child development 1975 gottman

Upload: sutherlandra

Post on 07-Apr-2018

226 views

Category:

Documents


0 download

TRANSCRIPT

  • 8/3/2019 Child Development 1975 Gottman

    1/11

    Social Interaction, Social Competence, andFriendship in ChildrenJohn Gottman, Jonni Gonso, and Brian RasmussenIndi(ma UniversUy

    , JOHN; GONSO, JONNI; and RASMUSSEN, BRIAN. Soctal Interaction, Social Comptence, and Friendship in ChUdmn. CHILD DEVELOPMENT, 1975, 46, 709-718. The relatioship betwe^i social skills, social interaction, and popularity was examined. The subjects we198 children in third and fourth grades in middle- and low-income schools. The relationshibetween number of friends, socioeconomic status, and grade level were studied in a 2 X 2 Xfactorial design with 2 sets of dependent measures: (1) social skills were ass^sed by experimenter testing each child individually on a set of tasks whidi included measures of tability to label emotions in facial expressions, knowledge of how to make friends, givihelp, and role-taking ability; and (2) social interaction in the classroom was assessed usingnaturalistic observational system. Popular and unpopular children differed in their knowedge of how to make friends and on the referential-communication task. In the classroopopular children distributed and received more positive reinforcement than unpopular dhdren and spent less time daydreaming. Both grade and social class factors were significaHowever, different pattems of results contributed to the main effect of friends and tgrade-level main effect. The importance of assessing social skills which are first validated reference to a criterion such as sociometric position was noted.

    There is a growing body of research onthe development of social-cognitive skills inchildren. Izard (1971) studied childien'sability to recog nize and labe l em otions. _An u m b er oj^ researchers using a wide variety oftasks ha ve e xamined ch ildren's ability to cpm-n iun ica te accu ra le ly t g . . a _ l ^ ^ ( A sh e r &Farke 1975; Flavell , Botkin, Fiy, Wright, &Jarvis 1968; Glucksbe rg & Krauss 19 67 ).Othe r lnvest iga tgrs_ha ye assessed chi ldren 'sability to take the role perspective of anotherperson on perceptual (Pif fgetS Inhelder 1956)andjscjcial tasks (Feffer & Go urevitc h 1 96 0) .One issue fevestigated in the present, |the covar iat ion among these2j3i l |e isntPrevidu s Tesea reft'lias typ ically exam ined ag etrends on one type of task without assessingthe relationships among performances on dif-ferent tasks.

    Another issue neglected in previous re-search is whether differences in social skillamong children are predictive of social suc-cess with peers. This is an important issue since

    peer popularity is itself related to indices later mental health. Unpopular children amore likely to be disproportionately reprsented later in life in a community-wide pschiatr ic register (Cowen, Pederson, BabigaIzzo, & Trost 19 73 ); the y a re also more liketo receive bad-conduct discharges from tarmed forces (Roff 1961). Kohn and Claus(1955) reported that the propSfHon of socisolates in adult manic depressives and schizphrenics was close to one-third, while in normcontrol groups the proportion was close to zeManic depressives were as likely as schizphrenics to have been isolates. In a survey research on suicide and attempted suicidStengel (1971) concluded that "social isot io n is th e co m m on d en o m in a to r o f " ^ n u m bof ifactors correlated with a high suicide rat(p . 2 8 ) . Roff, Sells, and Golden (197 2) cently studied a sample of 40,000 children 21 cities. Except for the lowest socioeconomclass, the relationship was highly positive btween percentage del inquent and low peer-aceptance scores taken 4 years earlier . I t

    The authors wish to acknowledge the cooperation of the principals and staff of Fairvieand Rogers schools in Bloomington, Indiana and the help of Greg Brown, Robert HatchJudy Jordan, Ellen Nakhnikian, Deborah Pizer, and Bart>ara Walker. Requests for repri

  • 8/3/2019 Child Development 1975 Gottman

    2/11

    710 Child Developm ^ittherefore important to study which socialskills relate to peer friendship ratings. Uaotti'fication of critical sIdlOb could be a &st step minterventions aimedt at faciKtating peer rela-' tions in children.

