Bush. 1954). This test uses ranges as a measure of variation instead of sums of squares. A significance test was provided for drawings. and for orders. The results for each index are presented in Table 2. Only the table for Insert Table 2 about here the .05 level of significance was available; accordingly. only significance values of .05 or less were reported. Fifteen out of 18 indexes differentiated the three draw— ings beyond the .05 level of significance. The remain— ing five indexes referred only to the human figures. For the 15 indexes. heavy line. and heavy pressure had highest anxiety scores for the automobile. For the 12 remaining 12 indexes. the automobile drawings had the lowest anxiety scores. Hypothesis I: GSR Data The GSR data were analyzed with the same analy- sis of variance technique used for the graphic anxiety indexes analysis. Table 3 summarizes the results for the frequency measure. The 3 value of 20.0 for drawings was significant beyond the .01 level. Table 4 summar- izes the results for the mean conductance scores. Similar results were obtained for the conductance measure. Again. a significant F of 16.67 for drawings was obtained, indicating that the mean conductance for the three drawings was significantly different. The results for the frequency and conductance data thereby support Hypothesis I. l3 Hypothesis II: Graphic Anxietinndexes As hypothesized. an examination of the means for the three drawings indicated that the woman drawing had the highest anxiety scores. While the automobile had the lowest scores. The mean anxiety scores for the woman. man. and automobile drawings were 19.4, 18.8. and 12.9. respectively. Duncan's New Multiple Range Test (Edwards. 1960) was used to determine which of the three means differed significantly from each other. The results indicated that both the man and woman drawings were significantly different from the automobile drawing. beyond the .001 level of signifi- cance; however. the man drawing only differed from the woman drawing at the .10 level of significance. Hypothesis II: GSR Data The woman drawing had the highest GSR frequency and the highest mean conductance. while the automObile had the lowest values for both of these measures. Dun- can's New Multiple Range Test (Edwards. 1960) was used to determine which of the three means differed signifi- cantly from each other. For the frequency measure the 14 results indicated that the man. woman. and automobile drawings*were all significantly different from each other at the .001 level of confidence. For the con- ductance measure the man drawing was significantly dif- ferent from the automobile drawing at the .005 level. while the woman drawing was significantly different from the automobile drawing at the .001 level. The conductance for the man drawing was significantly lower than the conductance for the woman drawing at the .01 level. The results for the frequency and conductance measures thereby support Hypothesis II. Hypophesis III: Graphic Anxiety Indexes ‘nggp, The E of .790 for orders was nonsignifi- cant. indicating that there were no significant differ- ences between the six orders of presentation. Separate analyses were also done for each of the separate indexes. using the LinkAWallace test. In the LinkAWallace test for order. there are six possible orders being compared (MAW. MWA. WAM. WMA. AWM. AMW). one with another. Thus. it is possible 15 to determine which. if any. of the six orders within gggh index is significantly different from the others. in terms of the degree of anxiety reflected. The results are presented in Table 2. Five of the graphic anxiety indexes (heavy pressure. balance. erasure. head:body ratio. and small head size) were significant for differences between the six orders of presentation. For the significant orders within each of these five indexes it makes a difference. in terms of the degree of anxiety present. as to the sequence in which the three drawings are executed. For these five indexes there were 21 separate order differences out of a possible 75. For example. for the index "bal- ance." order WAM yielded significantly higher anxiety scores than order AWM. For the 21 significant order differences. the number of times a specific order yielded higher anxiety scores (when compared with the scores for another pos- sible order) was compared with the number of times that order yielded lower scores. using a two tailed sign test. None of the differences were significant. except 16 for orders WAM and AWM. Order WAM had significantly more high anxiety scores than low anxiety scores (p= .002). while order AWM had significantly more low scores than high scores (p=.008). 0f the orders which had the higher anxiety scores. 12 of the 21 had woman as the first drawing. eight had man as the first drawing. while only one had automobile as the first drawing. The corresponding values for the lower anxiety scores were eight. seven. and six. respectively. A chi square of 9.1 for these values is significant at beyond the .02 level. The results for the above two analyses tend to indicate that for the five indexes sensitive to differ- ences between the six orders. the highest level of an- xiety occurs when the woman drawing is first. while the lowest level of anxiety occurs when the automdbile drawing is first. Thus. while the hypothesis for order (IIIb) was not supported for the summed graphic anxiety indexes. it was supported in part for five of the in- dividual indexes. Adaptation. The drawing x order interaction obtained in the analysis of variance for the summed 17 graphic anxiety indexes was used to test for adaptation. Table I summarizes the results of this analysis. The drawing x order interaction indicates whether or not there is any difference in the degree of anxiety repre- sented in a drawing when it is done either first. second. or third. An.