children's behaviour and their graphic representation of parents and self

13
The Arts in Psychotherapy 32 (2005) 107–119 Children’s behaviour and their graphic representation of parents and self Lisa C. Milne (DPsych) a,, Philip Greenway (Doc en Psych) b , Fiona Best b a Cardinia—Casey Community Health Service, Southern Health, Cranbourne, Vic. 3130, Australia b Faculty of Education, Monash University, Clayton, Victoria Human figure drawings have been shown to serve as indicators of many psychological features, including: the drawer’s perception of him or herself (Berryman, 1959); an insight into a person’s representation of self and others (Leibowitz, 1999); anxiety; (Fox, Davidson, Lighthall, Waite, & Sarason, 1958); and aggression and acting out behaviour (Koppitz, 1968, 1984). Drawings also enable us to understand social relationships. Researchers have used the Kinetic Family Drawing Test (K.F.D.; Burns & Kaufman, 1970) to assess family functioning and children’s perceptions of their relationships to other family members (Oster & Gould, 1987). Main, Kaplan, and Cassidy (1985) found that children’s family drawings could be used to assess the quality of the parent–child attachment relationship. Madigan, Ladd, and Goldberg (2003) noted that, although internal representations may be a primary means of carrying early attachment experiences forward, there is not much literature on the information carried by children’s representations of relationships, such as might be found in projective techniques or drawings. Fury, Carlson, and Sroufe (1997) pointed out that the drawings of children aged in the range 5–11 years, effectively represented attitudes, fears, and wishes that they could not or did not want to express verbally. These authors followed Kaplan and Main’s instructions for the classification of representations of attachments expressed in children’s drawings (cited in Fury et al., 1997) in a study of high-risk, racially mixed children aged 8–9 years. They confirmed Kaplan and Main’s position that drawings could be regarded as a potentially robust measure for tapping Corresponding author. Tel.: +61 3 98990158. E-mail address: [email protected] (L.C. Milne). 0197-4556/$ – see front matter © 2005 Elsevier Inc. All rights reserved. doi:10.1016/j.aip.2005.01.005

Upload: lisa-c-milne

Post on 02-Sep-2016

215 views

Category:

Documents


2 download

TRANSCRIPT

Page 1: Children's behaviour and their graphic representation of parents and self

The Arts in Psychotherapy 32 (2005) 107–119

Children’s behaviour and their graphicrepresentation of parents and self

Lisa C. Milne (DPsych)a,∗, Philip Greenway (Doc en Psych)b,Fiona Bestb

a Cardinia—Casey Community Health Service, Southern Health, Cranbourne, Vic. 3130, Australiab Faculty of Education, Monash University, Clayton, Victoria

Human figure drawings have been shown to serve as indicators of many psychologicalfeatures, including: the drawer’s perception of him or herself (Berryman, 1959); an insightinto a person’s representation of self and others (Leibowitz, 1999); anxiety; (Fox, Davidson,Lighthall, Waite, & Sarason, 1958); and aggression and acting out behaviour (Koppitz,1968, 1984). Drawings also enable us to understand social relationships. Researchers haveused the Kinetic Family Drawing Test (K.F.D.;Burns & Kaufman, 1970) to assess familyfunctioning and children’s perceptions of their relationships to other family members (Oster& Gould, 1987). Main, Kaplan, and Cassidy (1985)found that children’s family drawingscould be used to assess the quality of the parent–child attachment relationship.Madigan,Ladd, and Goldberg (2003)noted that, although internal representations may be a primarymeans of carrying early attachment experiences forward, there is not much literature onthe information carried by children’s representations of relationships, such as might befound in projective techniques or drawings.Fury, Carlson, and Sroufe (1997)pointedout that the drawings of children aged in the range 5–11 years, effectively representedattitudes, fears, and wishes that they could not or did not want to express verbally. Theseauthors followed Kaplan and Main’s instructions for the classification of representationsof attachments expressed in children’s drawings (cited inFury et al., 1997) in a study ofhigh-risk, racially mixed children aged 8–9 years. They confirmed Kaplan and Main’sposition that drawings could be regarded as a potentially robust measure for tapping

∗ Corresponding author. Tel.: +61 3 98990158.E-mail address:[email protected] (L.C. Milne).

0197-4556/$ – see front matter © 2005 Elsevier Inc. All rights reserved.doi:10.1016/j.aip.2005.01.005

Page 2: Children's behaviour and their graphic representation of parents and self

108 L.C. Milne et al. / The Arts in Psychotherapy 32 (2005) 107–119

into children’s representational models of attachment. They also reported that individualdrawing signs had relatively little predictive power and that an aggregation of signs faredbetter. Likewise, the individual subscales of the Child Behaviour Checklist were not aspowerful predictors of attachment history scores as was the Global Pathology Scale (Furyet al., 1997). Madigan et al. (2003)found, however, that the findings ofFury et al. (1997)failed to generalise to a low-to-moderate risk sample of 7-year-old children.