    An early study which related social be-havior to peer popularity was conducted byHirtup, Glazer, and Charlesworth (1967).TTiey observed ntirsery school children in theclassatx^m and categorized social behavior aspositiw or negative. The category of positivebehavior included giving attention and ap-proval, giving affection and personal accep-tance, submitting to another s wishes, andgiving things to another. Negative behaviorsinclua,ed noncompliance, interference, deroga-tion, and attack. Using a sociometric instru-ment in two nursery school classes, theseinvestigators found that positive behaviors dis-pensed to peers were related to acceptancescores in both classrooms. Negative behaviorsdispensed were related to rejection scores inone classroom bu t no t in th e other. J t is jnote-^o^*ty that a number of research studiesintervening with social isolates have usedtotal frequency of interaction as the majorcriterion in defining the isolate and as the de-pendent variable for assessing interventions(O'Connor 1969, 1972; Walker and Hops1973), with no reference to sociometric mea-sures. Furthermore, these studies have taughtsocial skills which were not tested empiricallyby studying the covariation of skills with peerratings. To test the validity of the dependent \measures used in these studies it would benecessary to study social interaction in con-junction with social skills and sociometric po-sition.

    In a more recent study Rubin (1973 )examined the relationship between social skillsand peer popularity. Kindergarten and second-, ifourth-, an(l sixth-grade children were testedon a variety of communication tasks. Popu-larity was assessed by asking each child tospecify his first, second, and third choices ofa classmate he "would like to spend his freeplaytime with." Rubin foimd two factors ina varimax solution of nine of his 12 variables.The first factor, which accounted for 56.9%of the total variance in the nine variables, "wasidentified as a *decentration factor"* (p. 108).

    loaded highly m the second factor but kadedless tha n .30 on the first factor. Therefo rRubin concluded that decentration skills forma unidimensional factor which is independenof popularity.Rubin's conclusion about the independence of popularity from (fecentration is nowarranted from his data. A varimax solutiondoes not guarantee that factors will be uncorrelated. Therefore it may well be that popularity is in fact correlated with decentrationA principal-components solution will guaranteeorthogonality of components. We reanalyzedRubin's data by the method of principal components, using the correlation matrix of all 12of Rubin's variables. Our first component wasimilar to Rubin's decentration factor, and onthis factor a significant loading of popularitywas found. Furthermore, another report byRubin (1972) based on these data found thathe partial correlation between performance onthe Clucksberg and Krauss (1967) referentialcommunication task and popularity, holdingIQ constant, was .58 (p < .01) for both kindergarten and second grade. These results areconsistent with our principal-components reanalysis of Rubin's data. Thus, there is reason

    to believe that certain interpersonal skills maybe related to how well a child is liked by peersThe present investigation examined thecovariation of (1) social skills assessed via aseries of tasks and (2) social behaviors assessed in the classroom, with friendshipchoices assessed by a sociometric measureThird- and fourth-grade children were studiedsince this age level appears to be one of transi-tion in the development of social abilities

    (Asher & Parke 1975 ; Elkind 1961; Feffer 1959Feffer & Gourevitch 1960; Flavell et al. 1968;Glucksberg & Krauss 196 7). Children camefrom two schools differing greatly in theeconomic and educational level of parentsThis provided an assessment of possible sociaclass correlates of the relationships betweenskills and peer popularity.MethodSUBJECTSTwo schools were selected for this study

  • 8/3/2019 Child Development 1975 Gottman

    3/11

    classrooms in each school participated in thet d y ^ there were 198 Ss.PROCEDURESocUjmetric Measure

    Children were asked to list their bestfriends (any number). This item was used be-cause of findings of a pilot study in which asociometric measure had been administered toa third-grade classroom (N = 25) in a largemidwestern city. A number of possible phras-ings were piloted. Children were asked to list(1) their best friends (any number), (2)three children they would like to work with,(3) three children they would go to for help,(4) three children who "really listen to you,"(5) three children who "really like you," and(6) three people "you'd like to play with best."Variables 1-6 were the number of other chil-dren who chose the subject on items 1-6. Aprincipal-components analysis found threecomponents with eigenvalues greater than one.The first component accounted for 54.08% ofthe variance and related primarily to friend-ship (high-loading variables were 1, 4, 5, and6 ) ; the second component accounted for25.98% of the variance and was interpretedas related to perceived academic ability (high-loading variables were 2 and 3). This socio-metric item was administered in the presentexperiment, and variable 1 (best friends) wasused as the sociometric measure of friendship.Social Skill Assessm ent1. Labeling emotions in facial expres-sions.tJsmg a set of photographs from Izard(19'7l), children were assessed in a match-to-sample task in which each child was asked tomatch each picture with index cards whichcontained the words "sad," "scared," "sur-prised," "disgusted," and "happy." A total-correct score was computed for each child.