§ of 1.36 was obtained. which is nonsignifi- cant. Thus. the hypothesis for adaptation (IIIa) was not supported for the summed graphic anxiety indexes. How— ever. an examination of the scores for each index sug- gested that for some indexes the three orders (first. second. and third) were associated with an increase in the amount of anxiety. while for other indexes there was a decrease for order. Thus. in computing a pooled anxiety score for each subject these opposite effects may very well have cancelled each other out. It was therefore decided to investigate adaptation effects separately for each index. To do this. the data were recast into a 3x3 table. where the rows. as before. referred to drawings (man, woman. or automobile). but the columns now referred to the order in which the drawing was done (first. second. or third). With the 18 data set up in this way. it is possible to investigate. for each of the indexes. whether there is any difference in the amount of anxiety when a drawing was done either first. second. or third. The results may be seen in Table 2. It should be noted that the general conception of adaptation can mean either an increase or a decrease. depending upon the particular function being discussed. Seven indexes yielded significant differences for the effects of adaptation (progressive diminution or reduction in the level of anxiety when a drawing is done first. second.or third). These indexes are: erasure. emphasis line. shading. reinforcement. large head size. head simplification. and hair shading. As Table 2 indicates. five of these indexes showed an in- crease as a function of whether the drawings were done first. second. or third. while one index showed a de- crease.1 Hair shading showed an increase for order two. followed by a decrease for order three. The re- sults for the individual indexes therefore support the hypothesis concerning adaptation (IIIa). 19 Hypothesis III: GSR Data nggp. Table 3 presents the results for the frequency data. while Table 4 presents the results for the conductance measure. An‘g of 1.62 was obtained for the frequency data. and and E of .591 was obtained for the conductance measure. Both values are nonsignif- icant. indicating that there is no difference in GSR frequency. or in mean conductance level for the six different orders of presentation. The hypothesis for order (IIIb) was therefore not supported for the two GSR measures. Adaptation. The drawing x order interactions obtained in the analysis of variance for GSR frequency. and for mean conductance were used to test for adapta- tion. Table 3 summarizes the results for frequency. while Table 4 summarizes the results for conductance. An‘g of 4.0 was obtained for frequency, which is signif- icant beyond the .05 level. An.§ of 9.22 was obtained for conductance. which is significant beyond the .01 level. Greater GSR frequency. and higher conductance levels (both indicating increased arousal). were 20 associated with the drawing done first. while lower conductance levels and decreased GSR frequency were associated with drawings done second. and with draw- ings done third. Thus. the hypothesis for adaptation (IIIa) was supported for the two GSR measures. Hypothesis IV The median GSR amplitude was obtained for each body part. for each man and each woman drawing. The number of times the GSR for each body part fell above or below the median was then determined. The chi square one sample test was used to test the significance of the differences for the various body areas. Table 5 lists the body parts for the man and woman drawings. with their associated levels of significance. For the man drawing. three body parts (shoulders. chest. and hips had significantly more GSR's above the median. indicating increased arousal. Another body area (feet) had significantly more GSR's below the median. 21 For the woman drawing. eight body parts (genital area, breast area, waist, skirt, hips. hair. body contour. and arms) had significantly more GSR's above the median. The body parts were then classified as either sexual, secondary sexual, or nonsexual. The basis for categorization was mainly in terms of anatomical and physiological criteria. For the man drawing the sexual body area was defined as that area within the waist. hip. and groin region. Thus. the body parts included in this category were: genital area. waist. and hips. The secondary sexual area included the following body parts: beard. chest hair, shoulders. chest. and hair. The nonsexual body area included: arms. neck, facial features. hands, fingers. feet. and toes. For the woman drawing the sexual body area was defined as for the man drawing. with the exception of body parts included in the breast area. Thus. for the woman drawing the sexual body area included: the geni- tal area. waist. skirt area, hips. and breast area. The secondary sexual body parts included the hair. the body contour. and the legs. The nonsexual area ’22 included the shoulders, arms, neck. facial features. hands, fingers, feet, and toes. The number of subjects the preponderance of whose GSR's to sexual, secondary sexual, and nonsexual body areas were above and below the median was tabulated. The percentage of subjects above and below the median for the man drawing was as follows: sexual, 66% above. and 34% below the median; secondary sexual, 71% above, and 28% below the median; nonsexual. 