Children’s drawings do seem to reflect their internal representations of themselves andothers and in practice clinicians seem to base their interpretations of drawings on this as-sumption.Gillespie (1994)asked children to draw a picture of a mother and a child, with aview to measuring the child’s sense of self as developed through infancy, in the context ofthe mother–child relationship. The exact nature of the sample used is not clear. However,Gillespie (1994)reports that she used hundreds of drawings, which came from childrenindividually, schools, and church communities, none coming from patient populations, norspecial groups such as children with delayed development. She analysed these drawingsfrom an object relations perspective, noting that children tend to identify with the childfigure in the drawing. She also observed that, as girls approached puberty, they tended toidentify more strongly with the mother figure. On the other hand, pubescent boys oftendrew non-human representations of both mother and child. This strategy, she speculated,might have represented an attempt to solve their dilemma of being uncomfortable in hav-ing to identify with either. Then often, as they matured and developed positive, internalrepresentations of women, they were able to represent mother-and-child in a human form(Gillespie, 1994). She also remarked that the drawings of children whose development isfree of clinical disturbance, show little differentiation in their representations of child andmother although as they developed their drawings do become more differentiated, even iffairly large individual differences may be apparent. In my clinical experience (Milne) chil-dren tend to show greater individual detail in the drawing of the self and less differentiationin that of people other than self, such as family members.

Clearly we need a rigorous system for scoring drawings, yet despite numerous attemptsat scoring drawings, researchers seem to encounter significant, systematic problems. Thisposition contrasts notably with that of clinical child psychologists, who seem to use drawingsas a useful way to add to and help in the interpretation of information obtained from clinicalinterviews and psychometric tests. However, without an objective scoring system, valid andreliable conclusions are virtually impossible.

On the assumption that children conceive of the figures they draw in terms of theirinternal representations of themselves and others, we sought to devise valid measuresof similarities and differences between parental figures in children’s drawings, with theexpectation that these would correlate with behavioural outcomes. The rationale for thishypothesis was twofold. Firstly, behavioural difficulties and characteristics derive fromattachment relationships, especially with a child’s primary caregivers (Karen, 1998).Main, Kaplan, and Cassidy (1985),Fury et al. (1997), and Madigan et al. (2003)allfound that children’s drawings of themselves and their parents provided indications of theparent–child attachment relationship. Poor parent–child relationships result in an array ofbehavioural concerns (e.g.,Greenspan, 1992). One might reasonably expect that the detailsin drawings might pick up a child’s view of themselves, of their parents, of how close ordistant their relationship might be, and on how strongly they identified with their parents.

Page 3: Children's behaviour and their graphic representation of parents and self

L.C. Milne et al. / The Arts in Psychotherapy 32 (2005) 107–119 109

Secondly, children’s figure drawings evidence some signs that consistently indicate clinicalfeatures, although the above studies had some difficulty in demonstrating this for individualdrawing indictors. However, some examples might be the omission of arms indicatinginadequacy and helplessness (Jolles, 1971); the presence of teeth indicating compensationfor anxiety about receiving and expressing affective needs; a figure leaning or slantingindicating insecurity about one’s identity (Leibowitz, 1999); and lack of differentiation inthe mother–child drawing indicating symbiosis (Gillespie, 1994).

The present study measured differences between a child’s drawing of him or herselfand that of mother and father, all figures being drawn separately and not in relation toeach other. We hypothesised that the presence in children’s drawings of similarities anddifferences between themselves and each of their parents would relate to correlates in theirbehaviour. We further reasoned that most children would experience some ambivalencetowards both mother and father, which would be reflected in the interaction of similaritiesand differences detectable in the drawings. Factors such as dependency on or identificationwith one or the other parental figure might influence a child’s tendency to draw themselvesas similar to their parent, and attempts at individuation or negative reactions to one or otherparental figure might influence the tendency to draw themselves as different from theirparent.

Method

Subjects

Subjects were 75 children, 33 males and 42 females, aged between 5 and 13 years.For males, the mean age was 10.22 years (S.D. = 2.05 years) and for females the meanage was 10.12 years (S.D. = 1.69 years). The mean age of the overall sample was 10.17years (S.D. = 1.85 years). The effective sample was reduced to 22 males and 34 females,as a result of questionnaires not being returned. There was no significant age differencebetween males and females. All children were attending primary school. All children ofthem were Australian-born and spoke English as their first language, and English was thepredominant language spoken at home. Beyond this, questions about ethnicity were notasked. However, the schools from which the children were selected have children froma range of European, British, and Australasian backgrounds. The parents completed thequestionnaires in English.