    2. Referential-cofnmunication accuracy.Children played a word game in which theyhad to communicate to a listener one of twowords in a word pair (e.g., "house-car") bysending the listener an appropriate clue word(e.g., "home"). The italicized word in the

    Gottman, Ck^o, and Rasmussoai 71pair is called tiie "referent,** and the other woiis called the "nonreferent** llie referent appeared with a nonreferent that was eithedissimikr (e.g., "hou^-cai") or similar (e.g."hoiise-home"). Word paiK and instructionwere taken from Asher and Parke (1975), buthe scoring procedures differed. An aduljudge blind to hypotheses about developmentadifferences scared the children's clue words onrelated and unrelated word-pair Usts, usingthree categories: good, bad, and not sure. Thetotal of "goods" on rekted and unrelated wordlists, respectively, became the child's twoscores on this task.

    3. ^er^ective-taking tasks,{a} Leftright 'percepiual decentering. To asseiss tM sskill two objects were placed between thesubject and the experimenter. The subjecwas asked which one was on his left, menwhich one was on his right. He was thenasked which one was on the experimenter'sright and which was on the experimenter'sleft. This was repeated with different objectsfor 12 pairs of objects (Elkind 1961). A total-correct score was calculated for each child.b) Two-m ountairL problem. The E safacing the cHld and set up two stacks of

    blocks, following a preset arrangement, toresemble mountain ranges. The child wasasked to pick from a set the appropriate indexcard with the three-dimensional perspectivedrawing which correctly displayed the moun-tain range" from the experimenter's perspective. Four cards were designed, each of thethree incorrect choices refiecting a differenttransformation error following Gibson (1966):a perspectival transformation error; a rightleft reversal error; and a picture of the blocksas the child actually saw them. The altemativechosen by the child was recorded. A child'sscore was either zero or one, depending onwhether his choice was correct.

    4. JBUndfolded-listener tasks. (a) Obstacle course. The child was asked to pretendthat he had to instruct a blindfolded puppet^through a miniature display of an obstaclecourse so that the puppet did not bump intoanything. The percentage of appropriate com1 Because of the illness of one observer, observational data were not completed for onemiddle-income fourth-grade dassroom, and the subjects for this classroom were dropped fromthe analysis.

  • 8/3/2019 Child Development 1975 Gottman

    4/11

    712 Child Developmentmunication "units," that is, nonvisual cues,was assessed (see Flavell et al. 1968, p. 88).

    b) Me n-with-hats sequence. Five plastictoy men were shown to S in a fixed sequence.The child was asked to give instructions to animaginary blindfolded child about the propersequence of the five plastic toy men. The menwere of different colors and wore hats of dif-ferent colors an d different textures (i.e., blackleather, blue corduroy, gold velvet, purplenetting, black and white fur). The child re-ceived a score ranging from zero to five basedon the number of texture cues.

    5. Making friends.The child was askedto pretend that the E was a new childin school with whom he wanted to makefriends . _]Responses w ere scored as follows:greeting, one point; asking for informatiori,two points; extending inclusion, three points;and giving information, four points. Thesecategories were used since a pilot study foundthat they covered all behaviors exhibited bythird- and fourth-grade children in a making-friends role play. Points were given differen-tially since the pilot study indicated thatchildren tended to proceed in a fixed sequence.Giving information was less likely than extend-ing inclusion, for example. Children in thepilot study followed the same sequence in therole play, though some children completedthe sequence while others did not, and thesequence was the same as that listed above.An example of information sought is "Wheredo you live?" or "How much is two and two?"Inclusion was usually a general offer ratherthan a specific offer to do something together(e.g., "Wanna come over to my house some-timeF*). Information given might be "I live nearthere" or "My favorite sport is basketball."6. Qivinghslp- Subjects were asked topretend that me E was another student in theirclass who needed help on three arithmeticproblem sets. Each problem set required thechild to switch teaching strategy. For the firstset counting on the fingers would do, but forthe second set it would not. For the third set,remainders had to be dealt with. The extentto which the child switched strategy was as-sessed. The child's response was scored zeroif he simply gave the answer, one point if hesupplied a strategy but not very clearly, andtwo points if a clear strategy was supplied; and

