12
KINETIC FAMILY DRAWING SCORING METHOD FOR CROSS-CULTURAL STUDIES PATRICIA WEGMANN, MD, and VIJA B. LUSEBRINK, PhD, ATR-BC* Introduction Culture shapes the psychology of individuals. Emotions are probably similar for all humans, how- ever, the expression of emotions is indeed culturally mediated. Most of the techniques used in psychother- apy and for assessment were developed by Americans or Europeans. More and more often, therapists are confronted with people from different cultures and feel the need to adapt their techniques to fit the cul- tural background of their patients. Research is needed to examine the influence of different cultures on psy- chological techniques and to verify the validity of assessment tests used on patients from these cultures. Family patterns and social interactions vary widely from one culture to another. Kinetic Family Drawing (KFD; Burns & Kaufman, 1970), a widely used pro- jective drawing assessment test to measure family dynamics, is a promising test for cultural studies. It was developed in the United States by two American authors, Burns and Kaufman, who empirically de- fined the criteria of analysis over several years. In their detailed review of the literature on the KFD technique, Handler and Habenicht (1994) listed stud- ies that investigate cultural norms. Many clinicians appear to use the KFD test with non-Caucasian pop- ulations without considering the impact of the child’s culture on the drawing (Cabacungan, 1985; Fukada, 1990; Ledesma, 1979; McNight-Taylor, 1974). Some authors compared the KFD test with other psycholog- ical tests to verify its validity for non-Caucasian- American children (Cho, 1987; Shaw, 1989). These authors made the assumption that the other psycho- logical tests used are valid for non-Caucasian-Amer- ican populations. Research that investigates the valid- ity of the KFD test with non-American children is rare. A few studies compare American-Caucasian KFDs with those drawn by children from other cul- tures; and they usually conclude that there are cultural influences on the drawing results. Chuah (1992) com- pared KFDs of Chinese-American families with those of Caucasian-American families. She concluded that KFDs illustrate cultural influences, and that KFDs are probably a useful tool to investigate the acculturation process (i.e., the assimilation of the American culture by the Chinese-American families). Nuttall, Chieh, and Nuttall (1988) compared KFDs of Chinese chil- dren from Beijing with those of American children. They concluded that the drawings reflect the social values and norms of the two cultures. Urrabazo (1986) compared KFDs of Hispanic children with Burns and Kaufman’s normative sample and ascer- tained differences in the activities of the parental fig- ures. Chartouni (1992) compared KFDs of American- Lebanese children with KFDs of American-Caucasian children. Chartouni concluded that the KFD test is a useful tool to show cultural differences in family life- Patricia Wegmann is a psychiatrist and psychotherapist in private practice (CH-1052 Le Mont, Switzerland), who is completing work on her medical PhD at Geneva University, Switzerland. Vija B. Lusebrink is Professor Emerita and resides in Palo Alto, California. The authors wish to thank Franc ¸ois Ferrero, Stephan Morgenthaler, Alain Wegmann, Holly Cogliati, Janis Yoshikawa and Sue Parker for their assistance on this study. The Arts in Psychotherapy, Vol. 27, No. 3, pp. 179 –190, 2000 Copyright © 2000 Elsevier Science Ltd. Printed in the USA. All rights reserved 0197-4556/00/$–see front matter PII S0197-4556(00)00051-4 179

KFD Scoring Method

  • Upload
    dorka9

  • View
    391

  • Download
    4

Embed Size (px)

DESCRIPTION

KFD Scoring Method

Citation preview

Page 1: KFD Scoring Method

KINETIC FAMILY DRAWING SCORING METHOD

FOR CROSS-CULTURAL STUDIES

PATRICIA WEGMANN, MD, and VIJA B. LUSEBRINK, PhD, ATR-BC*

Introduction

Culture shapes the psychology of individuals.Emotions are probably similar for all humans, how-ever, the expression of emotions is indeed culturallymediated. Most of the techniques used in psychother-apy and for assessment were developed by Americansor Europeans. More and more often, therapists areconfronted with people from different cultures andfeel the need to adapt their techniques to fit the cul-tural background of their patients. Research is neededto examine the influence of different cultures on psy-chological techniques and to verify the validity ofassessment tests used on patients from these cultures.

Family patterns and social interactions vary widelyfrom one culture to another. Kinetic Family Drawing(KFD; Burns & Kaufman, 1970), a widely used pro-jective drawing assessment test to measure familydynamics, is a promising test for cultural studies. Itwas developed in the United States by two Americanauthors, Burns and Kaufman, who empirically de-fined the criteria of analysis over several years. Intheir detailed review of the literature on the KFDtechnique, Handler and Habenicht (1994) listed stud-ies that investigate cultural norms. Many cliniciansappear to use the KFD test with non-Caucasian pop-ulations without considering the impact of the child’sculture on the drawing (Cabacungan, 1985; Fukada,1990; Ledesma, 1979; McNight-Taylor, 1974). Some

authors compared the KFD test with other psycholog-ical tests to verify its validity for non-Caucasian-American children (Cho, 1987; Shaw, 1989). Theseauthors made the assumption that the other psycho-logical tests used are valid for non-Caucasian-Amer-ican populations. Research that investigates the valid-ity of the KFD test with non-American children israre. A few studies compare American-CaucasianKFDs with those drawn by children from other cul-tures; and they usually conclude that there are culturalinfluences on the drawing results. Chuah (1992) com-pared KFDs of Chinese-American families with thoseof Caucasian-American families. She concluded thatKFDs illustrate cultural influences, and that KFDs areprobably a useful tool to investigate the acculturationprocess (i.e., the assimilation of the American cultureby the Chinese-American families). Nuttall, Chieh,and Nuttall (1988) compared KFDs of Chinese chil-dren from Beijing with those of American children.They concluded that the drawings reflect the socialvalues and norms of the two cultures. Urrabazo(1986) compared KFDs of Hispanic children withBurns and Kaufman’s normative sample and ascer-tained differences in the activities of the parental fig-ures. Chartouni (1992) compared KFDs of American-Lebanese children with KFDs of American-Caucasianchildren. Chartouni concluded that the KFD test is auseful tool to show cultural differences in family life-

Patricia Wegmann is a psychiatrist and psychotherapist in private practice (CH-1052 Le Mont, Switzerland), who is completing work on hermedical PhD at Geneva University, Switzerland.Vija B. Lusebrink is Professor Emerita and resides in Palo Alto, California.The authors wish to thank Franc¸ois Ferrero, Stephan Morgenthaler, Alain Wegmann, Holly Cogliati, Janis Yoshikawa and Sue Parker for theirassistance on this study.

