shy girls and boys: a new look

8
Pergamon 0021-9630(95)00066-6 J. Child Psychoi. Psychial. Vol. 37. No. 2. pp. 181-187. 1996 Eisevier Science Ud © 1996 Association for Child Psychology and Psychialry Printed in Great Britain. All rights reserved 0021-9630/96 $15.00 + 0.00 Shy Girls and Boys: A New Look Joan Stevenson-Hinde and Ann Glover University of Cambridge, U.K. A new look at shyness in young children involves selecting for shyness, using criteria involving both natural and laboratory contexts, categorizing children (including a group between the two extremes), and including the sex of the child in analyses. Children of both sexes differed significantly across low, medium, and high shy groups, with riegative mood, wonies and fears, and problem behaviour in preschool being associated with high shyness. Although boys in general and high shy boys in particular had the highest problem behaviour scores in preschool (including acting out behaviour), maternal interactions with high shy boys were positive: significantly more positive than with boys who were high shy in natural settings but low shy in the laboratory, or with high shy girls. Of the girls, maternal style was most positive for the medium shy girls, who also received the highest relative frequency of positive maternal actions during a brief task—significantly higher than medium shy boys, as well as high shy girls. Keywords: Shyness, preschool behaviour checklist, mother/child interaction, sex differences While many preschool-aged children show initial wariness on meeting a stranger, extreme shyness may be indicative of concurrent problems and, in the absence of any preventive intervention, subsequent disorder (see, e.g. Rubin, 1993). However, since a normative sample will contain only a small proportion of extremely shy children, studies of shyness in such samples may reveal less than a complete picture, particularly if the extremes are hidden within the whole group and linearity is assumed, as for example in a correlational approach. Indeed, a strong case has been made for not regarding "the difference between inhibited and uninhibited children . . . as a continuous dimension" (Kagan, Snidman & Arcus, 1993, p. 24). The present study involves a community sample, selected to contain a high proportion of shy children. On the basis of criteria involving both natural and laboratory contexts, children were categorized, to include a medium shy group as well as the extremes. Firstly, a profile of each group will be considered, in terms of negative mood, worries and fears, and problem behaviour in playgroup. Although worries and fears did not increase significantly with shyness in a normative sample (Stevenson-Hinde & Shouldice, 1995b), clinical studies indicate that extreme shyness should be asso- ciated with negative mood (e.g. Anderson, 1994). While 5-year-old children identified observationally as "so- cially withdrawn" were not rated by their teachers as Requests for reprints to: Dr J. Stevenson-Hinde, Medical Research Council Group on the Development and Integration of Behaviour, Madingley, Cambridge CB3 8AA, U.K. problematic (Rubin, 1993), self-reported anxious symp- toms of 6-year-olds were related to poorer functioning in a school setting (e.g. Ialongo, Edelsohn, Werthamer- Larsson, Crockett & Kellam, 1995). It is therefore possible that extremely shy children would have problems in pre-school. Secondly, we shall ask how mother/child interactions differ across the groups. In unselected samples, both behavioural interactions with others (e.g. Simpson & Stevenson-Hinde, 1985; Stevenson-Hinde & Hinde, 1986) and parental beliefs (e.g. Mills & Rubin, 1993) indicate that shy girls appear to be more acceptable than shy boys. However, it remains to be seen whether this difference holds for extreme shyness. Here, mother/child interactions will be assessed in two ways: (1) obser- vations of unstructured interactions over an entire home visit; and (2) in a second home visit, observations immediately following a particular challenge. This opens a "window" into how the pair operate together, as during reunions in a strange situation (Ainsworth, Blehar, Waters & Wall, 1978; see also, Stevenson-Hinde & Shouldice, 1995a for the present age group), or upon asking mothers to issue toddlers with the instruction "Clear up the toys now" (Achermann, Dinneen & Stevenson-Hinde, 1991). Subjects Following screening (see below), the sample involved 126 English-speaking mothers and their children (68 girls, 58 boys), recruited through playgroups (which the children had been attending for a mean duration of 15 months). All mothers were married and living in Cambridge or surrounding villages. Their ages ranged from 23 to 44 years, with a mean of 34 181

Upload: joan-stevenson-hinde

Post on 02-Aug-2016

219 views

Category:

Documents


6 download

TRANSCRIPT

Page 1: Shy Girls and Boys: A New Look

Pergamon

0021-9630(95)00066-6

J. Child Psychoi. Psychial. Vol. 37. No. 2. pp. 181-187. 1996Eisevier Science Ud

© 1996 Association for Child Psychology and PsychialryPrinted in Great Britain. All rights reserved

0021-9630/96 $15.00 + 0.00

Shy Girls and Boys: A New Look

Joan Stevenson-Hinde and Ann Glover

University of Cambridge, U.K.