    strategy, since a shift in strategy was consid-ered to indicate a high leved'Of he%>ing skifl.Clmsroom Behavior AssessmentObservers were assigned to classrooms.They spent a few days leaming the names ofthe children in the classroom and then beganobserving by picking a child, observing himfor 10 6-second intervals (using a clipboardwith a 6-second light-emitting diode), and thenmoving to the next child on the list. Observa-tion was sampled from four situations: (1)lecture or demonstration situations (approxi-mately 90 observations per child); (2) seatwork (approximately 90 observations perchild); (3) small-group work, mixing, orclassroom work which involved free access toother children (approximately 90 observationsper child); and (4) gym, recess, or playperiods outside the classroom which involvedmore vigorous physical interaction (approxi-mately 40 observations per child). The follow-ing categories were used:^ {a ) alone positive,(not interacting with other children or teacherbut listening to teacher when teacher is talk-ing, or doing assigned work); {b ) alone andoff task (i.e., daydreaming or "tuned out";(c) dispensing positive reinforcer verbally(e.g., giving approval or verbally complyingwith a request); {d ) dispensing positive rein-forcer nonverbally (e.g., giving something [atoken], giving affection, nonverbally comply-ing with a request); (e) dispensing negativereinforcer verbally; (/) dispensing negativereinforcer nonverbally; (g) receiving positivereinforcer verbally; {h ) receiving negative re-inforcer verbally; (i) receiving positive rein-forcer nonverbally; (/) receiving negativereinforcer nonverbally; {k ) entry behavior(asking for something, offering something);{I ) peer interaction neutral (includes whis-perin g); (m) teacher interaction: (1) teachergiving positive reinforcer; (2 ) teache r giv-ing negative reinforcer; (3) teacher neutral;and (n) child initiating interaction with teacher.The relative frequency of each category wastallied for each child, summing over all foursituations.

    Observers were first trained in a pilotschool until an acceptable interobserver agree-ment of 85% was reached. Interobserveragreement was calculated as the total numberof intervals for which agreement occurred

  • 8/3/2019 Child Development 1975 Gottman

    5/11

    85% was maintained throughout the study.Observers began observing in the Title Ischool and moved to the other school afterapproximately 350 observations (35 minutes)per subject, crossing grade level when switch-ing schools. One observer was used through-out the study for weekly random spot reli-ability checks only, and another observer alsofunctioned exclusively as a reliability checker,making reliability checks once a week at ran-dom unannounced times for each observer.Results

    To assess the relationships among socialskill variables, classroom interaction variables,and friendship, two 2 X 2 X 2 factorial mul-tivariate analyses of variance were performed,with two levels of social class, two levels ofgrade (third and fourth), and a median spliton the number of friendship choices received.The low-friends group received zero to fivechoices, with a mean of 2.82 friends, andthe high-friends group received six or morechoices, with a mean of 7.48 friends. Becausesome cells of the design had no variance onsome dependent variables, variable scores weresummed as follows. In the social skills tasksscores on the left-right and two-mountainproblems were summed and called "perceptualdecentering"; scores on the obstacle-courseand men-with-hats tasks were summed andcalled "blindfolded listener." In the classroomobservations entry and neutral interactionscores were summed and called "neutral in-teraction"; positive verbal and positive non-verbal reinforcements distributed were summedand called "distributes positive." Similarly,verbal and nonverbal behaviors were summedfor "distributes negative," "receives positive,"and "receives negative." Teacher-positive,teacher-negative, and teacher-neutral interac-tions were summed and called "receives atten-

    Gottman, Gon so, and Rasmussen 71tion from teacher.'* To control for differencein activity level among children, the proportion of positive to total reinforcement distributed was calculated.SOCIAL SKILL VARIABLES

    Using a multivariate analysis of variancof social skills with WiUcs's X crite rion , a significant main effect was obtained for friendsF(7,185) = 2.41, p < .022. No significaninteraction effects were obtained for gradlevel, social class, and the friends factor. Tabl1 summarizes the univariate F tests for eacsocial skill variable. High-friends children received significantly higher scores on the unrelated-word-pairs task and the knowledgeromaking-friends role play than did low-friendchildren.

    There was a significant grade-level maieffect, F (7,18 5) = 2.96, p < .006. Howevethe univariate F tests revealed a pattem diferent from that for the friends main effecGrade-level differences were found, with fourgraders performing better than third gradeon identifying affect in facial expressions, othe related-word-pair task, on perceptual dcentering, on the blindfolded-listener tasand on knowledge of giving help.

    As expected, there was a significant sociclass effect, F(7 ,1 85 ) = 12.74, p < .00Children in the middle-income school peformed better on all tasks than those in thlower-income school. There was also a signicant social class X grade level interactioF(7,185) = 2.70, p < .011. There were twsignificant univariate social class X grade levinteractions. Lower-income children increasand middle-incon ie rcHil& en idecreiase^ knowledge of making friends from third fourth grade. Table 2 presents means aT A B L E 1

    UNIVARIATE F T E S T S F O R S O C I A L S K I L L V A R I A B L E S F O R E F F E C T S S I G N I F I C A N T U S I N G W I L K S ' S I C R I T E R I O N