The Arts in Psychotherapy, Vol. 27, No. 3, pp. 179–190, 2000Copyright © 2000 Elsevier Science Ltd.Printed in the USA. All rights reserved

0197-4556/00/$–see front matter

PII S0197-4556(00)00051-4

179

Page 2: KFD Scoring Method

styles. Unfortunately, none of these authors describedtheir methods clearly. When they presented their scor-ing sheets, usually there were no clear definitions ofthe variables. Consequently, a duplication of theirstudies, or the application of these studies to childrenfrom other cultures, is not possible. This is all themore unfortunate because there is no consensus on thescoring method used for the KFD. We agree withHandler and Habenicht (1994) when she states, “un-fortunately, the KFD technique has not been subjectto adequate critical research evaluation” and, “theKFD still remains primarily a clinical instrument withinadequate norms and questionable validity” (p. 441).Many authors have criticized Burns and Kaufman fortheir subjective interpretation of their variables. Mostresearchers have modified, without consensus, theBurns and Kaufman (1972) scoring system by addingnew variables or by modifying their definitions. As aresult, the variables are differently interpreted,thereby making the comparison of results difficult orimpossible.

To be able to study cultural influences with theKFD test, we need a reliable scoring method withclearly defined variables. In order for KFD to be avalid research tool, the definition of the variablesmust be standardized. The purpose of this study is todescribe and validate a scoring method that can beeffectively used by other researchers or clinicians andwith people from varying cultures. In this article, wefocused our attention on the reliability of the variablesof three different populations.

We combined variables from Burns and Kauf-man’s (1972) method with variables found throughliterature review. Then, we collected KFD drawingsfrom children (7–10 years old) from three differentcontinents. Two judges scored all the drawings. Inorder to verify the accuracy of the definitions of thevariables, we calculated the percentage of identicalanswers between the judges. We then compared theresults among the three populations. We hypothesizedthat this scoring method is reliable and that its reli-ability is independent from the cultural background ofthe child.

Scoring System

We set a scoring system combining criteria used inother studies with Burns and Kaufman’s (1972) cri-teria. And we added a few new criteria of our own.We determined a set of 44 variables, including 24“objects and symbols.” Inspired by Burns and Kauf-

man’s writings (1970, 1972), we looked through thedrawings and subjectively chose the objects and sym-bols most frequently drawn by the children in thisstudy.

The 20 variables are organized into six categories:18 are descriptive features and 2 are numerical mea-surements.

Family Composition

The first category analyzes the family membersthat are drawn, or not drawn, and their sizes. Wefocused on the major figures: the self, the mother andthe father. This category contains four variables.

Major figure missing.Burns and Kaufman (1972)proposed a variable calledomission of figureswhereby they search for any family members missing.As the definition of family members may vary widelyfrom one culture to another—nuclear families or ex-tended families—we found it interesting to comparethe presence or absence of the major figures betweenthe cultures. Hence, we definemajor figure missingasa major figure that is left out of the drawing.

Major figure’s erasure.Burns and Kaufman(1972) proposed a variable callederasurewhere theysearch for any erasure on the figures of family mem-bers. With the same reasoning as above, we decidedto focus on the erasure of only the major figures. TheKFD authors did not defineerasure. We definemajorfigure’s erasureas a significant erasure of the self,mother or father where the Gestalt of the figure isaltered. We believe that an erasure of a detail that isredrawn in a similar way does not have the samemeaning as a significant erasure, which indicates am-bivalence towards the figure.

Extended family members added.In their culturalstudies, Nuttall et al. (1988) and Chuah (1992) ob-served whether extended family members weredrawn. As the definition of family varies from oneculture to another, we found it very promising to adda variable related to the extended family. Therefore,we defineextended family members addedwhere anymembers other than self, mother, father or siblings aredrawn. They are counted and scored accordingly.

Size of figures.Burns and Kaufman (1972) pro-posed to measure the size of figures in their KFD gridand analysis sheet. They felt it could be a usefulvariable for cultural studies. Unfortunately, they didnot define how to measure the figures, especiallywhen they are curved or partially hidden. We definesize of figuresas the measurement, of the major fig-

180 WEGMANN AND LUSEBRINK

Page 3: KFD Scoring Method

ures, in millimeters following the midline of the body.If a figure is curved, its size is measured as the short-est distance between the head and the feet. If a figureis partially hidden, the judge only measures what isdrawn, without estimating the probable size of thefigure.

Distance and Closeness

The second category explores the relative positionbetween the figure of the self and the parental figures.It contains five variables.

1. Distance between figures. Burns and Kaufman(1972) proposed to measure the distance betweenthe self and the other figures in their KFD grid andanalysis sheet. They did not explain how. We de-fine distance between the figures as the closestdistance between any body parts of any two majorfigures.

2. Compartmentalization. This variable proposed byBurns and Kaufman (1970) was not clearly de-fined despite the numerous examples of drawingsgiven in their books. We propose to definecom-partmentalizationas lines that organize space andstructure the entire drawing. In order to be consid-ered ascompartmentalization, all the figures mustbe placed in a compartment.