A new look at shyness in young children involves selecting for shyness, using criteriainvolving both natural and laboratory contexts, categorizing children (including a groupbetween the two extremes), and including the sex of the child in analyses. Children of bothsexes differed significantly across low, medium, and high shy groups, with riegativemood, wonies and fears, and problem behaviour in preschool being associated with highshyness. Although boys in general and high shy boys in particular had the highest problembehaviour scores in preschool (including acting out behaviour), maternal interactions withhigh shy boys were positive: significantly more positive than with boys who were high shyin natural settings but low shy in the laboratory, or with high shy girls. Of the girls,maternal style was most positive for the medium shy girls, who also received the highestrelative frequency of positive maternal actions during a brief task—significantly higherthan medium shy boys, as well as high shy girls.

Keywords: Shyness, preschool behaviour checklist, mother/child interaction, sexdifferences

While many preschool-aged children show initialwariness on meeting a stranger, extreme shyness may beindicative of concurrent problems and, in the absence ofany preventive intervention, subsequent disorder (see,e.g. Rubin, 1993). However, since a normative samplewill contain only a small proportion of extremely shychildren, studies of shyness in such samples may revealless than a complete picture, particularly if the extremesare hidden within the whole group and linearity isassumed, as for example in a correlational approach.Indeed, a strong case has been made for not regarding"the difference between inhibited and uninhibitedchildren . . . as a continuous dimension" (Kagan,Snidman & Arcus, 1993, p. 24). The present studyinvolves a community sample, selected to contain a highproportion of shy children. On the basis of criteriainvolving both natural and laboratory contexts, childrenwere categorized, to include a medium shy group as wellas the extremes.

Firstly, a profile of each group will be considered, interms of negative mood, worries and fears, and problembehaviour in playgroup. Although worries and fears didnot increase significantly with shyness in a normativesample (Stevenson-Hinde & Shouldice, 1995b), clinicalstudies indicate that extreme shyness should be asso-ciated with negative mood (e.g. Anderson, 1994). While5-year-old children identified observationally as "so-cially withdrawn" were not rated by their teachers as

Requests for reprints to: Dr J. Stevenson-Hinde, MedicalResearch Council Group on the Development and Integrationof Behaviour, Madingley, Cambridge CB3 8AA, U.K.

problematic (Rubin, 1993), self-reported anxious symp-toms of 6-year-olds were related to poorer functioning ina school setting (e.g. Ialongo, Edelsohn, Werthamer-Larsson, Crockett & Kellam, 1995). It is thereforepossible that extremely shy children would haveproblems in pre-school.

Secondly, we shall ask how mother/child interactionsdiffer across the groups. In unselected samples, bothbehavioural interactions with others (e.g. Simpson &Stevenson-Hinde, 1985; Stevenson-Hinde & Hinde,1986) and parental beliefs (e.g. Mills & Rubin, 1993)indicate that shy girls appear to be more acceptable thanshy boys. However, it remains to be seen whether thisdifference holds for extreme shyness. Here, mother/childinteractions will be assessed in two ways: (1) obser-vations of unstructured interactions over an entire homevisit; and (2) in a second home visit, observationsimmediately following a particular challenge. This opensa "window" into how the pair operate together, asduring reunions in a strange situation (Ainsworth,Blehar, Waters & Wall, 1978; see also, Stevenson-Hinde& Shouldice, 1995a for the present age group), or uponasking mothers to issue toddlers with the instruction"Clear up the toys now" (Achermann, Dinneen &Stevenson-Hinde, 1991).