    F R I E N D SS K I L L F p

    Facia l expressions 0.98Relate d wo rd pair 0.22Un related wo rd pair 10.20 .002

    G R AD E ( G ) SES (S) S X GF

    2.584.360.99

    P.038

    F19.7815.1613.60

    P.001.001.001

    F4.931.292.46

    .0

    .1

  • 8/3/2019 Child Development 1975 Gottman

    6/11

    714 ChildTABLE 2

    MEAKS AND STANDARD DEVIATIONS FOR SOCIAI, SKUX VARIABLES

    LOWER-INCOME SCHOOL MIDDUE-INCOME SCHOOL

    SOCIAL SKILLFacial expressions:1SDRelated word pair:J PSDUnrelated word pair:

    SD ...............Perceptual decentering:jfSDBlindfolded listener:1SDFriend making:ISDHelp giving:SD..... . . . . . . . .V.

    ThirdLowFriends14.69(2.09)

    1.81(1.09)8.09(1.40)6.97(2.66)

    83.28(66.54)3.16(1.53)2.44(3.51)

    GradeHighFriends15.32(2.04)2.48(1.24)9.00(1.16)8.55(2.39)

    84.19(55.87)3.77(2.14)1.94(2.61)

    FourthLow

    Friends14.78(2.92)2.38(1.67)8.19(1.20)8.43(2.90)

    99.73(61.14)4.00(1.76)2.89(3.90)

    GradeHighFriends14.92(1.62)2.17(1.40)9.08(0.90)8.92(1.62)

    117.08(72.20)4.67(2.96)2.67(3.34)

    ThirdLowFriends15.86(1.70)2.90(1.56)9.26(0.64)9.40(2.50)

    126.40(57.23)4.50(2.33)3.68(3.26)

    GradeHighFriends15.74(2.00)2.46(0.80)9.05(0.84)9.11(2.99)

    139.55(54.13)5.86(3.20)4.55(4.27)

    FourthIx)WFriends17.25(1.92)3.50(1.88)8.55(1.10)

    10.35(2.43)169.05(45.55)

    4.15(0.67)5.05(0.69)

    GradeHighFriends17.00(1.16)3.71(2.56)9.29(0.76)8.86(3.13)

    185.71(37.80)4.14(0.90)5.14(0.38)

    standard deviations for the social skill vari-ables.CLASSROOM BEHAVIOR VARIABLES

    Using multivariate analysis of variance ofthe classroom interaction variables, a margin-ally significant main effect was obtained forfriends, F(10,182) = 1.79, p < .066. Table3 presents the univariate F ratios for each

    classroom interaction variable. High-friendschildren were oflF task (daydreaming insteadof attending or working) less and distributedand received more positive reinforcement thanlow-friends children. There was also a signif-icant friends X grade level interaction for theclassroom variables, F( 10,182) = 2.26, p

  • 8/3/2019 Child Development 1975 Gottman

    7/11

    ment, the gap decreased for off-task behaviorand widened for positive reinforcement re-ceived from tihird to fourtfi grade.There was a significant grade-level mainF ( 10,182) 8.73, p < .001, for thesocial interaction variables. Fourth graderswere on task more, distributed more negativereinforcement, and received morepositive re-inforcement than third graders. There was asignificant social class main effect, F( 10,182)= 5.71, p < .001. Middle-income childrenwere on task more, distributed more positivereinforcement, distributed and received lessnegative reinforcement, had a higher propor-tion of positive to total interaction, and inter-acted with the teacher (initiated and received)less than lower-income children.

    Gottman, Ckinso, and R a s m u s ^ 715There was a si^tiificant social class Xgrade level interactitm, F( 10,182) ~ 27.13, p< .001;fromhird to fourth grade, low-incomechildren increased the time they spent bothon and off task and interacdng with theteacher and distributing negative reinforce-

    ment, while middle-intx)me cmldren decreasedon tiiese variables from third to fourth grade.Middle-income children increased from thirdto fourth grade on neutral interaction and ondistributing and receiving positive reinforce-ment, whereas low-income children decreasedon these variables. Table 4 presents means andstandard deviations for the classroom behaviorvariables.To study differences in the relationshipbetween verbal and nonverbal reinforcement

    TABLE 4M E A N S AND STANDARD DEVIATIONS FOR CLASSROOM BEHAVIOR VARIABLES

    VARIABLEAlone on task:XSDAlone off task:XSDNeutral interaction:J TSDDistributes positive:

    SDDistributes negative:XSDReceives positive:

    SDReceives negative:XSDInitiates to teacher:XSDReceives attention fromteacher:rSD