3. Encapsulation. This is a variable proposed byBurns and Kaufman (1972). They were interestedin its presence or absence in the drawing. Theygave many examples ofencapsulation, but theydid not clearly define it. We found it more mean-ingful to score the encapsulated figures. We defineencapsulationas the lines that enclose or encirclea whole figure, as if the figure were in a capsuleand separated from the others in its own con-strained space. The capsule may border on theedge of the paper. Two figures may be enclosedtogether. When lines are part of an object, weconsider them as anencapsulationonly if the fig-ure appears isolated from the others.

4. Barrier. This is a variable proposed by O’Brienand Patton (1974). It was widely used with slightlydifferent definitions by many authors (Chuah,1992; Habenicht, personal communication, April26, 1996; Elin & Nucho, 1979; Reynolds, 1978).We propose to scorebarrier when two figures areseparated by an object or by lines—including thelines of a compartment or of an encapsulation.

It is sometimes difficult to differentiate betweencom-

partmentalization, encapsulationand barrier. Com-partmentalization concerns all or most of the surfaceof the drawing. Barrier concerns the accessibility ofany two figures, whereas encapsulation usually con-cerns one figure, although sometimes many figuresmay be encapsulated separately or together.5. Figure ascendance. In their KFD grid and analysis

sheet, Burns and Kaufman (1972) proposed to in-vestigate the relative placement of the figures.They suggested that it could be useful informationfor cultural studies. But they did not explain howto analyse the location of the figures. In her anal-ysis sheet, Habenicht (personal communication,April 26, 1996) defined afigure ascendancevari-able. She scored the position of the head on thevertical axis. We assume that the position of thefigure on the vertical axis, as well as the positionon the horizontal axis, provides valuable informa-tion. We propose the use of a transparent grid thatdivides the page in eight parts. If the head is lo-cated within two (or more) zones, the zone withthe biggest part of the head is scored.

Interactions and Relationship

The third category analyses the interaction andrelationship between the major figures. It containsthree variables.

1. Level of interaction. Burns and Kaufman (1972)categorized actions between figures with a conceptcalled field of force. They gave examples, but didnot define how they scored these interactions.Koppitz (1983) and Lyons (1993) scored onlywhether there was interaction between figures.Rather than scoring interaction per se, most of theother authors deduced interaction from other vari-ables. We propose to introduce a level of interac-tion variable. It scores the degree of interactionbetween two figures. We scoreactive interactionwhen two figures are engaged in a shared activityinvolving action (e.g., playing ball, eating to-gether, speaking to each other) or when the twofigures share the same kind of activity (e.g., doinghousehold chores, sharing a picnic). The two fig-ures have apassive action togetherif they areinvolved in the same passive activity (e.g., watch-ing TV together, reading books in the same room,standing in the same place, one figure watching theother). Sitting around a table doing nothing is apassive action, whereas sitting around a table talk-

181KFD SCORING METHOD

Page 4: KFD Scoring Method

ing or eating is an active action. Sitting togethereach reading separate books is a passive action; ifthe book is shared it is an active action; if the tworeading figures are in separate places, there is nointeraction.

2. Facing. In their study, O’Brien and Patton (1974)scored if the figures were facing each other or theviewer. They found links between this variable andthe notion of “self-concept”. Inspired by Habe-nicht (personal communication, April 26, 1996),Cho (1987), and Chuah (1992), we included a fac-ing variable. This variable is scored if a figure islooking towards another figure, rather than to-wards objects or looking “out of the picture.” Thisother figure may, or may not, look back. Eachfigure is scored separately. If a figure is orientedtowards another figure but not looking at it, thisvariable is not scored. For instance because it isreading a book, the self may be oriented towardsthe father and not look at him.

3. Level of nurturance. Inspired by Burns’ (1982)scoring criteria for figure nurturingand by Habe-nicht’s (personal communication, April 26, 1996)KFD analysis sheet, we define alevel of nurtur-ancevariable. It examines the level of nurturancefor each major figure.Feeding/holdingis scored ifone figure is taking care of another in very closecontact. If a figure is taking care of another figurewithout close contact,cooking/setting tableisscored.Eating is scored when a figure is eatingand taking care of pet/gardening/houseworkisscored when a figure is taking care of a pet, a plantor the house.

Activities

The fourth category analyses the level of activityof the figures. It contains only one variable.

1. Activity level. We simplified Burns’ (1982)scor-ing criteria for figure activity levelfrom nine pos-sible answers to five.Running/sportimplies that alot of energy is spent.Walking/doingimplies thatsome movement is drawn, andstanding impliesthat the figure does not seem to move and is stand-ing. Sitting is scored if the figure is sitting, eventhough it is doing something (like eating) andlay-ing is scored if the figure is laying.

Sexual Identification

The fifth category assesses the sexual identifica-tion of the self. It contains two variables.

1. Self drawn like. Inspired by Habenicht’s (personalcommunication, April 26, 1996)similar treatmentof figuresvariable, we propose a self drawn likevariable. This criteria is scored if the self figure isdrawn like one of the parental figures and differ-ently from the other parental figure:Mother (orfather) is scored if the hair or clothing are drawnwith similar details, or if the self and the mother(father) are drawn in markedly similar stances.These similarities must contrast with details usedfor other figures and be obvious.

2. Self sharing activity with. To see if the child feelshim/herself closer to one parent than the other, weadded a self sharing activity with variable. It isscored if the self figure is doing the same, or samekind of, activity with one parent and not with theother.

Developmental Level

The sixth category analyzes the developmentallevel of the child. It contains five variables.

1. Space organization. According to Lowenfeld andBrittain’s (1987) description of developmentstages in art, we introduced aspace organizationvariable that scores the representation of space inthe drawing.No space organizationis scored whenthere is no sense of spacial organization (i.e., theobjects and figures are floating without any rela-tion to one other).One baselineis scored when thefigures and objects are organized in a line or whenthere is one baseline. The drawing suggests thatthe child has an understanding of 2D.3D attemptis scored when there is an attempt to represent 3D.For instance, there are many baselines or manypoints of view in the drawing (i.e., tilting the tableso we can see what it is on it).3D awarenessisused when the drawing shows perspective, whenthere is a sense of depth, and/or when there areobjects or figures overlapping.