SubjectsFollowing screening (see below), the sample involved 126

English-speaking mothers and their children (68 girls, 58boys), recruited through playgroups (which the children hadbeen attending for a mean duration of 15 months). All motherswere married and living in Cambridge or surrounding villages.Their ages ranged from 23 to 44 years, with a mean of 34

181

Page 2: Shy Girls and Boys: A New Look

182 J. STEVENSON-HINDE and A. GLOVER

years. All fathers were employed, with 60% in occupationalclasses 1 or 2 (professional/intermediate), 34% in class 3(skilled), and 6% in 4 or 5 (unskilled; Classification ofOccupations, 1970).

Assessments

When a child was 4-4.5 years old, each mother wasinterviewed at home and completed several questionnaires,including items concerning her child's shyness. A child wasincluded in the study if he/she (1) met shy criteria, or (2) metnot-shy criteria and also matched a shy child for sex, place infamily, mother's age and father's socio-economic status. Asecond home visit, followed by a lab visit, occurred within afew weeks of each other.

Screening for Approach/Withdrawal to People

Mothers completed the McDevitt and Carey (1978)Temperament Questionnaire. Screening items concernedinitial approach/withdrawal to unfamiliar people: the threerelevant items from the questionnaire (items 31, 43 and 86),and three of our own inserted into the questionnaire (The childplays readily with a new child. The child waits for a strangechild to make the first approach. The child speaks readily tonew adults). Since ratings could range from 1 to 6, a mean a4over the above items was considered to be high, and a means3.5 was considered to be not-high. In addition, these maternalratings had to be supported by interviewer's ratings (similar tothe laboratory ratings, see below) during the first home visit—both upon arrival and after interviewing mother for about 45min, when the interviewer invited the child to come and sit byher for a story.

Shyness

Independently of the above screening, shyness was assessedin three different ways in three different contexts.

Temperament questionnaire to mothers and fathers. TheTemperament Assessment Battery for Children (TABC:Martin, 1988) was given to mothers and to fathers, at differenttimes to ensure independence. It has eight particularlyappropriate approach/withdrawal items, concerning newpeople or places (rather than "new foods" or "new things"as with the McDevitt & Carey scale). Ratings vary from 1("hardly ever") to 7 ("almost always").

Observer ratings in laboratory. Upon visiting the laboratoryfor the first time, and after a brief introduction, each mother/child pair was left on their own in a room with chairs, a smalltable, and toys. After 3 min, a woman they had never seenbefore entered and talked briefly with the mother. She theninvited the child to approach her to see a toy in her hand,followed by an invitation to sit at a table for the PeabodyPicture Vocabulary Test (PPVT).

A rating was made from videotape—from the child's beinginitially addressed by the stranger through the first 30 sec ofthe PPVT. Both verbal responsiveness (replies, initiations,extensions) and inhibited behaviour (physical/visual avoidanceand tension in the face, body or voice) were rated, and then anoverall rating was made on 9-point scale, from 1 (relaxed/responsive) to 4 (a norm for the age) to 9 (high tension/noverbal response). Ratings were made by someone (N. Direkze)who was blind to all the other assessments. Inter-observerreliability was r = .81, A = 20. (For further analyses involvingthis dimension in a different, longitudinal sample, seeStevenson-Hinde & Shouldice, 1995b).

Temperament scate to playgroups leaders. Playgroupleaders completed the EAS Temperament Survey for Children

(Buss & Plomin, 1984), which consists of 20 items, each ratedon a 5-point scale. Only the Shyness scale, consisting of fiveitems, is used here.

Negative Mood

Temperament questionnaire to mothers. The McDevitt andCarey (1978) Temperament Questionnaire contains 12 mooditems, rated from 1 to 6 and pertaining to mood in a variety ofsituations, including being disciplined, being tired or ill, andplaying with peers (the TABC has no mood scale).

Observer ratings at home. At the end of the first home visit,the interviewer rated the child's mood over the whole visit on ascale ranging from 1 (unhappy, irritable or solemn) to 4(typical for age) to 7 (happy, cheerful nearly all the time withany low mood a brief exception and the result of some knownevent). These ratings were reversed, to indicate "negativemood". For inter-observer reliability, a second observer (A.Tamplin) sat passively in the background for 12 of the homevisits (r = .97, N = 12).