Procedure

Subjects were selected from several primary schools in suburban Melbourne. We wroteto parents and asked whether they were interested in participating in the project. We sentthose that gave consent for them and their child to participate the Behaviour AssessmentSystem for Children (BASC) parent rating scales, appropriate to their child’s age (Reynolds& Kamphaus, 1992) and the Child Behaviour Check List (CBCL;Achenbach, 1991). Themother completed the questionnaires in all cases, although we did not specifically requestthis. Children whose parent gave consent for them to participate were invited to do the

Page 4: Children's behaviour and their graphic representation of parents and self

110 L.C. Milne et al. / The Arts in Psychotherapy 32 (2005) 107–119

drawing component of the task, which was completed at school during class time. We gavethese children a pack containing some sheets of plain white A4 paper, the shorter sidetowards them, and a standard set of 12 coloured pencils and a grey lead pencil. The childrenwere free to orient the paper in any direction they wanted. They were then asked by theresearcher to draw seven pictures. All participants were given the task in the same order:the first picture was a free picture where participants were invited to “draw anything youwant to”; for the second picture participants were asked to “draw a picture of a person”; forthe third “a picture of yourself”; for the fourth “a picture of your mother”; for the fifth “apicture of your father”; for the sixth “a picture of a house”; finally, for the seventh picture,participants were asked to draw “a picture of a tree.” Questions about how to draw, e.g.,“should I use colour?”, “can I draw me doing. . .?” etc., were answered neutrally, invitingthem draw the pictures in any way they wanted. Only drawings of the child themselves andtheir mother and father were used because the present study was part of a larger researchproject.

Drawing scoring

The drawings were scored by a registered psychologist experienced in the use of drawingsfor clinical purposes as well as for research. The psychologist was blind to the age andgender of participants and did not participate in the data collection or the scoring of thequestionnaires. The exception to this was the drawing of the self, where in many casesthe gender of the drawer was apparent. Children were asked not to name their drawings.Once all of the data was collected, the drawings and questionnaires were numbered witha unique identifier. A list of drawing details was compiled (seeTable 1), some of whichare self-explanatory, i.e.,Head, Hair, Eyes, Eye brows, Eye lashes, Nose, Mouth, Teeth,Facial hair, Jewellery, Lips,Arms, Legs,Hands. If any of these details were present in bothdrawings, i.e., in both the drawing of the parent and that of the child, 1 point forsimilaritywas given. If a detail was present in one but not the other, 1 point fordifferencewas given.In addition, if a detail was present in both drawings under comparison, but clearly drawndifferently, 1 point fordifferencewas given.

Table 1Criteria for scoring drawings as “similar” or “different”

Drawing detailsHeight TeethColour Facial hairSame colour Torso shapeHead ArmsHair LegsEyes HandsEye brows ClothesEye lashes JewelleryLips Activity/postureNose SymmetryMouth Shading

Page 5: Children's behaviour and their graphic representation of parents and self

L.C. Milne et al. / The Arts in Psychotherapy 32 (2005) 107–119 111

Height

If the height of drawings under comparison were within 1 cm of each other drawings 1point was given forsimilarity, otherwise 1 point was given fordifference.

Colour

If the participants did not use colour for both drawings 1 point was given for theirsimilarity. If they had used colour for both drawings they were also given 1 point for theirsimilarity; if they used colour in one drawing but not the other but not in the other, the twodrawings were considered to contain colour as a unique detail.

Same colour

If both drawings under comparison contained colour, but the colours in each drawingwere different, 1 point for thissimilaritywas given for the variableColouras just described,but 1 point fordifferencewas given for use of different colours.

Torso

If the torsos in both drawings being compared were the same, 1 point was given forsimilarity, if they were not, 1 point was given fordifference.

Clothes

If the clothes in two drawings being considered were the same garments and drawnsimilarly on each figure, e.g., a participant drew both figures wearing a skirt and top of thesame style, 1 point forsimilaritywas given, otherwise 1 point was given fordifference.

Activity/posture

If the two figures being compared had been drawn in the same posture, or doing thesame thing, e.g., standing with their hands behind their backs, 1 point was awarded forthis similarity. If the figures were standing in different postures, or engaged in differentactivities, e.g., one playing with a ball the other not, a point was given fordifference.

Symmetry

Some drawings tended to lean to one side, or were less well formed on one side, whileother drawings were more or less symmetrical. If both drawings being compared showedsymmetry or were asymmetrical in the same way, e.g., they were lop-sided in the samedirection, the two were considered similar and 1 point awarded forsimilarity. In cases inwhich symmetry differed, the two drawings were considered different and awarded 1 pointfor difference.

Page 6: Children's behaviour and their graphic representation of parents and self

112 L.C. Milne et al. / The Arts in Psychotherapy 32 (2005) 107–119

Shading

If the shading was of similar grade, intensity and neatness in two drawings being com-pared, they were judged similar in that respect and 1 point was given forsimilarity. Otherwise1 point was given fordifference.

Four variables were created from the drawings:Child–Mother Similarities(C-MSimil) Child–Mother-Differences(C-M Diff), Child–Father Similarities(C-F Simil), andChild–Father-Differences(C-F Diff).Child–Mother Similarities(C-M Simil) andChild–Mother-Differences(C-M Diff): The

drawing of the child themselves and that of their mother were examined for each of thevariables inTable 1. The first criterion for scoring was that the detail be present in bothdrawings. If so, one point was given toC-M Simil for this similarity, if not one point wasgiven toC-M Diff for thisdifference. For each variable, the score was the sum of the pointsgiven to each detail and variable. An identical procedure was used to calculateC-F SimilandC-F Diff.