2. Incomplete body. The completeness of the bodiesdrawn is another criteria for scoring developmentlevel. Each major figure is analyzed separately.The incomplete bodyvariable is scored if arms,legs, trunk, hands or feet are missing; if stick fig-ures are drawn; if there is a poor integration of

182 WEGMANN AND LUSEBRINK

Page 5: KFD Scoring Method

body parts; or if a part of the body is cut-off thepage. This variable is not scored if parts of thebody are not drawn because they are obscured byan object or another figure. Missing facial featuresare scored separately.

3. Incomplete face. It is scored if eyes or mouth aremissing; if part of the face is cut-off the page; or ifthere is a poor integration of the facial features.This variable is not scored if the nose is missing orif the face is hidden (e.g., because the figure’s backis turned).

4. Sexual Differentiation. It is scored if the gender ofall figures (except young children and babies) isobvious, even though stick figures are drawn.

5. Akinesis. We set anakinesiscriteria inspired byMostkoff and Lazarus’ (1983)evasioncriteria andby Reynolds’ (1978)motionless or stick figures.This variable rates whether activities or physicalstances are drawn. It is scored if all figures arestanding facing forwards, as if they were sitting fora snapshot without any orientation to each other orto tasks. It is also scored if the figures are drawnstatic, despite objects drawn involving an activityor despite the child’s verbal description that thefigures are engaged in activities.

Method

Once our scoring method was defined and docu-mented in a manual, we collected KFD drawings fromchildren from three different continents. This studyincluded over 120 children. Thirty-seven children livein a rural town situated near a university city in Swit-zerland (Froideville, Vaud); 42 children are Taiwan-ese and live in a university city situated in a ruralregion in Taiwan (Hsinchu); and 44 children are Cal-ifornian, living in a university city (Palo Alto). TheTaiwanese and the Swiss children were not exposedto cultural influences other than their own (except forTV and McDonald’s). For the American children, wechose for the study only those children who, alongwith their parents, were born in the USA and hadnever lived abroad.

The children of this study were between 7 and 10years old. They were “nonclinical” children, i.e., theyattended the usual school classes. Each cultural groupcompleted the KFD test at the same time, in their ownclassroom during their usual art class. The data werecollected as part of a battery of tests, including theBridge Drawing (Hays & Lyons, 1981). For the KFDdrawing, they were asked to follow Burns and Kauf-

man’s (1970) instructions: “Draw a picture of every-one in your family doing something. Try to drawwhole people, not cartoons or stick people. Remem-ber to make everyone doing something” (pp. 19–20). The instructions were translated in Mandarinfor the Taiwanese children and in French for theSwiss children. They were given a pencil and a sheetof plain white paper (8“3 11”, A4 in Taiwan andSwitzerland). When the drawings were completed,the children identified their family members on thedrawings.

Two judges, independent from the study, scored allthe drawings. The judges were graduate students fromUniversity of California Berkeley Extension Post-Masters Certificate Program in clinical art-therapy.They were trained in art-therapy techniques and un-derstood the variables’ definitions. Because they didnot have much clinical experience with KFD scoringsystems, they could easily accept our proposed scor-ing system. To train the judges to use our scoringmethod, we asked them to score, following our man-ual, a few drawings not included in the study. Wediscussed their answers with them. When they hadscored the 121 drawings we compared the scores ofthe two judges. First we calculated the percentage ofidentical answers to obtain the degree of reliability foreach variable. Then, to compare the reliability resultsamong the three groups of children, we applied thex2

test to verify whether the results were independentfrom the cultures.

Results

We decided that a good reliability result is 80% ormore of identical answers between both judges. Re-sults between 75% and 80% are acceptable. For thesizes and distances we decided to accept: for shortmeasurements (up to 40 mm) a difference of 4 mm,for longer distances (above 40 mm) a 10% differencewith a maximum of 10 mm difference between bothmeasurements.

When we compiled the percentage of agreementsbetween both judges for the three samples as a whole,we obtained a high degree of agreement for 80% ofthe variables. The 24 objects and symbols reachedmore than 85% agreement between judges, 14 of the20 variables reached 80% or more agreement, 3 vari-ables attained an acceptable result, and we obtainedan insufficient percentage of agreement between thejudges for only 3 variables. However, when we sep-arated the agreement results by population, we saw

183KFD SCORING METHOD

Page 6: KFD Scoring Method

that these good global results were sometimes mis-leading, and that they may hide conflicting resultsbetween the three samples. Twenty-two of the 24objects and symbolsstill had a high percentage ofagreement between judges for each population (seeTable 1). Thegrid symbol and theemphasis on wallsymbol attained good results for the Swiss and theAmerican samples only. They had a lower result forthe Taiwanese population (79% forgrid and 71% foremphasis on wall).The difference foremphasis onwall is highly significant with the Chi-square test(x2 5 14.5,p . 0.01).

Differences in reliability results are more frequentwith the descriptive and measurable variables (seeTable 2). Only 7 of the 14 variables that obtainedgood global results showed good results for each of

the three populations without significant differenceamong them with the Chi-square test:major figuremissing, major figure’s erasure, extended familymembers added, distance between figures, encapsu-lation, self sharing activity withand incomplete face.Four other variables had good global results; butwhen we calculated by population, we found goodresults for one or two populations and only acceptableones for the other(s):size of figure, level of interac-tion, level of nurturanceand space organization.These four variables have global results around 80%and the Chi-square test did not indicate any signifi-cant difference among the three samples.