Observer ratings in the laboratory. A rating was made fromvideo-tape over the same period and by the same rater as forthe above laboratory assessment of shyness. The originalratings (which were later reversed to indicate negative mood)ranged from 1 (clear and prolonged negative mood) to 4(neutral) to 7 (clear and prolonged positive mood). Inter-observer reliability was r = .86 (N = 20).

Worries and Fears

From a list of common childhood fears and worries, motherswere asked to indicate the degree of fear or worry on scalesfrom 0 to 2, which were then summed (see also Stevenson-Hinde & Shouldice, 1995b).

Problem Behaviour in Playgroup

The Preschool Behaviour Checklist (McGuire & Richman,1986) was also sent to playgroup leaders. This consists of 22items relating to problem areas (e.g. unclear speech, wetting,poor concentration), each marked for frequency of occurrenceand then assigned a 0 ,1 , or 2. Scores may therefore range from0 to 44, with a score of 12 or more indicating problemsneeding attention. Of the 22 items, seven clearly relate towithdrawal problems: #9 (reluctant to speak), #11 (rarely playswith others), #13 (easily upset), #15 (unoccupied), #17(tearful), #19 (emotionally withdrawn), and #21 (very fearful);and six to acting out problems: #6 (difficult to control), #10(temper tantrums), #14 (fighting), #16 (interferes with work/play of peers), #18 (teases/spiteful), #20 (destructive).

Mother/Child Interactions at Home

Maternal style over the first home visit. The first visit,which lasted for approximately an hour, involved interviewingthe mother and then talking with the child. No instructionswere given as to what they should/should not do together.Mother-child interactions were observed throughout, with aview to completing rating scales relating to the provision of asecure base (from Stevenson-Hinde & Shouldice, 1995a).Immediately after each visit, ratings were made for each of thefollowing six dimensions, on a scale from 1 to 7, with 4 as themode for the whole sample:

Positive mood (unhappy/irritable/worried—^to—happy/cheerful);

Sensitive (inappropriate/ignoring—to—appropriate re-sponses to child's signals);

Page 3: Shy Girls and Boys: A New Look

SHY GIRLS AND BOYS 183

Gentle (sudden, over-intense voice/rough physical con-tact—to—soft voice/gentle physical contact);

Meshing (intense, unpleasant friction—to—smooth, co-ordinated interactions towards mutual goals);

Enjoy child (negative comments/attitude toward child—to—enjoyment of child's company);

Relaxed (tense/uncomfortable—to—positive, 'secure'atmosphere generated by mother).

These ratings were converted to z-scores and averaged toobtain an overall index of maternal style. Over the wholesample, intercorrelations among the six dimensions rangedfrom r = .22 to .72.

For inter-observer reliability, a second observer (A.Tamplin) sat passively in the background for 12 of the homevisits. Reliabilities were: r = .89 for positive mood, .89 forsensitive, .92 for gentle, .97 for meshing, .91 for enjoy child,and .87 for relaxed.

Family Drawing Task

During the second home visit, a video camera wasintroduced. The child was given a large piece of paper and abox of coloured crayons and asked to "Draw a picture of yourfamily". The observer suggested that mother sit with the childand talk him/her through the drawing. The first 3 min ofinteractions were coded from video-tape (by AG). The meanfrequency of occurrence of interactions or "turns" over thisperiod was 64.6 (SD = 16.4; range = 27-113). A "turn" wasdefined as a 3 sec pause, or a clear-cut change in vocalinflection, or a change of topic or activity. Each "turn" wasassigned to one and only one of the following types ofinteraction:

Child's positive interactions, with examples:

Answer questionsComment on drawing

Enquire

Smile/nod/hum

"He IS taller than I am."We are all holding hands."

"What colour are my eyes?"

lild's negative interactions:

Rebel

Refuse

Ignore

Indicate anxiety/helplessness

Look anxious

Bang crayon on table.

"No!"

"I can't draw people"—you do it.

Mother's positive interactions:

Guide "Now who will you drawfirst?"

Reinforce "That's really good."

Reply "Yes, he has got a beard."

Smile/nod

Gentle physical contact Cuddle, pat.

Mother's negative interactions:Criticise "That head is much too

big."Interfere/interrupt "Here, let me do it for you."

"Hurry up, we haven't gotall day."

"Just do one more and then

Show impatience

Cajoleyou can have somecrisps."