Two further variables were created to examine the unique details of the drawings.Child–Mother Unique(C-M Unique): The child’s drawing of themselves was comparedwith that of their mother. When a particular detail was drawn in one, but not in the other(e.g., if one figure was drawn with a nose, and the other figure is drawn without a nose) onepoint was given toC-M Unique, otherwise zero was given for that detail. This conditionwas, in fact, one of the criteria forC-M Diff andC-F Diff; however, we thought it may alsobe important to examine unique features in a separate variable. The score was the sum totalof unique details. An identical procedure was used to calculateChild–Father Unique(C-FUnique).

Instruments

Behaviour Assessment System for Children (BASC;Reynolds & Kamphaus, 1992) isa system for evaluating children aged 4–18 years of age. We elected to use the parentform that gathers descriptions of children’s observable behaviour. One advantage of thismethod is that it providesT-scores for both clinical scales (Aggression, Anxiety, AttentionProblems, Atypicality, Conduct Problems, Depression, Hyperactivity, Somatization, andWithdrawal) and adaptive scales (Adaptability, Leadership and Social Skills). The parentform for children (6–11 years) consists of 138 questions with four-choice responses—never,sometimes, often, always—with scores ranging from 0 to 3. It takes from 10 to 20 min tocomplete. The parent form for adolescents (12–18 years) consists of 126 questions, withthe same questions with four-choice response format.

Child Behaviour Checklist (CBCL;Achenbach, 1991) is a checklist completed by par-ents of 4–18-year-old children. It contains some introductory questions about the child’sparticipation in sport, hobbies, organisations, chores, friendships and family relationships,and school performance, and gives parents an opportunity to express any concerns that theymight have about their child and lastly asks parents to describe the best things about theirchild. The quantitative part of the CBCL consists of 113 items, or problems, which the par-ent endorses as either: 0, not true; 1, somewhat or sometimes true; or 2, very true or oftentrue. For this study, we examined the syndrome and total problems scales only. There are

Page 7: Children's behaviour and their graphic representation of parents and self

L.C. Milne et al. / The Arts in Psychotherapy 32 (2005) 107–119 113

three internalizing subscales:Withdrawn; Somatic Complaints; Anxious/Depressed. Thereare two externalizing subscales:Delinquent BehaviourandAggressive Behaviour. Thereare four other subscales:Social Problems; Thought Problems; Attention Problems; andSexProblems. Finally, theTotal Problems Scale, is the sum total of all problems endorsed.T-scores were used to enable data from males and females and difference age groups to becompared (Achenbach, 1991).

Results

The descriptive statistics for the BASC and CBCL are shownTable 2. Independentsamplest-tests were run to determine gender differences. Males and females differedsignificantly on two variables only: BASC-Atypicality (p= 0.016), where males obtainedsignificantly higher scores, and CBCL-Anxiety/Depression (p= 0.001), where males alsoobtained significantly higher scores. The descriptive statistics for the drawing variables

Table 2Descriptive statistics for behavioural variables

Variable Females (N= 34) Males (N= 22)

M S.D. M S.D.

BASC-Hyperactivity 46.82 11.26 51.00 11.41BASC-Aggression 48.32 10.00 54.70 9.58BASC-Conduct Problems 48.65 9.11 53.61 9.74BASC-Anxiety 45.65 9.18 49.78 8.49BASC-Depression 47.53 9.33 49.91 11.81BASC-Somatization 48.41 9.93 47.17 8.47BASC-Atypicality 44.56 5.64 47.91 8.54BASC-Withdrawal 46.32 9.03 47.83 9.47BASC-Attention Problems 47.35 11.28 55.13 10.97BASC-Adaptability 50.91 11.39 47.22 11.77BASC-Social Skills 52.59 9.22 47.35 8.78BASC-Leadership 54.44 8.15 49.04 9.46BASC-Externalizing Composite 47.74 10.45 53.61 9.96BASC-Internalizing Composite 46.41 9.55 48.78 10.17BASC-Behavioural Symptoms Index 44.09 12.20 52.52 11.48BASC-Adaptive Skills Composite 51.06 12.25 47.04 8.30

CBCL-Withdrawal 53.38 5.71 55.22 6.91CBCL-Somatic 56.26 7.28 54.83 6.81CBCL-Anxiety/Depression 53.68 4.75 59.35 8.23CBCL-Social Problems 55.35 9.10 56.13 6.95CBCL-Thought Problems 52.12 4.44 53.87 6.29CBCL-Attention Problems 53.65 5.80 56.52 8.73CBCL-Delinquency 55.47 7.47 55.87 5.59CBCL-Aggression 54.41 7.54 57.57 6.36CBCL-Internalizing 50.62 10.64 54.83 10.96CBCL-Externalizing 50.62 10.68 54.65 7.69CBCL-Sex Problems 52.32 5.85 53.65 7.20

Page 8: Children's behaviour and their graphic representation of parents and self

114 L.C. Milne et al. / The Arts in Psychotherapy 32 (2005) 107–119

Table 3Descriptive statistics for drawing variables

Variable Females (N= 34) Males (N= 22)

M S.D. M S.D.