Six variables revealed statistically significant dif-ferences of reliability among the three samples.Sometimes these differences were found only for one

Table 1Reliability of the Objects and Symbols Variables: Identical Answers Between Both Judges

USA Taiwan Switzerland

global %c x2dA/Ta %b A/T % A/T %

House 42/43 98 41/41 100 35/37 95 98Emphasis on Roof 42/43 98 36/42 85 36/37 97 93Emphasis on Wall 42/43 98 29/41 70 34/37 92 87 14.5*Smoking Chimney 43/43 100 40/42 95 36/37 97 98Antenna 43/43 100 38/42 90 34/37 92 94Door 40/42 95 37/42 88 35/37 95 93Stairs 43/43 100 41/42 98 35/37 95 98Grid Window 42/43 98 39/42 93 36/37 97 96Bed 42/43 98 35/41 85 30/37 81 88Light 40/43 93 37/42 88 34/37 92 91Roof Figure 43/43 100 42/42 100 37/37 100 100Cross-shaped Body 41/43 95 39/42 93 36/37 97 95Cut-off Head 42/43 98 40/42 95 37/37 100 98Ball 41/43 95 40/42 95 35/37 95 95Jump Rope 42/43 98 40/42 95 37/37 100 98Remote Control Device 42/43 98 41/42 98 34/37 92 96Clouds 42/43 98 41/42 98 37/37 100 98Sun 43/43 100 41/42 98 36/37 97 98Moon 43/43 100 39/42 93 37/37 100 98Stars 42/43 98 42/42 100 37/37 100 99Snow 42/43 98 42/42 100 35/37 95 98Pet 41/43 95 42/42 100 36/37 97 98Coffin 42/43 98 42/42 100 37/37 100 99Grid 40/43 93 33/42 79 32/37 86 86 3.7

a A 5 agreement: Number of identical answers; T5 total: Number of answer.b % 5 percentage of identical answers.c Global %5 percentage of identical answers for the (3) populations.d x2 test result for the three populations.* Level of significancep , .01.

184 WEGMANN AND LUSEBRINK

Page 7: KFD Scoring Method

Table 2Reliability of the Descriptive and Measurable Variables: Identical Answers Between Both Judges

USA Taiwan Switzerland

Global %c x2dA/Ta %b A/T % A/T %

OmissionS 42/43 98 42/42 100 37/37 100 99 1.8M 43/43 100 42/42 100 37/37 100 100F 43/43 100 41/42 98 36/37 97 98 1.1

ErasureS 30/37 81 32/36 89 29/32 91 87 1.6M 34/36 94 34/38 89 23/25 92 92 0.6F 30/35 86 37/38 97 19/22 86 91 3.5

Extended 38/42 90 41/42 98 36/37 97 95 2.9Figures’ Sizes

S 29/33 88 29/36 81 23/28 82 84 0.7M 30/33 91 29/37 78 17/22 77 83 2.5F 26/29 90 31/38 81 16/21 76 83 1.7

DistancesS/M 23/24 96 27/30 90 19/20 95 93 0.9S/F 25/27 93 31/31 100 18/19 95 96 2.2M/F 24/29 83 31/35 89 20/20 100 89 3.8

Compartmentalization 41/43 95 33/42 79 33/37 89 88 5.7*Encapsulation

S 34/37 92 30/36 83 28/32 88 88 1.2M 32/35 91 34/38 89 24/25 96 92 0.9F 31/32 97 36/38 95 21/22 95 96 0.2O 37/42 88 36/42 86 30/36 83 86 0.3

BarrierS/M 25/30 83 30/33 91 15/22 68 82 4.7*S/F 22/29 76 28/32 88 14/20 70 79 2.5M/F 22/32 69 36/37 97 17/22 77 80 10.2

Figure’s AscendanceS 37/37 100 29/36 81 28/32 88 90 7.6**M 34/35 97 33/38 87 23/25 92 92 2.6F 31/32 97 32/38 84 19/22 86 89 3.1

InteractionS/M 26/31 84 30/33 91 17/22 77 85 1.9S/F 24/28 86 30/32 94 18/20 90 90 1.1M/F 24/30 80 35/37 95 19/22 86 88 3.3

FacingS/M 26/30 87 33/33 100 13/21 62 86 15.2**S/F 28/28 100 30/32 94 18/20 90 95 2.6M/S 26/30 87 31/33 94 16/22 73 86 4.9*M/F 27/30 90 34/37 92 19/22 86 90 0.5F/S 25/28 89 30/32 94 17/22 85 88 1.1F/M 28/30 93 32/38 84 22/22 100 91 4.6*

a A 5 agreement: Number of identical answers; T5 total: Number of answer.b % 5 percentage of identical answers.c Global %5 percentage of identical answers for the three populations.d x2 5 x* test result for the three populations.* Level of significancep , .05.** Level of significancep , .01.

185KFD SCORING METHOD

Page 8: KFD Scoring Method

of the figures tested.Barrier results were good for theTaiwanese sample, insufficient for the Swiss sampleand were not homogeneous for the US sample (S/M83%, S/F 76%, M/F 55%). The Chi-square test con-firms significant differences for the pairs S/M (x2 54.7,p . 0.05) and M/F (x2 5 10.2,p . 0.05); it didnot for the pair S/F. Theincomplete bodyvariable didnot display homogenous results. The Swiss samplehad good results for the three major figures. The USsample showed a good result for the mother, an ac-ceptable for the self and an insufficient for the father.The Taiwanese sample obtained acceptable results forthe mother and the father figures and a good result forthe self. The Chi-square test demonstrated a statisticaldifference only for the mother figure (x2 5 4.6, p .0.05). For the other figures, other than the major ones,the results were insufficient for the three populations.The facingvariable rates six combinations of figures.Results for these six combinations were not homoge-neous. Half of these combinations had good resultsand showed no significant differences. The other half

showed significant differences: The Swiss sample at-tained insufficient results for S/M (62%,x2 5 15.2;p . 0.05) and M/S (73%,x2 5 4.9; p . 0.05). TheTaiwanese sample got a lower but still good result forthe combination F/M (84%;x2 5 4.6, p . 0.05).Compartmentalizationattained good results for theAmerican and the Swiss samples. The Taiwanese re-sult was significantly, statistically lower (79%,x2 55.7, p . 0.05). Figure ascendanceobtained goodresults for each population. The American sample hadhigher agreement between the judges when the selffigure was rated (x2 5 7.6 ,p . 0.01). For the motherand the father figures, there was no significant differ-ence between the populations.Akinesisgot an insuf-ficient global result; the result for the Swiss samplewas excellent (94%), whereas the results were insuf-ficient for the US (62%) and the Taiwanese (64%)samples. These differences were confirmed with theChi-square test (x2 5 12.1).