Frown/sneer

Mother's neutrat interactions:Instruct "Draw his feet on the end of

his legs."To obtain 'relative frequencies' (Table 2), the interactions

within each of the above sets were combined and thenexpressed as the proportion of total child or total motherinteractions, as appropriate. For reliability assessment, anothertrained person (J. Early) coded a randomly selected sub-sample (N = 16). Inter-observer reliabilities for the above fiveheadings ranged from r = .75 to .99.

Results

Shyness and Negative Mood—Correlations

When boys and girls were considered separately,correlations were all positive, with neither sex producingconsistently higher coefficients than the other. Indeed,for mothers' ratings the correlation between shyness andnegative mood was exactly the same for girls as for boys(r = .43, N = 56 and 66, respectively). Thus, boys andgirls will be considered together here (Table 1).

Firstly, intercorrelations involving shyness were all£.50 and highly significant, even though some involved

Table 1Pearson Correlations Involving Ratings of Shyness and Negative Mood

Shyness Negative mood

M Lab Obs M Home Obs

Shynessby motherby fatherby lab observerby playgroup leader

Negative moodby motherby home observerby lab observer

.86***

.54****

.73***

.43***

.63***

.23*

—.50***.62***

.32**

.57***

.27**

—.52***

.21*

.17

.46***

.29**

.30***

.17.27**.07 .18*

N = 119-126.* p< .05, ** p< .01, *** p < .001, two tailed.

Page 4: Shy Girls and Boys: A New Look

184 J. STEVENSON-HINDE and A. GLOVER

different contexts and different instruments. Unusuallyhigh agreement occurred between mothers and fathers(r = .86), possibly reflecting the clarity of the TABCitems. Intercorrelations involving negative mood wereall considerably lower, ranging from r = .07 to .27 (Table1).

Secondly, shyness was significantly correlated withnegative mood, whether the correlations involved thesame rater (i.e. mother or lab observer) or different raters(i.e. mother vs home observer; father vs home observer).Furthermore, correlations involving different raters anddifferent settings (e.g. mother vs playgroup leader) werenearly all significant, although less high.

Children's Feelings and Interactions According toShyness Status

Children were divided into categories of shyness—high, medium, low, and mixed. To be in the high group,a child had to screen as shy at home (as described above)AND in the laboratory test receive the highest shynessratings (7-9). To be in the low group, a child hadto initially screen as "not shy" at home AND receivethe lowest shyness ratings in the laboratory(1-3). The medium shy group consisted of all childrenwho received laboratory ratings at the "norm" of 4 orslightly higher (4-6) in the laboratory, regardless of thehome category.

The remaining children were "mixed": initiallyscreened as "shy" at home but then receiving the lowestlaboratory ratings (1-3) or vice-versa, i.e. "not shy" athome but high laboratory ratings (7-9). This "mixed"group contained only nine children, and will beconsidered in a separate section. Here, the threegroups—low, medium and high shy—^will be comparedby means of two-way ANOVAs (shyness group x sexof child; see Table 2).

Negative mood. Whether mood was assessed bymothers on the temperament questionnaire, by theinterviewer in the first home visit, or by the observerin the laboratory test, a significant overall groupdifference occurred, with high shyness associated withnegative mood. Negative mood was always highest forhigh shy boys, although in pairwise comparisons, theynever differed significantly from high shy girls.

Worries and fears. Similarly, there was a significantgroup effect for worries and fears, as reported bymothers, with the highest values associated with highshyness.

Problem behaviour in playgroup. There were sig-nificant main effects according to both group and sex,with problems associated with high shyness and being aboy. Of the nine high shy boys, three had a totalPreschool Behaviour Checklist (PBCL) score above thecut-off for definite problems (al2). In between-groupcontrasts, more behaviour problems were reported forthe high shy boys than for each of the other fivesubgroups. The high shy girls showed significantly moreproblem behaviour than the medium shy girls.

Of the variety of problems indicated by the 22 PBCLitems, seven clearly relate to withdrawal and six toacting out problems (as indicated in the methodssection). For each set, item scores were summed and

divided by the number of items, to make the setscomparable, i.e. a possible range from 0 to 2, as in theoriginal scoring for each item. As with the PBCL total,scores involving the withdrawal set of items showed thehigh shy group differing significantly from both themedium and low groups, with medium and low notdiffering significantly. Unlike the PBCL total, with-drawal scores for high shy girls were similar to high shyboys, and the main effect for sex was not significant (F =.87, p = .35).