55%.above, and 45% below the median. For the woman drawing, 82% 0f the subjects gave GSR's above the median for sexual body areas, and 18% gave GSR's below the median. Seventy- five percent of the cases were above the median, and 25% below the median for the secondary sexual body areas. The corresponding values for the nonsexual areas were 58% and 42%,respective1y. A chi square for single samples was used to test the significance of the scores above and below the median. The results for the man drawing indicated that there was no significant difference between the frequencies above and below the median for the nonsexual body parts (chi 23 square?.74, p=between .05 and .3). However, signifid cantly more subjects gave GSR's above the median than below the median for the sexual area (chi square=6.4, p=between .01 and .001), and for the secondary sexual body area (chi square=ll.60, p=<.001). Thus, for the man drawing the secondary sexual body areas seem most anxiety-arousing, followed by the sexual body area. and lastly, the nonsexual body areas. Similar results were obtained for the woman drawing, except that the sexual body areas were asso- ciated with the most anxiety (chi square=30.4, p=<.001). followed by the secondary sexual body areas (chi square =13.2, p=<.001). As with the man drawing, there was no significant difference between the frequency above and below the median for the nonsexual body parts (chi square =1.77. p=between .2 and .1). Thus, Hypothesis IV was supported for the woman drawing, and to a lesser degree. for the man drawing. Hypothesis V Pearson Product Moment Correlations were run be- tween the summed graphic anxiety indexes scores for the 24 96 subjects and the frequency and conductance measures, for the man, woman, and automobile drawings. The re- sults. presented in Table 6. indicate that the corre- lations are not essentially different from zero. The correlation between the graphic indexes and the conductance measure for the automobile was .18, which approaches significance (p=.08). Thus, using the summed indexes, Hypothesis V is not supported for the man and woman drawings, but there is a small indication of sup— port for Hypothesis V for the automobile drawing. Separate Product Moment Correlations were done for each index and the two GSR measures. since it is possible that pooling the 23 graphic indexes obscured relationships between individual anxiety indexes and the GSR variables. The results are presented in Table 6. All of the resulting correlations were rather low. but 12 correlations were significant at the .05 level and beyond. while three were significant at beyond 25 the .01 level. For the man drawing, shading, erasure, reinforcement, emphasis line, and distortion correlated with GSR frequency. Four of these indexes (shading. erasure, reinforcement. and emphasis line) correlated significantly and negatively with GSR frequency. This finding is similar to Handler and Reyher's (1964) find- ings for shading, erasure, reinforcement, and emphasis line. There were no significant correlations for any of the indexes and the conductance measure for the man drawing. For the woman drawing significant correlations were obtained for the frequency measure and reinforce- ment, heavy line, emphasis line, body simplification. and light pressure. Emphasis line. light pressure. and reinforcement were significantly and negatively correlated with GSR frequency. Shading. hair shading. and erasure were also negatively correlated with GSR frequency for the woman drawing, but the correlations were nonsignificant. Two indexes (line discontinuity. and emphasis line), correlated significantly with the conductance measure for the woman drawing. For the automobile drawing only placement correlated with GSR 26 frequency, while line discontinuity and emphasis line were significantly correlated with the conductance measure. Thus. the results for the separate indexes tend to support Hypothesis V. The GSR frequency correlations for shading. hair shading, erasure, reinforcement, and emphasis line were consistently negatively correlated, indicat- ing that presence of these indexes is associated with significantly fewer GSR's. Hypothesis VI The correlation for each index with each of the GSR measures may be seen in Table 6. Rank order correlations were done for the obtained correlations between each of the two GSR variables and the rank order suggested in Hypothesis VI. Separate correla- tions were done for each drawing. The obtained cor— relations for the man drawing were .20 for GSR fre- quency, and -.08 for the conductance measure. The respective correlations for the woman drawing were .25 and .17, respectively. and those for the automo- bile drawing were .20 and .07, respectively. None 27 of these correlations is significantly different from zero. These results indicate that there is essentially a zero relationship between the hypothesized rank order of the anxiety indexes and the obtained rank order for GSR frequency, and for mean conductance. Hypothesis VI was therefore not supported. Discugsion Hypothesis I The finding that more anxiety is associated with the man and woman drawings than with the automo- bile drawing is in agreement with Handler and Reyher's (1964) findings. They found that the figures reflected more graphic anxiety signs than did the automobile drawing. These results suggest that as a more neutral stimulus. the automobile is a useful control figure in clinical practice, as was indicated by Handler and Reyher (1964), and Reyher (1959). The results for the separate indexes are also similar to Handler and Reyher's (1964) findings. The 28 major difference is more apparent than real. That is. in the results obtained for Hypothesis I (see Table 2), four of the five indexes that were significant in the Opposite direction (present in the nonstress drawings more frequently) in the previous study were nonsignifi— cant in the present study. These indexes were: shading. erasure, reinforcement, and emphasis line. A possible explanation for the diverse findings in the two studies lies th procedural differences for the two experiments. In the former study, drawings were obtained both under stress and nonstress conditions, and each stress draw- ing was compared with its nonstress counterpart. In the present study. the drawings were obtained pply under stress conditions. and the man, woman. and auto- mobile drawings were compared with each other. Thus. it is possible that in the present study stress washed out differences between the three drawings for shading. erasure, reinforcement, and emphasis line. This con- clusion may be especially true if these above indexes are more sensitive to external stress than to intra- psychic stress. 29 To determine whether the external stress ac- tually did wash out differences between the three draw- ings for erasure. shading, reinforcement. and emphasis line, the scores for these indexes were compared with the stress and nonstress scores in the previous study (Handler & Reyher. 1964). using the Link-Wallace method of analysis of variance. For erasure and emphasis line. the scores for drawings in the present study, and the scores for the stress situation in the previous study were significantly different from the nonstress scores of the previous study at beyond the .05 level of confi- dence. The scores for erasure, and emphasis line in the present study were not significantly different from the stress drawings in the former study. For shading. the corresponding scores were significantly different from each other, beyond the .05 level, although the drawings of the present study were more similar to the stress drawings of the former study than they were to the nonstress drawings. There were no significant differences for reinforcement. These findings suggest that for erasure. emphasis line. and to some extent 3O shading, the hypothesis concerning the washing out of differences,.mentioned above. is correct. Handler and Reyher (1964) found that there was more shading, hair shading. erasure, emphasis line, and reinforcement in the nonstress rather than in the stress drawings. They suggest that these indexes might reflect an adaptive and flexible response. rather than indicating anxiety. The present findings for adapta- tion are consistent with this argument. Table 2 indi- cates that erasure. shading, emphasis line, reinforce- ment. hair shading. large head size, and head simplifi- cation showed adaptation, and furthermore. that with adaptation there was increased rather than decreased presence of these indexes, for all but head simplifica- tion. The consistent negative correlations obtained for shading, hair shading, erasure, reinforcement. and emphasis line indicate the same trend. Absence rather than presence of these indexes correlates nega- tively with GSR frequency. It is very likely that these indexes are more sensitive to external stress (albeit in the opposite direction) than to intrapsychic stress. 31 Thus, they may be less reliable indexes of sympolic conflict than those indexes which differentiated be- tween the three drawings. In addition, it is also possible that those indexes which differentiated all three nonstress drawings from their stress counterparts in the study by Handler and Reyher (1964), and differ- entiated between the three drawings in the present study. reflect pppp intraphsychic and external stress. These indexes were: line pressure increase. delineation line absence, and body simplification. This conclusion is based on the fact that in the present study the above three indexes differentiated between the three drawings despite the external stress, and in the previous study they differentiated between stress and nonstress for the three drawings. In the present study, the automobile had the lowest anxiety scores for all but two of the signifi- cant indexes (heavy 1ine, and heavy pressure), which is as it should be. since the automobile has been shown to be