Three variables attained insufficient results.Sexualdifferentiationhad an global result of 75%. The Tai-

Table 2Continued

USA Taiwan Switzerland

Global %c x2dA/Ta %b A/T % A/T %

NurturanceS 32/37 86 35/36 97 30/32 94 92 3.1M 27/35 77 35/38 92 22/25 88 86 3.5F 28/32 88 34/38 89 21/22 95 90 0.9

Activity LevelS 24/37 65 26/36 72 17/32 53 64 2.7M 21/35 60 10/38 26 8/25 32 40 9.5**F 23/32 72 28/38 74 13/22 59 70 1.5

Self Drawn Like 29/36 81 23/33 70 13/21 62 72 2.5Self Share Activity 33/36 92 33/33 100 19/21 90 94 3.1Space Organization 34/43 79 35/42 83 29/37 78 80 0.4Incomplete Body

S 29/38 76 29/36 81 27/32 84 80 0.7M 31/35 89 27/38 71 22/25 88 82 4.6*F 23/32 72 28/38 74 19/22 86 76 1.7O 27/42 64 24/41 59 21/35 60 61 0.3

Incomplete FaceS 35/38 92 31/34 89 28/32 88 90 0.4M 33/36 92 32/38 84 25/26 96 90 2.6F 30/32 94 32/38 84 21/32 94 81 2.5O 37/42 88 34/41 83 28/35 80 84 1.0

Sexual Differentation 31/42 74 34/41 83 23/35 66 75 3.0Akinesis 26/42 62 27/42 64 33/35 94 72 12.1**

186 WEGMANN AND LUSEBRINK

Page 9: KFD Scoring Method

wanese sample was good (83%), whereas the US(74%) and the Swiss (66%) were insufficient. TheChi-square test did not confirm these differences.Selfdrawn like got an insufficient global result. The re-sults were insufficient for the Taiwanese (70%) andthe Swiss (62%) samples, but it was good for theAmerican sample (81%). These differences were notconfirmed with the Chi-square test.Activity levelat-tained insufficient results for all samples.

Discussion

When we compiled the three samples as a whole,the good reliability results were misleading: onlythree variables seemed to have insufficient agreementbetween judges. However, when we calculated thereliability results for each population and comparedthem, we noticed that there are statistically significantdifferences among the three populations. A goodoverall reliability result for one variable may hide aninsufficient percentage of agreement between judgesfor one population; or conversely, an insufficientglobal result may hide a good reliability result.

For the objects and symbolswe obtained goodglobal results. When these results were detailed, wenoticed that the reliability was good for the threesamples except for thegrid and theemphasis on wallsymbols. These symbols attained an good reliabilityresult for the Swiss and the American samples onlyand got a lower result for the Taiwanese population.The grid symbol still attained an acceptable result(79%) and the Chi-square test does not conclude to ahighly significant difference (x2 5 3.7).Emphasis onwall did not reach a sufficient percentage of agree-ment for Taiwanese sample (71%) and this resultindicates highly significant differences with the Chi-square test (x2 5 14.5). We assumed thatemphasis onwall was not clearly defined. As the Taiwanese chil-dren draw more often houses than the American andSwiss children, they draw more walls, thus increasingthe opportunities for disagreement between judges.

For the twenty variables that are notobject orsymbol the situation is more complex. Only sevenvariables attained good reliability results for eachsample:major figure missing, major figure’s erasure,extended family members added, distance betweenfigures, encapsulation, self sharing activity withandincomplete face. The Chi-square test did not indicateany significant difference among the three samples.We concluded that these seven variables are reliableand that their reliability is independent of the thee

cultures tested. Four variables reached an averageresult barely above 80% of agreement between bothjudges: size of figure, level of interaction, level ofnurturanceand space organization. When the reli-ability was computed for each sample, the result forone population was below the threshold of 80%. Asthe Chi-square test did not show significant differ-ences among the results of the three samples and, aseach of the results were above 75%, we consider theresults for each population as good. These 11 vari-ables attained good reliability results without signifi-cant differences between the three samples. They maybe used for cross-cultural studies with the three pop-ulations tested. We noticed that even the variablesthat rely on subjective appreciation, such aserasure(changing a Gestalt), orencapsulation(being sepa-rated from the others), obtained a high percentage ofagreement between both judges.

For six variables, the reliability compiled by pop-ulations brought to our attention statistically signifi-cant differences between the three samples. Some-times these difference were found only for one of thefigures tested.

Compartmentalizationattained a good global reli-ability result. The results for each sample show sig-nificant differences, with a lower but still acceptableresult for the Taiwanese population (x2 5 5.7). Tai-wanese children tend to draw X-ray houses with eachfamily member placed in a different room. When thehouse occupies the whole surface of the paper, thedrawing gives an impression of compartimentaliza-tion and was usually rated as such by the judges. Mostof the disagreement between the judges occurredwhen X-ray houses were drawn only in a part of thepage and did not involve all of the paper. The dis-crepancy between reliability results attract our atten-tion on this Taiwanese feature and raise the questionif these X-ray houses may be considered as compar-timentalization or not. We wonder if our definition ofthe variable compartimentalization includes two dif-ferent features. This question is all the more interest-ing since the interpretation of the variablecompart-mentalization, proposed by Burns and Kaufman(1972), is debated. Some authors agree with them:Reynolds (1978) defined it as a sign of “inhibition ofstrong emotions; isolation; inability to communicateopenly.” Other authors suggested another definition;for example, McPhee and Wegner (1976) concludedin their study that

187KFD SCORING METHOD

Page 10: KFD Scoring Method

Compartimentalization seemed to appear re-peatedly in those drawings (both adjusted anddisturbed) depicting the family members en-gaged in roles associated with the typical Amer-ican household. Perhaps in an effort to imposeclarity and order on this assortment of activities,the subjects employed the use of lines to sepa-rate the various figures—thus producing thestyle defined as Compartmentalization. (p. 490)

Hence, comparison with other variables that give in-formation about family member interactions and ac-tivities might help us to better interpret this variable:Does it indicate separation and/or lack of communi-cation; or, does it indicate a tendency to organize,structure space and roles of family members?