    LOWER-INCOME SCHOOLThirdLowFriends

    48.11(21.99)7.83(6.28)

    28 07(12.58)9.87(8.46)0.08(0.17)2 09(2.10)0.46(1.17)0 97(0.59)

    1.76(1.21)

    GradeHighFriends46.41(20.62)4 06(3.96)

    32.21(13.09)10.10(8.11)0.13(0.24)3.00(3.18)0.24(0.31)1 20(0.94)

    1.90(1.52)

    FourthLowFriends74.83(14.41)6.84(6.56)

    10.22(13.64)1.99(1.93)0.69(0.73)1.46(1.25)0.44(0.44)108(0.92)

    1.78(1.47)

    GradeHighFriends70.75(14.84)8.64(7.76)

    11.14(8.75)1.96(0.93)0.77(0.58)1.68(0.96)0.43(0.35)1.03(1.65)

    2.69(2.53)

    MI0DLE.-INCOME SCHOOLThirdLowFriends73.40(14.13)

    7.58(5.97)13.19(9.69)1.31(1.41)0.22(0.27)1.07(1.03)0.16(0.25)0.69(0.70)

    1.58(1.93)

    GradeHighFriends78.48(14.00)

    3.92(5.27)12.08(9.34)1.02(0.80)0.27(0.24)1.10(0.98)0.10(0.19)0.74(0.71)

    1.46(1.33)

    FourthLowFriends

    49.32(9.10)4.09(4.02)

    34.55(9.03)4.59(2.96)0.03(0.02)4.46(2.80)0.10(0.45)1.05(0.99)

    0.86(0.99)

    GradeHighFriends38.00(7.99)

    3.49(2.21)32.71(9.39)17.40(34.82)0.02(0.01)5.76(4.63)0.01(0.01)0.67(0.74)

    0.93(0.78)

  • 8/3/2019 Child Development 1975 Gottman

    8/11

    716 Child Developm entdispensed and friendship as a functimi of socialclass, regression equations were computedseparately for each school, with friends as thedependent variable. Table 5 shows that verbalreinforcement accounts for the most variancein the relationship of reinforcement dispensedto friendship in the middle-income school andnonverbal reinforcement accounts for the mostvariance in this relationship in the lower-income school.IMsciissicm

    The findings support the conclusion thatpopular children are more skillful than un-popular children and interact differently withtheir peers. The classroom interaction data areconsistent witK H ar tu p et al 's (1967 ) findingof correlation between acceptance scores apddistribution of positive reinforcement in tv/pmiddle-income nursery school classes. Popularchildren were also'^ore knowledgeable abouthow to make friends. The strongest findingwas the relationship Between performance onthe unrelated-word-pairs task and the friend-ship main effect. I'his finding is consistentwim Rubin's (1972) report of high positivepartial correlations (with IQ held constant)between popularity and performance on theClucksberg and Krauss (1967) referential-communication task. Asher and Parke (1975)reported that second graders made very fewenors in referential communication on unre-lated word pairs. However, a strong mono-tonic developmental effect was observed fromsecond to sixth grade on the related-word-pairslist. This finding was replicated in the presentinvestigation. Asher and Parke argued thatchanges in performance on the related-word-pairs list may be indicative of "comparisonprocesses." In comparison processes the childmust produce a good clue word by editingfrom a total list of possible associates to thereferent that clue word which is also less an

    associate to the nonreferent. Themmablx.flltts by taldng^ihe i^ F sp ^t iv e of ^Igteuer* Differential effects on the related- andunrelated-woHi-pairs lists in the present inves-tigation suggest that comparison processes maynot be operative in having or making friends.Instead, a more primitive skill may be in-volved, in which the high-fdends child ismore able than the low-friends child to com-municate to a listener what he is talking aboutin situations where the fine discriminationtapped by the related-word-pairs task is un-necessary. The unrelated-word-pair skill maybe more related to Vygotsky's (1962) investi-gations of private or unedited inner speech asegocentric thought. Part of the function servedby the familiar classroom activity of "showand tell" may be to train the child to bringhis audience along with him in showing ortelling about something, a skill which may betapped by the unrelated-word-pair task.