C-M Simil 6.40 2.88 5.80 2.33C-M Diff 5.56 2.85 5.87 2.69C-M Unique 0.70 0.85 1.00 1.14

C-F Simil 6.44 2.19 5.35 2.68C-F Diff 6.23 2.26 7.26 2.74C-F Unique 0.88 0.79 1.23 1.20

are shown inTable 3. Independent samplest-tests showed only one gender difference,C-FUnique (p= 0.008), drawings of boys of themselves showing significantly more uniquefeatures in relation to their drawings of their fathers.

Partial correlation analyses were conducted to test the relationships between the BASCand CBCL behavioural variables and the drawing measures of similarities in the drawingsof the self and mother (C-M Simil) and self and father (C-F Simil). We expected gender toaccount for some of the differences and similarities in the drawings of the child and parent(i.e., girls would be more likely to draw themselves as being more similar to their mother,and boys more like their father). For this reason, the genders were analysed separately. Inorder to account for possible developmental differences, such as younger children havingfewer overall details on the drawings, we also controlled for age, by entering age, in months,as a control variable in the partial correlation analysis. Further, in analysingC-M Similwecontrolled forC-M Diff for the following reason. If the presence of ambivalence in a child’sinternal representation of his or her mother was reflected in drawings in terms of the differ-ences and similarities between him or herself and mother, controlling for differences, whenanalysing similarities, might effectively assess the strength of a child’s identification withher. We proposed an analogous rationale for child–father similarities. We also controlledfor features of the drawings that were unique to either mother or child (C-M Unique) on thegrounds that these details were not caught up in a child’s dynamic efforts to identify withor differentiate him or herself from his or her parent. Again, similar reasoning was appliedto the analysis of child–father similarities.

In sum, the BASC and CBCL behavioural measures were correlated withC-M Similcontrolling for Age,C-MDiffandC-MUnique. The BASC and CBCL behavioural measureswere correlated withC-F Similcontrolling for Age,C-F Diff andC-F Unique. The resultsof the partial correlation analyses are shown inTables 4 and 5.

To account for the problem ofα slippage due to conducting multiple statistical tests wedid two things. Firstly, we used the Bonferroni correction procedure, taking into accountthe number of tests conducted and anα level of 0.05 (Uitenbroek, 1997). Secondly, wehypothesized that children drawing themselves as similar to their parents would indicatetheir identification with their parents, which would show itself in the mother describingher child, as having positive behavioural characteristics. Her doing so would confirm thehypothesis. If the mother were to report negative behavioural characteristics of the childher doing so would disconfirm our hypothesis. Thus, all correlations cannot be treated as if

Page 9: Children's behaviour and their graphic representation of parents and self

L.C. Milne et al. / The Arts in Psychotherapy 32 (2005) 107–119 115

Table 4Partial correlation and significance between negative behavioural variables and mother/father similarities

C-M Similarities(controllingfor C-M Differences; C-MUnique; and Age)

C-F Similarities(controllingfor C-F Differences; C-FUnique; and Age)

Females(d.f. = 30)

Males(d.f. = 19)

Females(d.f. = 30)

Males(d.f. = 19)

BASC-Hyperactivity 0.039 (0.834) −0.529 (0.020) 0.025 (0.892) −0.485 (0.030)BASC-Aggression −0.140 (0.453) −0.312 (0.194) −0.104 (0.577) −0.333 (0.151)BASC-Conduct Problems −0.257 (0.163) −0.366 (0.123) −0.003 (0.985) −0.249 (0.289)BASC-Anxiety −0.059 (0.751) −0.251 (0.300) 0.153 (0.411) −0.387 (0.092)BASC-Depression −0.259 (0.160) −0.311 (0.196) 0.201 (0.279) −0.630 (0.003)BASC-Somatization −0.102 (0.585) −0.008 (0.974) 0.360 (0.046) −0.196 (0.409)BASC-Atypicality −0.192 (0.301) −0.069 (0.779) 0.113 (0.546) −0.318 (0.172)BASC-Withdrawal 0.235 (0.203) 0.102 (0.679) 0.286 (0.119) −0.254 (0.279)BASC-Attention Problems −0.033 (0.859) −0.517 (0.024) −0.015 (0.935) −0.697 (0.001)BASC-Externalizing Com-

posite−0.130 (0.487) −0.496 (0.031) −0.018 (0.924) −0.452 (0.045)

BASC-Internalizing Com-posite

−0.181 (0.330) −0.267 (0.269) 0.286 (0.118) −0.530 (0.016)

BASC-Behavioural Symp-toms Index

−0.094 (0.615) −0.387 (0.102) 0.121 (0.518) −0.616 (0.004)