With the barrier variable we reached an overallreliability result near 80%. The Taiwanese drawingshad a better reliability result than the Swiss andAmerican ones, with a statistically significant differ-ence for the pair mother/father (x2 5 16) and for thepair self/mother (x2 5 4.7). If we calculate the reli-ability for the American and Swiss drawings only, theresults will be below 80% (77% for self/mother; 77%for self/father and 66% for mother/father). As theTaiwanese drawings more often containedcompart-mentalization, the barriers between the pairs of fig-ures were more obvious. Thus it led to a higher per-centage of identical answers between judges.Studying the American and Swiss drawings that gotdifferent scores for this variable, we noticed that mostdivergences occurred when an object or another fig-ure only partially separated the pair of figures. Wecan improve our definition if we take partial separa-tion into account. In addition, it would be useful toseparatebarrier from the lines ofcompartmentaliza-tion as these two variables might indicate differentfeatures.

The good reliability result of thefacingvariable ismisleading. Only the result of the Swiss drawings forthe self/mother pair of figures was insufficient. Actu-ally, the variable was not present at all in the Amer-ican drawings and only rarely in the Taiwanese, but itwas present a few times in the Swiss drawings. TheSwiss children tried, more often than the others, torepresent figures facing each other. The way the vari-able is defined does not allow for the ambiguity of thedrawing done by children that do not master the skillto draw figures in profile. Studies would be useful toconfirm if this variable is actually more often present

in Swiss drawings then in American or Taiwanesedrawings.

The reliability results for the variableincompletebodyare intriguing as they differ from one figure tothe other. For the self figure, the reliability is goodwhereas for the father figure it is insufficient. Forthese two figures there are no significant differencesbetween the three samples. On the other hand, for themother figure, the Taiwanese reliability result is sig-nificantly statistically lower with an insufficient per-centage of agreement between the jugges, while thereliability result is good for the two other populations.We have no hypothesis for the reason for the discrep-ancies of the results. However, we suppose that thedisagreements between judges are related to the waychildren at these ages draw hands and feet. Oftenthese features are suggested and not clearly drawn.We assume that if we had defined more clearly thisvariable, especially the minimum features needed forhands and feet, it would have improved the reliabilityresults. The result of this variable for other non-majorfigures is not reliable. A drawing could contain to-gether complete and incompleteother figures. Thescoring of the other figures is debatable because theymust be scored all together as complete or incom-plete.

Figure ascendancealso attained intriguing results.The American sample had significantly statisticallyhigher agreement between the judges when the selffigure was rated. For the mother and the father fig-ures, this variable did not show significant differencebetween the populations. We have no hypothesis forthis inconsistency between the figures results.

We obtained low reliability results forakinesisexcept with the Swiss children. The scoring sheetsshow that one judge scoredakinesis on only onedrawing. The other judge scored this variable manytimes, especially with the Taiwanese and Americandrawings. This leads to more disagreement aboutthese cultures than the Swiss culture. The definitionof this variable is not clear as it was obviously under-stood differently by each judge. Nevertheless, itwould be interesting to clarify the definition to findout if this variable is actually distributed differently inthe three cultures.

The variableactivity level, obtained a low reliabil-ity result for all samples. We made the hypothesis thatthe variable’s definition was not accurate enough.During the schematic stage, at least in the early phase,children have a poor ability to draw figures in aclearly recognizable activity. Hence the poor results

188 WEGMANN AND LUSEBRINK

Page 11: KFD Scoring Method

of activity levelcan be attributed to the developmentallevel of the children of the study. The judges probablyhad to guess the type of action drawn, thus adding animportant subjectivity to the result. It was probablydifficult for them to decide if the figure was simplystanding or actually doing something. The majority ofdisagreements were between the categories of an-swersstandingand doing. If we combine these twocategories together and keep as multiple choices:moving, standing, sittingandlaying, the result is goodwith no significant difference with the Chi-square testamong the cultures (Self: 82%,x2 5 0.9; Mother:91%,x2 5 0.3; Father: 87%,x2 5 2.7).

We obtained an insufficient percentage of agree-ment between judges for thesexual differentiationvariable with the Swiss and American children but notwith the Taiwanese. For this variable, all the figuresof the drawing were scored together. The drawingsshowing partial differentiation got arguable scoring.The Swiss and American drawings often containedfigures that were sexually recognizable and some thatwere not. In the Taiwanese drawings, most of thefigures present a sexual differentiation and reach a83% of identical answers between the judges. Weassume that if we had proposed a third possibility,partial differentiation, or if the major figures werescored separately, we would have had a better reli-ability result.

Conclusion

To analyze drawings from children of differentcultural backgrounds, we propose a revised scoringmethod for the KFD test. We verified the reliability ofthe method and demonstrated the importance ofclearly defined variables and the necessity to test thereliability of the variables with each cultural sample.We demonstrate that clinicians must be very carefullwhen applying the KFD test with children from an-other cultural background than the one it was de-signed for. Indeed, a variable may be reliable with onecultural sample, but not necessarily with another.Many variables obtained statistically significant dif-ferent reliability results from one population to theother. Most of the time these differences were due toa different distribution of the variable between thesamples. If a variable is more frequent in one popu-lation and if its definition is not clear, then we mayhave more disagreement between the judges for thispopulation than for the others.