The sex difference in the PBCL total was reflected inthe acting out items, where boys were significantlyhigher than the girls, and the interaction effect wasnearly significant (F = 2.89, p = .06). That is, while thelow shy boys and girls had equal acting out scores, thehigh shy boys' scores were significantly higher than thehigh shy girls' scores (p < .01). The high shy boys alsohad significantly more acting out problems than themedium shy boys (p < .05) and the medium shy girls (p <.001). As these results imply, within the PBCL thewithdrawn and acting out items are correlated for boys(r = .31, p < .05, N = 57) but not for girls (r = .04, N =65).

Interactions with mother—family drawing task. Here,no significant main effect or interaction effect occurred,either for the relative frequency (Table 2) or for absolutefrequency of interactions with mother (not tabled). Thelowest relative frequency of positive behaviour (andhighest negative behaviour) was shown by the high shygirls.

Maternal Interactions at Home According to Child'sShyness Status

Family drawing task. Unlike the children, mothers'interactions did reveal significant effects (Table 2). Inbetween-groups contrasts medium shy girls receivedsignificantly more positive interactions than either highshy girls or medium shy boys. Indeed medium shy boysreceived the lowest proportion of positive interactionsbut the highest proportion of "instructs" from mothers,and significantly more than the medium shy girls.

Unstructured interactions over an entire visit.Maternal style—the sum of six ratings dealing withpositive, "secure base" behaviour—also showed asignificant interaction effect, with mothers being mostpositive to high shy boys, and least positive to high shygirls, who differed significantly from each of the othergroups. In spite of the sex difference indicated in Table2, maternal style did not differ significantly betweenboys and girls over the whole sample (i.e. when themixed shy group of nine children was included).

The "Mixed Shy" Group

Three girls and six boys fell into this group: Two girlshad been "not shy" at home but high shy in thelaboratory, with ratings of 7 and 7.5. The remaining girland all six boys had been "shy" at home but low shy inthe laboratory, with ratings from 1 to 3. Since the onlysubgroup with a sufficient A involved the six boys, theywere compared with the other three groups of boys. Inone-way ANOVAs, the mixed group of boys differed

Page 5: Shy Girls and Boys: A New Look

SHY GIRLS AND BOYS 185

•I

60

I

o

XuCO

§-

X

s;

1 OS

G'

s "

5

•2 '

o

B

oA03*

B

tA

a

PII

3.

OA

*

cn

aA

19

cn

A A A A A

cn

in 'cn ;

cn tscn cn

A A

XX

O^

A A

XX

p

00rn

o00

cn cn cn

(N ON

cn CN

vqcn

cn

vd

oqCN

vqd

o o00 CN

CN 0000 ^

CN 0000 '-I cn

CT\ O CSVO T—I CN

00 O tNTl ^^ cn

00 CTv cnvo O CN

*O

wj voI

CNCS OO00 rt

I"

I00 S

-1 I

If0 ;=60 S '

1 I.

I| p . | .

:E 8 t

0, M :s ;

5 6 l S 'II11II5i CL, Z £

8SJ

CA> .

o

u

u.£T3

acdBcdu

o

o

T3

5J

1.

r3 <H c

[V a^ (A w U (Ai!

Page 6: Shy Girls and Boys: A New Look

186 J. STEVENSON-HINDE and A. GLOVER

significantly from the other three groups of boys asfollows (please see Table 2):

Negative mood ratings by home observer weresignificantly higher than for medium shy boys (p <.01) or low shy boys (p < .001);

Problem behaviour in presciiool was similar to thehigh shy boys, including acting out behaviour beingsignificantly higher than that of the medium shy boys (p< .05).

Maternal style at home was negative, and significantlylower than with all other boys: low shy (p = < .01),medium shy (p < .05), and high shy (p < .01).