    The results also suggest that there arerelationships among measures of friendship,social skills, and social interaction. However,the relationships obtained do not support thehypothesis that social skill variables whichdiffer across age will also necessarily be re-lated to a criterion variable of successful socialinteraction such as friendship. For example,the grade-level univariate F ratios for therelated-word-pair, blindfolded-listener, andperceptual-decentering tasks were large. How-ever, these effects were not significant at the.05 level for the friends main effect. There istherefore no reason to suspect that "skills"logically related to making friends (such asblindfolded-listener role-taking skills) are infact related, without prior validation. It istherefore improper to label performance on atask a "social skill" simply because older children do better than younger children on thetask. It may be necessary to validate the taskby showing that performance on the task ha

    T A B L E 5M U L T I P L E REG RES S I O N O F N U M B ER O F F RI EN D S O N PO S I TI V E V ERBA L ( D 4- V ) A ND P O S I TI V E N O N V ERBA L

    ( D - | - N ) REI N F O RCE M EN T D I S P EN S ED TO P EERS AS A F U N CTI O N o r S O CIA L CLA S S A ND GRA D E LEV EL

    D-f-VProportionof Variance Direct ionof

    D + NProportionof Variance DirectionofWeights

    COVARIANCE(D -F V, D + NProportionof VarianceAccounted For

  • 8/3/2019 Child Development 1975 Gottman

    9/11

    discriminant validity with reference to a cri-terion such as sociometric status, which itselfhas been shown to have predictive validity.The finding that nonverbal reinforcementdispensed is related to friendship in the lower-

    income school whereas verbal reinforcementdispensed is related to friendship in the mid-dle-income school may be an important resultin designing interventions for socially isolatedchildren. It may be necessary to design dif-ferent interventions for children in differentschools. While it could be argued that theeffective number of subjects for analyzingsociometric position in this study is the num-ber of classrooms, not the number of children,school-wide sociometric data conelate ex-tremely highly with within-classroom socio-metric data, and the consistency is evengreater for the extremes of friendship choicesreceived (Bell & French 1950; Gronlund &Whitney 1956).Significant interactions between socialclass and grade level would seem to be parsi-moniously explained by suggesting that mid-dle-income-school children's orientation shiftsmore toward a peer-interaction mode whereaslow-income-school children's orientation shiftsmore toward task behavior and interaction withthe teacher. For example, from third to fourthgrade the relative frequency of neutral inter-action with peers moves from 30.14% to10.68% in the lower-income school and from12.64% to 33.63% in the middle-incomeschool. It is unclear why such shifts wouldtake place, but it is likely that lower-income-school fourth-grade teachers stress remedia-tion of academic skills whereas middle-income-school fourth-grade teachers relax in this area.On-task behavior increases in the lower-income

    school from 47.26% to 72.79% from third tofourth grade and decreases in the middle-income school from 75.94% to 43.66% fromthird to fourth grade. However, it is interest-ing that the making-friends scores of middle-income children decrease while those of lower-income children increase; also, middle-incomechildren catch up to lower-income children onthe identification of affect in faces. Perhap.ssocial skill development lags the shifts in em-phasis in the two schools.It is insufficient to study only the co-variation among social skills, social interaction,

    Gottman, Q in so , and Rasmi:sen 717designed to train low-friends children on thosevariables which were shown to be related tothe friendship main effect. These interventionstudies should use sociometric status as thedependent variable. In a recent reyfejM,.jofiatevyentions for socially isolated childrenAsher, Oden, and Gottman (in press) foundthat all interventions in which sociometric position was the dependent measure resulted insmall ga^^ins. In those few cases where interventions included a reasonably long term followup, these initial gains returned to baselineThe same conclusion was drawn in an earliereview by Bonney (1971). The present investigation may suggest one possible reason fothis unfortunate return-to-baseline effect. McFall (1975, p. 7) discussed this issue for sociaskill training in the area of assertion: "Investigators have relied on little more than theiclinical intuitions to decide what assertive responses to teach in their training programsThis is unfortunate, since there is an integrarelationship between the validity of the responses selected for trainingthat is, the program's contentand the potential effectivenesof a training program's techniques."

    The present investigation can function asa discriminant validity study for the design ofintervention programs. A coaching study wassubsequently performed, training low-friendchildren on those variables identified in thepresent investigation. There were two experimental and two control children. Long-termfollow-up indicated that the experimental children changed significantly. Interrupted timeseries analyses showed no significant increasein the total frequency of interaction for anychild; however, experimental children redistributed their interaction to their new friends.*This study is currently being repeated with alarger N. Previous intervention studies in thiarea have been focused on contingency management (Walker & Hops 1973) or modeling(O'Connor 1969, 1972). While coaching as atechnique has been used successfully withadults in areas of assertion training (M cFal& Twentyman 1973) and social skill training(Goldsmith & McFall 1975 ), there has beenno systematic study of the effects of coachingchildren in social skill training.ReferencesAsher, S. R.; Oden , S. L.; & Go ttman, J. M. C hil