A more interesting reason for variations in the

reliability results among the populations is when avariable was defined so that two different featureswere scored together. For instance, the Taiwanesechildren often drew X-ray houses with the rooms laidout over the whole page. This feature looked likecompartmentalizationand was scored as such. Dis-agreement occurred when X-ray houses were notspread over the whole page. This leads us to wonderif these X-ray houses are to be considered as com-partmentalization or some other feature. This brings anew argument to the debate of the interpretation to begiven to this variable: does it indicate a lack of com-munication or a tendency to structure and organize.Additional studies are needed to answer the questionwhether the different distribution of the variables be-tween populations is a consequence of an insufficientvariable definition, or a sign of cultural influence onthe personality. Before this method could be appliedwith confidence in clinics, further studies are neededto learn if the variables can predict phenomena ofclinical importance. The purpose of this article is thedescription and the verification of the applicability ofthis scoring method for cross-cultural studies, thus wefocused on the cultural differences in the reliability ofthe variables. We did not analyse here the differencesof the distribution of the variables among the popu-lations. Such analysis and the discussion of their clin-ical implications will help to verify the relevance ofthis method in clinics. It will be the topic of a futurearticle.

In addition to finding cultural differences in thereliability of the variables, this study shows that thecontribution of cross-cultural studies is to questionand thus to clarify the definitions and interpretationsof the KFD variables. Cross-cultural studies of thetest improve our understanding and use of it by chal-lenging us to verify the validity of the definition of thedifferent variables, even when the test is used withpopulation for whom it was designed.

References

Burns, R. (1982). Self-growth in families: Kinetic Family Draw-ings (K-F-D) research and application. New York: Brunner/Mazel.

Burns, R., & Kaufman, S. (1970). Kinetic Family Drawings (K-F-D). An introduction to understanding children through kineticdrawings. New York: Brunner/Mazel.

Burns, R., & Kaufman, S. (1972). Actions, styles and symbols inKinetic Family Drawings (K-F-D): An interpretive manual.New York: Brunner/Mazel.

Cabacungan, L. (1985). The child’s representation of his family in

189KFD SCORING METHOD

Page 12: KFD Scoring Method

Kinetic Family Drawings (KFD): A cross cultural comparison.Psychologia, 28, 228–236.

Chartouni, T. (1992). Self-concept and family relations of Ameri-can-Lebanese children. Unpublished doctoral dissertation, An-drews University, Berrien Springs, MI.

Cho, M. (1987). The validity of the Kinetic Family Drawing as ameasure of self-concept and parent/child relationship amongChinese children in Taiwan. Unpublished doctoral dissertation,Andrews University, Berrien Springs, MI.

Chuah, V. (1992). Kinetic Family Drawings of Chinese-Americanchildren. Unpublished doctoral dissertation, Andrews Univer-sity, Berrien Springs, MI.

Elin, N., & Nucho, A. (1979). The use of Kinetic Family Drawingas a diagnostic tool in assessing the child’s self-concept.ArtPsychotherapy, 6, 241–247.

Fukada, N. (1990). Family drawing: A new device for cross-cul-tural comparison. Paper presented at the Twenty-Second Inter-national Congress of Applied Psychology, Kyoto, Japan.

Handler, L., & Habenicht, D. (1994). The kinetic family drawingtechnique: A review of the literature.Journal of PersonalityAssessment, 62, 440–464.

Hays, R., & Lyons, S. (1981). The Bridge Drawing: A projectivetechnique for assessment in art therapy.The Arts in Psycho-therapy, 8, 207–217.

Koppitz, E. (1983). Projective drawings with children and adoles-cents.School Psychology Review, 12, 421–427.

Ledesma, L. (1979). The Kinetic Family Drawings (KFD) of Fili-pino adolescents. Unpublished doctoral dissertation, BostonCollege, Boston, MA.

Lowenfeld, V., & Brittain, W. L. (1987). Creative and mentalgrowth (8th ed.). Englewood Cliffs, NJ: Prentice Hall.

Lyons, S. (1993). Art psychotherapy evaluations of children incustody disputes.The Arts in Psychotherapy, 20, 153–159.

McKnight-Taylor, M. (1974). Perceptions of relationships in lowincome Black families. Unpublished doctoral dissertation, Uni-versity of Virginia, Charlottesville, VA.

McPhee, J., & Wegner, K. (1976). Kinetic-Family-Drawing Stylesand emotionally disturbed childhood behavior.Journal of Per-sonality Assessment, 40, 487–491.

Mostkoff, D., & Lazarus, P. (1983). The Kinetic Family Drawing:The reliability of an objective scoring system.Psychology inthe Schools, 20, 16–20.

Nuttall, E., Chieh, L., & Nuttall, R. (1988). Views of the familyby Chinese and U.S. children: A comparative study of KineticFamily Drawings.Journal of School Psychology, 26, 191–194.

O’Brien, R., & Patton, W. (1974). Development of an objectivescoring system for the Kinetic Family Drawing.Journal ofPersonality Assessment, 38, 156–164.

Reynolds, C. (1978). A quick-scoring guide to the interpretation ofchildren’s Kinetic Family Drawings (KFD).Psychology in theSchool, 15, 489–492.

Shaw, J. (1989). A developmental study on the Kinetic FamilyDrawing for a nonclinic, Black-child population in the Mid-western region of the United states. Unpublished doctoral dis-sertation, Andrews University, Berrien Springs, MI.

Urrabazo, R. (1986). Machismo: Mexican American male self-concept: An interpretation and reflection on the Thematic Ap-perception test and Kinetic Family Drawing results of MexicanAmerican teenagers. Unpublished doctoral dissertation, Gradu-ate Theological Union, Berkeley, CA.

190 WEGMANN AND LUSEBRINK