Discussion

The present results indicate the value of selecting forshyness, using criteria involving both natural andlaboratory contexts, categorizing children (including agroup between the two extremes), and including the sexof the child in analyses. These points are highlighted byour unexpected finding, that some children (three girlsand six boys) were placed into one extreme at home butthe other extreme in the laboratory. If nothing else, theyindicate the value of assessing behaviour in more thanone context—either context alone would have groupedthem incorrectly thereby contaminating the "pure"groups. All of the "mixed" shy boys were extremelyshy by home assessments (maternal temperament ratingsand observer ratings), yet on meeting a stranger in thelaboratory (with mother in the room) they received thelowest shyness ratings (1-3 on a 9-point scale). Incomparison with the other three groups of boys, theirown behaviour never differed significantly from, andindeed resembled, that of the high shy boys, includingthe highest levels of problem behaviour in playgroup(Table 2). Yet their mothers behaved in a significantlyless positive way to them than to the other boys,including the high shy group. All six "mixed shy" boyshad problems stemming from birth. Three were Caesar-ean births, surrounded with anxiety (e.g. previousmiscarriages, slow to breathe), one was a traumaticdelivery (facially presented) with subsequent speechproblems (and a tendency to "bark" at strangers whengreeted by them), and two had early-appearing neuro-logical problems, including suspected cerebral palsywhich is not now apparent. Thus, their shyness in thehome setting appears to be associated with causes foranxiety over their well-being.

Over the entire sample different assessments ofshyness were significantly intercorrelated, ranging fromr = .50 to .86 (Table 1). As might be expected, these arehigher than intercorrelations involving samples withfewer extremes (e.g. Stevenson-Hinde & Shouldice,1995b). Intercorrelations of mood ratings were consider-ably lower, ranging from r = .07 to .27. Such low valuesmay be partly due to the fact that extremes of mood werenot selected for. However, the correlations are lowenough to suggest a further possibility. Whereas shynessis defined in terms of a particular context (i.e. a novel orchallenging social event) as well as behaviour indicatingpotential movement (approach/withdrawal), "mood" isless clearly specified, possibly making comparisonsacross situations less straightforward.

Nevertheless, for each of three independent moodassessments (Table 2), negative mood increased sig-nificantly across the low, medium, and high shy groupsfor both sexes, mirroring the association between anxietyand low affect found in clinical samples (e.g. Anderson,1994). Other significant differences across groupsinvolved increasing worries and fears as well as problembehaviour in playgroup, assessed by the PreschoolBehaviour Checklist (PBCL: McGuire & Richman,1986). Here, the mean total score for the high shy boyswas 8.2, with 1/3 of these boys above the cut-off (ofal2) indicating problems. Looking at subsets of thePBCL, high shy children of both sexes had withdrawalproblems in preschool. Surprisingly, the highest in-cidence of acting out behaviour occurred not with thelow shy, but with the high shy boys. Yet in spite of thenegative mood, worries and fears, and problem beha-viour of these high shy boys, they did not behavenegatively with mothers in the drawing task and motherswere positive with them. Maternal "style" (the sum ofsix rating scales) was most positive with high shy boysand was significantly higher than with the high shy girls.Perhaps shy sons can become "mothers' darlings", in away that shy daughters cannot.

As reviewed in the Introduction, results with norma-tive samples indicate that shyness in girls is moreacceptable than shyness in boys. Here, it was the mediumshy group of girls and boys who supported thosefindings. That is, mothers gave a significantly higherproportion of positive interactions to medium shy girlsthan to medium shy boys. Additionally, mothers weresignificantly more positive to medium shy girls than tohigh shy girls, who received a negative standardizedscore for maternal style (z = - .57). Parallels withnormative samples occur if one considers that most ofthe high scorers in an "unselected" sample wouldprobably fall into a "medium shy" group as defined inthe present sample. Thus, previous studies which foundpositive associations with shyness in girls should not betaken to imply that this holds for extreme shyness.Indeed, high shy children of both sexes may be at risk forlater disorder.

In conclusion, had shyness been treated as acontinuum, three sets of findings would have beenobscured: (1) with girls, the nonlinearity of the maternalinteractions, with medium shy girls receiving the mostpositive interactions; (2) the existence of the "mixedshy" children, who were high shy in one context but lowshy in another, and their differences from the "pure"groups; and (3) significant differences not just betweenthe high and low shy groups, but also between the highand medium shy groups. Whether shyness is "truly"categorical is not the issue here, for shyness is a conceptof our own making. The point is that a categoricalapproach reveals differences and suggests future direc-tions that might be otherwise hidden (see also Kagan,1994).