a more neutral control figure. For heavy line. and line pressure increase the automobile had signifi- cantly higher scores than did the figures. At first 32 glance these findings suggest that heavy line and heavy pressure denote the absence of anxiety. However, another interpretation is that heavy line and heavy pressure re- flect external stress to a greater degree than they re- flect intrapsychic stress. On the other hand. light line and light pressure both differentiated between the three drawings, but these indexes are present in the figures significantly more than in the automobile drawing. Thus, a differential response to stress is suggested for these four indexes, light line and light pressure reflecting intrapsychic stress, while heavy line and heavy pressure seem to reflect external stress. Similar results for light and heavy line were found in the previous study. The data are also consistent with the results of Reznikoff and Nicholas (1958). and with those of Gutman (1952). Reznikoff and Nicholas (1958) found that heavy line significantly differentiated the drawings of paranoid schizophrenics from undifferenti- ated schizophrenics. A paranoid typically views the world about him as a source of his fears, rather than recognizing his own intrapsychic conflicts. Gutman (1952) found that patients who did not improve in therapy tended 33 to draw continuous and reinforced lines, but those who improved tended to draw light, sketchy lines. This is consistent with the widely shared view among therapists that patients who experience anxiety about themselves improve more quickly than those who see their problems as emanating from an outside source. The above discussion concerning intrapsychic and external sources of stress suggests that the pres- ence of anxiety indexes in a drawing cannot always be used to measure intrapsychic stress. Instead, the pres- ence of shading, erasure, emphasis line, reinforcement. heavy line. heavy pressure, and to a lesser extent de- lineation line absence and body simplification, may be a reaction to an external source of stress. Hypothesis II Significant results for Hypothesis II indicates that the female drawing, drawn by a male is fraught with more anxiety than is either the man or the automdbile drawing, and that the man drawing is more anxiety produc- ing than the automdbile drawing. Additional data for sup- port of Hypothesis II is indicated in the results for 34 order for the graphic indexes. For the indexes sensitive to differences between the six orders, the highest level of anxiety occurred when the woman was drawn first, while the lowest level of anxiety occurred when the automobile drawing was first. The finding that the woman drawing yielded the highest level of anxiety is important in establish- ing a baseline for clinical interpretation in a diag- nostic setting, since it seems that the presence of anxiety indexes in a woman drawing drawn by a male might be a natural state of affairs. Thus, it may be erroneous for a clinician to weave elaborate hypotheses about a patient's relationships with, and feelings to- ward women on the basis of anxiety indexes in the female drawing unless he takes into account the fact that nor- mal males, too. have a certain degree of anxiety asso- ciated with their drawings of a woman. It is possible that older males, whose natural adolescent problem con- cerning masculinity and sex have been dealt with and solved will not show such anxiety indexes in their drawings of a woman. Additional research along these 35 lines is indicated. It is conceivable that DAP norms can be established for a variety of age groups, and for both sexes, much like norms on the various intelli- gence scales. Hypothesis III It is not very surprising to find adaptation with the GSR, for it is a well known and well documented phenomenon (Woodworth & Schlosberg, 1954). Such adapta- tion effects seem to be another indication that the GSR measures are probably tapping reactions to external stress. However, it is quite surprising to find adapta- tion effects for seven of the individual graphic anxiety indexes. Such findings indicate that shading, hair shading, erasure, reinforcement, emphasis line, head simplification. and large head size may be unreliable indexes of anxiety unless adaptation effects are taken into account. If these indexes are used to infer an- xiety. the order of drawing must be taken into account. since the first drawing might differ from the second and third in degree of anxiety reflected merely because 36 it was drawn first. For example, if a clinician obtains two or more drawings from a patient, and the second draw- ing has a good deal more shading. erasure, reinforcement. and emphasis line than the first drawing, it would be erroneous to conclude that the content of the second drawing was associated with more anxiety for the patient than the first drawing. Such results would be expected on the basis of adaptation, and for these indexes, adapt— ation is associated with increased rather than decreased presence. Similarly, the second drawing might show less head simplification, and a smaller size head than the first drawing, since for these two indexes adaptation is manifested by decreased rather than increased presence. The findings for adaptation seem to indicate that