CBCL-Withdrawal −0.312 (0.088) 0.036 (0.884) −0.065 (0.729) −0.317 (0.173)CBCL-Somatic −0.266 (0.148) 0.112 (0.649) 0.124 (0.507) −0.184 (0.437)CBCL-Anxiety/Depression −0.271 (0.140) −0.338 (0.158) 0.146 (0.432) −0.694 (0.001)CBCL-Social Problems −0.308 (0.091) −0.267 (0.253) 0.018 (0.922) −0.528 (0.017)CBCL-Thought Problems −0.120 (0.522) −0.064 (0.794) 0.002 (0.993) −0.489 (0.025)CBCL-Attention Problems −0.227 (0.220) −0.191 (0.434) 0.024 (0.896) −0.520 (0.019)CBCL-Delinquency −0.076 (0.684) −0.420 (0.073) −0.104 (0.578) −0.317 (0.174)CBCL-Aggression −0.135 (0.471) −0.191 (0.433) 0.025 (0.896) −0.434 (0.056)CBCL-Internalizing −0.242 (0.190) −0.282 (0.241) 0.214 (0.247) −0.476 (0.034)CBCL-Externalizing −0.196 (0.289) −0.594 (0.007) −0.008 (0.968) −0.523 (0.018)CBCL-Sex Problems −0.267 (0.147) −0.372 (0.116) −0.164 (0.377) −0.424 (0.063)CBCL-Total −0.229 (0.216) −0.428 (0.066) 0.096 (0.608) −0.623 (0.003)

Significance is parenthesized.

Table 5Partial correlation and significancea between positive behavioural variables and mother/father similarities

C-M Similarities(controllingfor C-M Differences; C-MUnique; and Age)

C-F Similarities(controllingfor C-F Differences; C-FUnique; and Age)

Females(d.f. = 30)

Males(d.f. = 19)

Females(d.f. = 30)

Males(d.f. = 19)

BASC-Social Skills 0.017 (0.927) 0.364 (0.125) 0.013 (0.945) 0.572 (0.008)BASC-Leadership 0.016 (0.931) −0.107 (0.663) −0.053 (0.776) 0.277 (0.236)BASC-Adaptability 0.171 (0.358) 0.346 (0.146) −0.003 (0.987) 0.572 (0.008)BASC-Adaptive Skills

Composite0.024 (0.900) 0.407 (0.084) −0.123 (0.511) 0.729 (<0.000)

a Significance is parenthesized.

Page 10: Children's behaviour and their graphic representation of parents and self

116 L.C. Milne et al. / The Arts in Psychotherapy 32 (2005) 107–119

only chance operated, so one cannot take only the null hypothesis into account (Perneger,1998). The presence of correlations indicating good behaviours is not enough to prove themas validly confirming the hypothesis, as there should be an absence of negative, behaviouralcharacteristics. The two correlations work against one another in order to confirm or dis-confirm the hypothesis. The Bonferroni correction does not seem to take account of thispsychologically meaningful relationship as pointed out byPerneger (1998)in a medicalcontext. Accordingly one is able to treat testing for positive behaviours as one set of tests,and testing for negative behaviours as a separate set of tests.

The Bonferroni adjustment, resulted in two separate reducedα levels as follows: forpositive behaviours the reducedα was 0.013 (4 tests,α level at 0.05); and for negativebehaviours the reducedα was 0.002 (24 tests,α level at 0.05). In order to explore possibilitiesin the data, we were also interested in the possible psychological meaning in correlationsgreater than 0.5.

Overall boys yielded more significant results than girls, for whom there were no corre-lations above 0.5 for eitherC-M Simil, norC-F Simil.

The picture for boys was strikingly different. For boys, there were negative correlationsabove 0.5 betweenC-M Simil and BASC-Hyperactivity, BASC-Attention Problems,BASC-Externalizing Composite, and CBCL-Externalizing Composite. There werealso negative correlations greater than 0.5 betweenC-F Simil and BASC-Depression,BASC-Attention Problems, BASC-Internalizing Composite, BASC-Behavioural SymptomsIndex, CBCL-Anxiety/Depression, CBCL-Social Problems, CBCL-Attention Problems,CBCL-Externalizing Composite, andCBCL-Total. There were positive correlations greaterthan 0.5 betweenC-F SimilaritiesandBASC-Adaptability, BASC-Social Skills, and theBASC-Adaptive Skills Composite.

When the Bonferroni adjustment to theα level was made, the results were more con-servative, obviously, with boys only obtaining significant results. For boys, there were nosignificant behavioural correlates, positive or negative, withC-MSimil, when the more con-servative estimate is made, although CBCL Externalizing came close (0.007).C-F Similcorrelated with the negative behaviours, one cognitive and the other anxiety-related, namely,theBASC-AttentionProblemsandCBCL-Anxiety/Depressionand the positive behaviour, theBASC-Adaptive Skills Composite.