Acknowledgements—We are grateful to the following: AnneShouldice for running the laboratory procedures and super-vising their coding; Natalie Direkze for laboratory coding; andJoanne Early and Alison Tamplin for reliability coding ofhome observations and home ratings. Robert Hinde and Anne

Page 7: Shy Girls and Boys: A New Look

SHY GIRLS AND BOYS 187

Shouldice made helpful comments on the manuscript as well.This work was supported by the Medical Research Council,London.

ReferencesAchermann, J., Dinneen, E. & Stevenson-Hinde, J. (1991).

Clearing up at 2.5 years. British Journal of DevelopmentalPsychology, 9, 365-376.

Ainsworth, M. D. S., Blehar, M. C , Waters, E. & Wall, S.(1978). Patterns of attachment. Hillsdale, NJ: Erlbaum.

Anderson, J. C. (1994). Epidemiological issues. In T. H.Ollendick, N. J. King & W. Yule (Eds), Internationalhandbook of phobic and anxiety disorders in children andadolescents (pp. 43-65). New York: Plenum Press.

Buss, A. H. & Plomin, R. (1984). Temperament: Earlydeveloping personality traits. Hillsdale, NJ: Erlbaum.

Classification of Occupations (1970). London: Her Majesty'sStationery Office.

Ialongo, N., Edelsohn, G., Werthamer-Larsson, L., Crockett,L. & Kellam, S. (1995). The significance of self-reportedanxious symptoms in first grade children: prediction toanxious symptoms and adaptive functioning in fifth grade.Journal of Child Psychology and Psychiatry, 36, 427—437.

Kagan, J. (1994). Galen's prophecy: temperament in humannature. New York: Basic Books.

Kagan, J., Snidman, N. & Arcus, D. (1993). On thetemperamental categories of inhibited and uninhibitedchildren. In K. H. Rubin & J. B. Asendorpf (Eds), Socialwithdrawal, inhibition, and shyness in childhood (pp. 19-28). Hillsdale, NJ: Erlbaum.

McDevitt, S. C. & Carey, W. B. (1978). The measurement oftemperament in 3-7-year-old children. Journal of ChitdPsychology and Psychiatry, 19, 245-253.

McGuire, J. & Richman, N. (1986). Screening behaviourproblems in nurseries; the reliability and validity of thePreschool Behaviour Checklist. Journal of Child Psy-chology and Psychiatry, 27, 7-32.

Martin, R. P. (1988). The temperament assessment battery forchildren. Brandon, VT: Clinical Psychology Publishing.

Mills, R. S. L. & Rubin, K. H. (1993). Socialization factors inthe development of social withdrawal. In K. H. Rubin & J.B. Asendorpf (Eds), Social withdrawal, inhibition, andshyness in childhood (pp. 117-148). Hillsdale, NJ: Erlbaum.

Rubin, K. H. (1993). The Waterloo Longitudinal Project:correlates and consequences of social withdrawal fromchildhood to adolescence. In K. H. Rubin & J. B. Asendorpf(Eds), Social withdrawal, inhibition, and shyness in child-hood (pp. 291-314). Hillsdale, NJ: Erlbaum.

Simpson, A. E. & Stevenson-Hinde, J. (1985). Temperamentalcharacteristics of three- to four-year-old boys and girls andchild-family interactions. Journal of Child Psychology andPsychiatry, 26, 43-53.

Stevenson-Hinde, J. & Hinde, R. A. (1986). Changes inassociations between characteristics and interactions. In R.Plomin & J. Dunn (Eds), The study of temperament:changes, continuities and challenges (pp. 115-129). Hills-dale, NJ: Erlbaum.

Stevenson-Hinde, J. & Shouldice, A. (1995a). Maternalinteractions and self-reports related to attachment classifi-cations at 4.5 years. Child Development, 66, 583-596.

Stevenson-Hinde, J. & Shouldice, A. (1995b). 4.5 to 7 years:fearful behaviour, fears and worries. Journal of ChildPsychology and Psychiatry, 36, 1027-1038.

Accepted manuscript received 24 May 1995

Page 8: Shy Girls and Boys: A New Look