the subject or patient is adjusting to a new anxiety pro- voking task. or to the stresses of the testing situation. Thus. differences in the pattern of the adaptation effect can often be diagnostically significant. That is, it may be possible for the clinician to use the adaptation effect as a measure of whether the patient is beginning to relax and as a measure of whether or not rapport has 37 been established. If a patient's drawings do not show adaptation effects, the examiner might do well to specu— late about his interaction with the patient, and about the quality of rapport established. If there seems to be no problems in this area, one can then be more cer- tain that it is the task of drawing itself which is the cause of the anxiety. The particular content of the second or third drawing might be associated with more conflict than was the first drawing. The use of the automobile as a control figure, as suggested by Handler and Reyher (1964), would be valuable here in sorting out the anxiety responses to external stress and associated adaptation from intrapsychic stress associated with the content of the drawing. The six different orders of presentation do not appear to influence degree of anxiety, as measured by the two GSR variables. There does seem to be some effect of order for the graphic indexes. but these differences seem to depend mainly upon whether the woman drawing is done first, or whether the automobile is done first. 38 Hypothesis IV The results for Hypothesis IV are consistent with observations of adolescents, and with what is known about the sexual problems associated with maturation. The results indicate that although adolescent males are concerned with both sexual and secondary sexual signs. they are more consciously concerned with secon- dary sexual signs,—-signs that are more public and transmittable evidence of masculinity, sexuality, and adequacy, than they are with male primary sexuality. per se. Mussen, Conger. and Kagan (1963) note that for the adolescent, possession of desirable physical characteristics is associated with physical attractive— ness. They emphasize that members of the Western cul- ture equate certain anatomical attributes with strength or weakness, and with sexual attractiveness or unattrac- tiveness, and that the adolescent equates his ability to establish satisfying heterosexual relationships with aspects of his physique. Mussen, Conger, and Kagan (1963) emphasize that the American boy is concerned 39 with his height, breadth of his chest and shoulders. muscular development, and amount of facial and bodily hair. When adolescent boys were asked about the kinds of bodily changes they wanted, they wished for broader shoulders, more rugged builds. large chests and facial and body hair (Frazier & Lisonbee, 1950). When a male draws a female. however, it is the sexual areas which are more anxiety arousing ra- ther than the secondary sexual areas. although the secondary sexual areas are also highly anxiety produc— ing. Here again it seems that the relative newness of the sexual drive is an anxiety arousing problem for the adolescent male. Moreover. he is more inter- ested in her primary sexual characteristics. and more troubled over the problems and fears accompanying this interest. For many subjects there seemed to be specific sexual or secondary sexual body areas associated with greater GSR amplitude. Perhaps this is analogous to the colloquial conception of "a breast man." "a leg man." etc. It may well be that for each male, different 40 secondary sexual body areas of the female represent sexuality to a greater or to a lesser degree. The demonstration of such variability from person to person may be viewed by some as indicating that the DAP is too capricious and unreliable. How— ever, the results for Hypotheses I, II, III, and IV indicate that there is a great deal of consistency among subjects despite the variability. More impor- tant. such variability is the essence of what is meant by "projection" in the context of projective testing. Hypothesis V The lack of correlation between the summed anxiety indexes and the GSR measures seems to be due to pooling the indexes which correlate negatively with the GSR. together with those indexes which correlate positively with the GSR measures. The fact that the correlations for the individual indexes are low may very well be due to the fact that the GSR measures seem to be tapping external stress moreso than intra- psychic stress, while the graphic indexes probably tap 41 both external stress and intrapsychic conflicts on a symbolic level. Others (Mordkoff, 1964; Speisman. Lazarus. Davison, & Mordkoff, 1964) have found rather high correlations between GSR variables and stress. The stress situations described by these authors seem to a great extent external rather than intrapsychic. Preliminary findings by Reyher also bear upon this point. Using the technique of Free Imagery while con- tinuous GSR's were being recorded, Reyher noted that there were few GSR's when the material was unconsciously but.symbolically expressed.2 In another.study involving hypnotically implanted conflicts, more direct recognition of the conflict in the posthypnotic state resulted in increased GSR frequency. The tendency toward significance for the correla— tion between the automobile and the conductance measure must be seen in light of the results for Hypotheses I and II. That is, for the summed graphic anxiety indexes. and for the two GSR measures the figures had significantly higher scores. indicating greater anxiety and conflict. while the automobile drawing had the lowest scores. 