Discussion

The results partially supported the hypothesis that similarities and differences inchildren’s representation of themselves and each of their parents would reflect theirbehaviour. Specifically, when age and the number of differences were controlled, thenumber of similarities in children’s drawings of themselves and each of their parents wasinversely related to “negative” behavioural characteristics, and to a lesser extent, positivelyrelated to more adaptive behaviours. However, this was only true for boys—there were nosignificant findings for girls.

For boys, fewer similarities between the drawing of themselves and their mothers wasassociated with a variety of negative externalizing behaviours, such as hyperactivity andinattention. This is a complex finding. For an adolescent boy to individuate he must separate

Page 11: Children's behaviour and their graphic representation of parents and self

L.C. Milne et al. / The Arts in Psychotherapy 32 (2005) 107–119 117

from his mother, in order to develop heterosexual interests outside his family (Blos, 1979).There is evidence that this process begins much earlier than early adolescence (Kaplan,1978). From this point of view, the greatest anxiety for the infant boy is that he might mergewith his mother, resulting in loss of a sense of self and some distortion to his masculineidentity. Seemingly mothers differentiate between their infant boy and girl as well, tendingto be more permissive of separation from their sons (Kaplan, 1978). In the present study,this moving away of boys from their mother may indeed have resulted in differences in theboys’ representations of themselves and their mothers, which were statistically controlled bypartialing out the effects ofC-MDiffs. If this were so, the number of similarities between theboys’ self and mother drawings may be indicative of their identification with their mothers,along with attempts at separation, which were reflected in differences in their self and motherdrawings. In all cases, mothers completed the BASC and the CBCL, so that their perspectiveof their sons formed the dependent measures. It may be that fewer similarities between aboy and his mother, as measured by his drawings, reflected his resistance toward her, whichmay indeed have appeared to her as his showing difficult behaviours. Conversely, boys whodepicted themselves, in some ways, as being like their mother may be more comfortablewith her, and may comply more easily with her, resulting in her inclination to see him ashaving fewer behaviour problems and more adaptive characteristics.

Lamb (1997), having reviewed the relative contributions of maternal and paternal care,concluded that mothers and fathers influence their children in similar rather than differentways. The crucial variables identified by Lamb as being associated with positive outcomeswere parental warmth, nurturance, and closeness, regardless of parent gender. Boys whodrew themselves and their fathers with similar features showed more positive behaviouralfeatures and fewer internalizing and externalizing problems. This suggests that boyswho identify with their fathers are more interpersonally constructive and competent thanboys who do not. This finding emphasizes the importance of fathers to sons.Rohner andVeneziano (2001)reviewed contemporary and earlier research literature on the role of fatherlove, defined as the extent of paternal acceptance or rejection. They concluded that fatherlove is as important, and in some case more important than mother love, in terms of infants’,children’s, and young adults’ psychological well-being, health, and behavioural difficulties.They demonstrated that paternal, as well as maternal, rejection has been associated withan array of adjustment problems, including low self-concept and self-esteem, social andemotional withdrawal, conduct problems, aggressive behaviour, substance abuse problems,academic difficulties, and clinical conditions such as depression and borderline personalitydisorder features (Rohner & Veneziano, 2001). If, as we hypothesized, similarities betweendrawings of the self and a parent reflect something of the relationship with that parent,then our findings support the notion that fathers have a very important role in the theirchildren’s well-being, for boys at least. Rohner and Veneziano identified studies that foundthat the positive father–child relationship was associated with a variety of specific effects,but concluded that this phenomenon was not as yet well understood.

The lack of significant results for girls was unexpected. It may simply mean that drawingsdo not have the same projective value in representing parent–child relationships for girls asthey appear to have for boys. In the past, we found that boys as they approach puberty stopusing colour in their drawings, probably as a means of inhibiting their emotional reactivity,but that girls continued to use colour (Milne & Greenway, 1999). At this stage it would

Page 12: Children's behaviour and their graphic representation of parents and self

118 L.C. Milne et al. / The Arts in Psychotherapy 32 (2005) 107–119

appear that for girls, comparison of the child’s self and parent drawing gives no insight intotheir behaviour.t-tests run between males and females on the behavioural variables showedonly two gender differences, and there was no evidence that girls were behaving any worseor better than boys. It does not seem then, that girls were behaving differently from boys,rather, that they do not graphically represent the parent–child relationship in the same wayas boys. The similarities and differences in drawings may reflect the possibility that thegenders express themselves in distinctly different ways. Boys for example may projectaspects of the parent–child relationship in their drawings, whereas girls may use a morestylized approach that does not vary much with the nature of the parent–child relationshipor does so only when clinical factors are present. Another possibility lies in the behaviouralvariables, from the BASC and the CBCL. As mothers completed both the BASC and theCBCL, a possible explanation is that mothers are more tolerant of their daughters’ attemptsto individuate, or form a negative identification with them, than they are of their sons’. Thatis, girls who draw fewer similarities in their parent and child drawings either do not projecttheir distancing from their mother as obviously as boys in their behaviours, or mothers areless concerned, and therefore, do not endorse their daughters’ behaviour as problematic.The genders probably express psychological distress in characteristically different ways,and it may be that the BASC and the CBCL, with their emphasis on observable behaviours,tend pick up more typically male characteristics, and perhaps research needs to look atvariables such as confidence, self-esteem, and more subtle social variables to better tapsgirls’ potential difficulties.