  • 8/3/2019 Child Development 1975 Gottman

    10/11

    718 Ohfid Devekpm eirtKatz (Ed.) , Current topics in early chUd-hood education. Vol. 1. Hillsdale, N.J.: Erl-baum, in press.Asl^r, S. R., & Parke, R. D. Influence of samplingand comparison processes on the developmentof communication effectivaaess. Journal ofEducational Psychology, 1975, 67, 64.-75.Bell, G. B., & French , R. L. G onsistency of indi-vidual leadership petition in small groups ofvarying m embership. Journal of Abnorma l andSocial Psychology, 1950, 45, 764-767.Bonney, M. E. Assessment of ^orts to aid sociallyisolated elementary school pupils. Journal ofEducational Research, 1971, 64(8), 359-364.Gowen, E. L.; Pederson, A.; Babigan, H.; Izzo,L. D.; & Trost, M. A. Long-term follow-up ofearly ^ e c t e d vulnerable children. Jourrud ofCormtking and Clinical Psychology, 1973,41, 438-446.Elkind, D. Piaget's conceptions of right and left:Piaget replication study IV. Journa l of G eneticPsychology, 1961, 99, 269-276.Feffer, M. H. The cognitive implications of role-taking behavior. Journal of Personality, 1959,27, 152-168.Feffer, M. H ., & Gourevitch, V . Gognitive aspectsof role-taking in children. Journal of Person-ality, 1960, 28, 383-396.Flavell, J. H.; Botkin, R. T.; Fry, G. L.; Wright,J. W .; & Jarvis, P. E. The developmera ofrole-taking and communication skills in chil-dren. New York: Wiley, 196$.Gibson, E. J. Experimental psychology of leamingto read. In J. Money & G. Schiffman (E ds .),The disabled reader: education of the dyslexicchild. Baltimore: Johns Hopkins UniversityPress, 1966.Glucksberg, S., & Krauss, R. M. W hat do peop lesay after they have leamed to talk? Studies ofthe development of referential communication.MerriU-Palmer Quarterly, 1967, 13, 309-342.Goldsmith, J. B., & McFall, R. M. Developmentand evaluation of an interpersonal skill-training program for psychiatric inpatients.Journal of Abnormal Psychology, 1975, 84( 1) ,51-.58.Gronlund, N. E., & W hitney, A. P. Relation be-tween pupils' social acceptability in the class-room, in the school, and in the neighborhood.Social Review, 1956, 64, 267-271.Ha rtup , W. W.; G lazer, J. A.; & Gharlesworth, R.

    Izard, G. E. The face of emaion. New YoAppleton-Gentury-Grofts, 1971.

    Kohn, M., & Glausen, J. Sodal isolation and sdphrenia. American Sodolog^cd Review, 1920, 265-273.McFall, R. M. Assertion training. Unpublmanusaript. University of Wisconsin, Mson, 1975. (To be published in B. B. Wolm[Ed.], International encyclopedia of neuogy, psychiatry, psycho analysis, and psychogy, in preparation.)McF all, R. M., & Twen tyman, G. T. Four expments on the relative contribution of reheal, modeling, and coaching to assertraining. Journal of Abnormal Psycholo1973, 81, 199-219.O'Gonnor, R. D. Modification of social wdrawal through symbolic modelling. Jourof Applied Behavior Analysis, 1969, 2, 15O'Gonnor, R. D. The relative efficacy of modellshaping, and the combined procedures the modification of sodal withdrawal. Jourof Abnormal Psychology, 1972, 79, 327-3Piaget, J., & Inhelde r, B . The child's conceptionspace. London: Routledge & Kegan P1956.Roff, M. Ghildhood social interactions and yoadult bad conduct. Journal of Abnormal aSocial Psychology, 1961, 63, 333-337.Roff, M.; Sells, B.; & Golden, M. M. Social adjment and personality development in childMinneapolis: University of Minnesota Pr1972.Rubin, K. H. Relationship between egocencommunication and popularity among pe

    Developnwntal Psychology, 1972, 7(3), 3Rubin, K. H. Egocentrism in childhood: a unitconstruct? Child Development, 1973, 44, lC110.Stengel, E. Suicide and attempted suicide. Middsex: Penguin, 1971.Vygotsky, L. S. Thought and language. GambriMass.: M.I.T. Press, 1962.Walker, H. M., & Hops, H. Th e use of group aindividual reinforcement contingendes in modification of social withdrawal. In L.

    Hamerlynck, L . G. Handy, & E. J. M(Eds . ) , Behavior change, methodology, c

  • 8/3/2019 Child Development 1975 Gottman

    11/11