42 indicating the least amount of anxiety. Thus, the correlation between the graphic indexes and the con- ductance measure indicates that they both reflect a generally lowered level of anxiety. This finding is consistent with the rest of the data, and seems to indicate the failure of the conductance measure to reflect anxiety. The significant negative correlations obtained for the GSR frequency and shading, hair shading, era— sure, reinforcement, and emphasis line are in agreement with the results for Hypothesis I. Again. it seems that the presence of these indexes is associated with significantly fewer GSR's. Increased distortion, increase in omission. and increased placement in the upper left hand corner are associated with increased GSR frequency for the man drawing. They are very likely good measures of external stress if one takes the viewpoint that GSR frequency reflects external stress to a greater degree than it reflects symbolic, intrapsychic stress. The same may be said of heavy line and body simplification 43 for the woman drawing. The significant negative corre- lation between light pressure and GSR frequency for the woman drawing is inconsistent with the rest of the data. Therefore. it was decided to compare the rankings of the correlations obtained for the man draw- jxm; and the frequency measure with the correlations- for the woman drawing with the frequency measure. A rank order correlation of .66 was obtained, which in- dicates that the overall results for the man and woman drawings are rather similar. This correlation indicates that the rankings for each of the indexes is similar for both drawings, and that the results for the 23 indexes can be generalized to both drawings. As the results of this investigation show, con- ductance and GSR frequency are equally good measures when comparing a stresser operating over time--i.e., comparing GSR frequency and conductance for each of the three drawings. However, Table 6 shows that GSR frequency enters into more significant relationships than conductance. This finding indicates the superi- ority of the GSR frequency measure in reflecting more 44 specific sources of anxiety, for a more differential analysis. Hypothggis VI The attempt to predict the best anxiety indexes. in terms of rank order, was a failure, in part because of the obtained negative correlations. In terms of postdiction, however. everything is not lost. It seems that shading, hair shading, erasure, reinforcement, and emphasis line are good anxiety indexes. as are distor- tion, body simplification, and heavy line. and to a lesser extent, placement, and omission. These indexes seem valuable for predicting external stress rather than for predicting intrapsychic stress, however. If one takes the view that the GSR primarily taps external stress, then the data for Hypothesis VI yields no in- formation concerning the best indexes of intrapsychic stress. The fact that shading. erasure, reinforcement. and emphasis line did not significantly differentiate the automobile from the figure drawings. and the fact that the differences between the three drawings were 45 washed out by the stress indicates that these indexes are poor predictors of intrapsychic stress. On the other hand. those indexes which differentiated between the three drawings despite the external stress situation (see Table 2) should prove to be good indexes of intra- psychic stress. Some additional research along these lines seems indicated. However. the results in Table 2 seem in agreement with a review of the literature con- cerning DAP anxiety indexes (Handler, 1963). When the significant results for the indexes in Table 2 were com- pared with the summarized results for 48 studies. only five of the 15 significant indexes (light line, hair shading, placement, transparency, and head:body ratio) yielded results which were inconsistent with the review of the literature. That is, significant results in the present study agreed with other findings for these in- dexes. There were relatively few studies, for these 15 significant indexes, Which yielded nonsignificant results. Only two out of eight indexes which yielded nonsignificant results in the present study (head sim— plification. and large size) were inconsistent with 46 the review of the literature. The results for the other six indexes (shading. erasure, reinforcement, emphasis line, small head size, and large head size) were essentially in agreement with the summarized re- view of the literature. It should be emphasized that although this study has dealt with a list of anxiety indexes, the author does not advocate a one to one sign approach, or any figure drawing analysis approach which attempts to reduce analysis to a simple formula. Such an approach seems quite empty and misleading. However, this does not mean that one must swing in the Opposite direction and resort solely to an intuitive, impressionistic approach. Rather, the ideal approach seems to be one which allows for the