While this research is at an early stage, it is likely to have some clinical implications.One may speculate that the drawings of a boy referred for behavioural difficulties mayreveal something of his relationship with his father, and hence the potential source of theunderlying behavioural difficulty. There may also be potential to screen for boys at riskof behavioural problems using this drawing technique. The first step towards the clinicalusefulness of this method would be to establish a normative data set.

The present study has several limitations, firstly, due to the comparatively small samplesize, and the need to consider each gender separately, the results need to be replicated beforefirm conclusions can be drawn. Secondly, we managed the differences in age of the subjectsby statistically controlling for it. This may not be the most effective way of dealing withthis problem.Fury et al. (1997)andMadigan et al. (2003), for example used same-agechildren in their studies. However, even using same-age children does not fully accountfor developmental differences observed in drawings. We also relied on mother’s reportsof behaviour. While this is valid, and certainly gives information about the mother–childattachment relationship, it may not give the full story. All drawings were done in the samesequence for ease of group administration; however, a counterbalancing sequence wouldhave controlled for any effect of order. Consideration also needs to given to fathers’ reportsof behaviour. Future research may also need to look at other ways of assessing children’sbehaviour.

References

Achenbach, T. M. (1991).Manual for the child behaviour checklist/4–18 and 1991 profile. Burlington, VT:University of Vermont Department of Psychiatry.

Page 13: Children's behaviour and their graphic representation of parents and self

L.C. Milne et al. / The Arts in Psychotherapy 32 (2005) 107–119 119

Berryman, E. (1959). The self-portrait: A suggested extension of the HTP.Perceptual andMotor Skills,9, 411–414.Blos, P. (1979).The adolescent passage. New York: International Universities Press, Inc.Burns, R. C., & Kaufman, S. H. (1970).Kinetic family drawings (K.F.D): An introduction to understanding childrenthrough kinetic drawing. New York: Brunner/Mazel.

Fox, C., Davidson, K., Lighthall, F., Waite, R., & Sarason, S. B. (1958). Human figure drawings of high and lowanxious children.Child Development, 29, 297–330.

Fury, G., Carlson, E. A., & Sroufe, A. (1997). Children’s representation of attachment relationships in familydrawings.Child Development, 68, 1154–1164.

Gillespie, J. (1994).The projective use of mother-and-child-drawings: A manual for clinicians. New York: Brun-ner/Mazel.

Greenspan, S. I. (1992).Infancy and early childhood: The practice of clinical assessment and intervention withemotional developmental challenges. Madison, CT: International Universities Press.

Jolles, I. (1971).A catalogue for the interpretation of the H–T–P. Los Angeles: Western Psychological Services.Kaplan, L. (1978).Oneness and separateness: From infant to individual. London: Jonathan Cape Ltd.Karen, R. (1998).Becoming attached: First relationships and how they shape our capacity to love. Oxford: Oxford

University Press.Koppitz, E. M. (1968).Psychological evaluationof children’s humanfiguredrawings. New York: Grune & Stratton.Koppitz, E. M. (1984).Psychological evaluation of human figure drawings made by middle school pupils. New

York: Grune & Stratton.Lamb, M.E. (1997). The development of the father–infant relationship. In M. E. Lamb (Ed.),The role of the fatherin child development(3rd ed., pp. 104–120). New York: John Wiley & Sons, Inc.

Leibowitz, M. (1999).Interpreting projective drawings. New York: Brunner/Mazel.Madigan, S., Ladd, M., & Goldberg, S. (2003). A picture is worth a thousand words: Children’s representations

of family as indicators of early attachment.Attachment and Human Development, 5, 19–37.Main, M., Kaplan, N., & Cassidy, J. (1985). Security in infancy, childhood and adulthood: A move to the level of

representation.Monograph for the Society for Research in Child Development, 50, 66–104.Milne, L. C., & Greenway, P. (1999). Colour in children’s drawings: The influences of age and gender.Arts inPsychotherapy, 26, 261–263.

Oster, G. D., & Gould, P. (1987).Using drawings in assessment and therapy. New York: Brunner/Mazel.Perneger, T. V. (1998). What’s wrong with Bonferroni adjustments.British Medical Journal, 316, 1236–1238.Reynolds, C. R., & Kamphaus, R. W. (1992).Behaviour assessment system for children: Manual. Circle Pines,

MN: American Guidance Service, Inc.Rohner, R. P., & Veneziano, R. A. (2001). The importance of father love: Historic and contemporary evidence.Review of General Psychology, 5, 382–405.

Uitenbroek, D. G. (1997).SISA binomial. Southampton: D.G. Uitenbroek. Retrieved January 01, 2004, from theWorld Wide Web:http://home.clara.net/sisa/binomial.htm.