full use of a clinician's skill. based on intuition, hunch, or what have you, within a scientific framework of controls, and checks and balances. There is no harm in generating many hypo- theses about a patient from his drawings, just so long as these are logically and systematically checked out. 47 The omnipresent split in clinical psychology between "the artist," and ”the scientist" seems nowhere to make less sense than in the task of figure drawing analysis. References Block, J. Measurement dimensions in a palmer resistance situation. Psychol. Rep., 1962, 11, 319-331. Edwards, A. Experimental design in psychological re- search. New York: Rinehart, 1950. Edwards, A. Experimental design in psychological research. (Revised ed.) New York: Rinehart, 1960. Frazier, A. & Lisonbee, L. Adolescent concerns with physique. Sch. Rev., 1950, 58, 397-405. Gutman, Brigette. An investigation of the applicability of the human figure drawing in predicting im— provement in therapy. Unpublished doctoral dissertation, New York Univer.. 1952. Handler, L. The effects of stress on the Draw A Person Test. Master's thesis, Michigan State Univer— sity, 1963. Handler, L. & Reyher, J. The effects of stress on the Draw A Person Test. J. consult. Psychol., 1964, 28, 259-264. 48 49 Hoyt, T. & Baron, M. Anxiety indices in same sex draw- ings of psychiatric patients with high and low MAS scores. J. consult. Psychol., 1959, 23, 448-452. Mogar, R. Anxiety indices in human figure drawings: A replication and extension (extended report). J. consult. Psychol., 1962, 26, 108. Mordkoff, A. The relationship between psychological and physiological responses to stress. psy- chosom. Med., 1964, 26, 135-150. Mosteller, F. & Bush, R. Selected quantitative tech- niques. In G. Lindzey (Ed.), Handbook of so- cial pgygholoqy. Vol. I. Reading, Mass.: Wesley, 1954. Pp. 289-334. Mussen, P., Conger, J., & Kagan. J. Child development ppg personalipy. New York: Harper & Row, 1963. ReYher, J. Use of figure drawings in differential diag- nosis. Paper read at the State Diagnosticians' Conference, Michigan, 1959. Reznikoff, M. & Nicholas, Alma. An evaluation of human figure drawing indications of paranoid pathology. J. consult. ngchol., 1958. 22, 395-397. 50 Speisman. J., Osborn. Janet, & Lazarus. R. Cluster analyses of skin resistance and heart rate at rest and under stress. psychosom. Med., 1961, 23, 326-343. Speisman. J., Lazarus. R., Davison, L., & Mordkoff, A. Experimental analysis of a film used as a threat- ening stimulus. J. consult. Psychol., 1964, 28, 23-33. Swensen, C. Empirical evaluations of human figure draw- ings. .ggychol. Bull., 1957. 54, 431—466. Woodworth, R. & Schlosberg, H. Experimental psychgloqy. New York: Holt. Rinehart, & Winston, 1954. Footnotes 1While an increase with order hardly looks like adaptation, data presented by Handler and Reyher (1964) indicate that the increased presence of some anxiety indexes demonstrates lowered anxiety and adap- tation to stress. This point will be discussed more fully later in this paper. 2J. Reyher, Personal communication. Michigan State University, 1964. 3J. Reyher, Personal communication. Michigan State University, 1964. 51 52 Table 1 Analysis of Variance for the Summed Anxiety Indexes Source SS df MS F Between Orders 141.17 5 28.23 .790 Between Subjects 3213.71 90 35.71 Total Between (95) Between Drawings 2550.58 2 1275.19 131.80* Drawings X Order 96.75 10 9.68 1.36 Subjects X Drawings 1276.67 180 7.09 Total Within (192) Total (287) W * p<.01. 53 Table 2 Results for Drawing Differences. Order Differences. and for Adaptation, for EaCh of the 23 Indexes a Anxiety Drawing a Order Adaption Index Signif. Direction Signif. Signif. Heavy Line i Yes A most .No No Light Line Yes A least No No Heavy Pressure Yes A most Yes No Light Pressure Yes A least No No . b Shading No No Yes Hair Shading Yes W more No Yes b Erasure No Yes Yes Placement Yes A least No No . b Reinforcement No No Yes . . b Empha51$ Line No No Yes Vertical Imbalance Yes A least Yes No Delineation Line Absence Yes A least No No Line Discontinuity Yes A least No No Transparency Yes A least No No Omission Yes A least No No (Table continued on next page) 54 Table 2 (continued) W Anxiety Drawing Order Adaption . . . . a . . . . Index Signif. Direction Signif. Signif. Distortion Yes A least No No Head c Simplification No No Yes Body Simplification Yes A least No No Size (Small) Yes A least No No Size (Large) No No No Head Size (Small) No Yes No . b,c Head Size (Large) No No Yes Head:Body _ Ratio Yes W most Yes No vf‘ v v— vfiVv—v— v v Note.--All significance values are reported as .05 or less. aW = woman; A = automobile. b . Increase WIth order. c . Decrease Wlth order. 55 Table 3 Analysis of Variance for GSR Frequency Source SS df MS F Between Orders .0354 5 .0071 1.62 Between Subjects .3920 90 .0044 Total Between (95) Between Drawings .0240 2 .0120 20.0** Drawings X Order .0243 10 .0024 4.0* Subjects X Drawings .1158 180 .0006 Total Within (192) Total (287) *p<.05. **p<.01. Analysis of Variance for the Mean Conductance Scores 56 Table 4 Source SS df MS F Between Orders 318 5 63.6 .591 Between Subjects 9686 90 107.6 Total Between (95) Between Drawings 60 2 30.0 16.67* Drawings X Order 166 10 16.6 9.22* Subjects X Drawings 317 180 1.8 Total Within (192) Total (287) *p