A Longitudinal Study of the Roots of Preschoolers' Conscience: Committed Compliance and Emerging...

19
A Longitudinal Study of the Roots of Preschoolers' Conscience: Committed Com^iance and Emerging Internalization Grazyna Kochanska University of Iowa Nazan Aksan University of Wisconsin Amy L. Koenig University of Iowa KOCHANSKA, GRAZYNA; AKSAN, NAZAN; and KOENIG, AMY L, A Longitudinal Study ofthe Roots of Preschoolers' Conscience: Committed Compliance and Emerging Internalization. CHILD DEVEL- OPMENT, 1995,66,1752-1769. The focus of this study is the complex relation between compliance and internalization in childhood. It is a replication and a longitudinal extension of earlier work, where we distinguished between 2 forms of compliance: committed, when tbe child eagerly embraced and endorsed the mother's agenda, and situational, when the child was cooperative, but lacked the sincere commitment and feeling of internal obligation, 99 children, seen previ- ously at 26-41 months, were studied again at 43-56 months. Compliance and internalization were assessed in multiple observational contexts and using maternal reports. As at toddler age, the 2 forms of compliance had distinctly different developmental tr^ectories, and again, only committed compliance was significantly associated with measures of intemalization. Moreover, committed but not situational compliance at toddler age predicted internalization at preschool age. Shared positive affect within the mother-child dyad at toddler age predicted some measures of intemalization at preschool age. Further evidence of significant difFerences in children's com- pliance to maternal "dos" versus "don'ts" is reported. We have recently returned to one ofthe clearly disparate from the true internaliza- central developmental issues, the origins of tion of rules. The latter, in his view, resulted internalization and conscience {Grusec & from cognitive development and the transi- Goodnow, 1994; Kochanska, 1993). We also tion from the world dominated by control- have rediscovered the importance ofthe tod- ling parents to the world of peers (Piaget, dler period as the context for emerging in- 1932). Similarly, Kohlberg (1969) argued ternalization (Emde, Biringen, Clyman, & that child early experienees with parents Oppenheim, 1991; Kagan & Lamb, 1987; had no beneficial impact on internalization. Kochanska, 1994). Consequently, many is- Other models, however, have portrayed in- sues in early moral development are being ternalization as gradually unfolding out of revisited. early compliance to parents (Gralinski & A long-standing controversy concerns f«PP' ^?^^' ^".^^^^'^Ji 1991; Kopp, 1982; the relation between compliance and in- ^^^j"' ^^^^'^ ^^y*^"' "°^^"' ^ Amsworth, ternalization. The cognitive-developmental model sharply contrasted these constructs. We recently advanced a model tiiat rec- Piaget saw child early compliance to parent onciles these competing approaches by pos- as reflecting constraint by authority and iting a differentiated view of compliance This research was supported by grants from the National Science Foundation (DBS- 9209559), the MacArthur Foundation, and the University of Iowa Faculty Scholar Award to Grazyna Kochanska. We are particularly grateful to Juli Fratzke and Kim Vandegeest for their important contributions to coding and data collection, and to Kathy DeVet, Margi Goldman, Kathy Murray, Eric Poole, and numerous graduate and undergraduate students for their help with all aspects of this research. We also appreciate the enthusiastic commitment ofthe motliers and children in the Toddler Study. All correspondence should be addressed to Grazyna Kochan- ska, Department of Psychology, University of Iowa, Iowa City, IA 52242. [Child Decelopment, 1995,66,1752-1769. © 1995 by the Society for Research in Child Development, Inc, All rights reserved, 000&-.'J920/95/6606-0018$01.00]

Transcript of A Longitudinal Study of the Roots of Preschoolers' Conscience: Committed Compliance and Emerging...

A Longitudinal Study of the Roots ofPreschoolers' Conscience: CommittedCom^iance and Emerging Internalization

Grazyna KochanskaUniversity of Iowa

Nazan AksanUniversity of Wisconsin

Amy L. KoenigUniversity of Iowa

KOCHANSKA, GRAZYNA; AKSAN, NAZAN; and KOENIG, AMY L, A Longitudinal Study ofthe Roots ofPreschoolers' Conscience: Committed Compliance and Emerging Internalization. CHILD DEVEL-OPMENT, 1995,66,1752-1769. The focus of this study is the complex relation between complianceand internalization in childhood. It is a replication and a longitudinal extension of earlier work,where we distinguished between 2 forms of compliance: committed, when tbe child eagerlyembraced and endorsed the mother's agenda, and situational, when the child was cooperative,but lacked the sincere commitment and feeling of internal obligation, 99 children, seen previ-ously at 26-41 months, were studied again at 43-56 months. Compliance and internalizationwere assessed in multiple observational contexts and using maternal reports. As at toddler age,the 2 forms of compliance had distinctly different developmental tr^ectories, and again, onlycommitted compliance was significantly associated with measures of intemalization. Moreover,committed but not situational compliance at toddler age predicted internalization at preschoolage. Shared positive affect within the mother-child dyad at toddler age predicted some measuresof intemalization at preschool age. Further evidence of significant difFerences in children's com-pliance to maternal "dos" versus "don'ts" is reported.

We have recently returned to one ofthe clearly disparate from the true internaliza-central developmental issues, the origins of tion of rules. The latter, in his view, resultedinternalization and conscience {Grusec & from cognitive development and the transi-Goodnow, 1994; Kochanska, 1993). We also tion from the world dominated by control-have rediscovered the importance ofthe tod- ling parents to the world of peers (Piaget,dler period as the context for emerging in- 1932). Similarly, Kohlberg (1969) arguedternalization (Emde, Biringen, Clyman, & that child early experienees with parentsOppenheim, 1991; Kagan & Lamb, 1987; had no beneficial impact on internalization.Kochanska, 1994). Consequently, many is- Other models, however, have portrayed in-sues in early moral development are being ternalization as gradually unfolding out ofrevisited. early compliance to parents (Gralinski &

A long-standing controversy concerns f«PP' ^?^^ ' ̂ ".^^^^'^Ji 1991; Kopp, 1982;the relation between compliance and in- ^ ^ ^ j " ' ^^^^'^ ^^y*^"' " ° ^ ^ " ' ^ Amsworth,ternalization. The cognitive-developmentalmodel sharply contrasted these constructs. We recently advanced a model tiiat rec-Piaget saw child early compliance to parent onciles these competing approaches by pos-as reflecting constraint by authority and iting a differentiated view of compliance

This research was supported by grants from the National Science Foundation (DBS-9209559), the MacArthur Foundation, and the University of Iowa Faculty Scholar Award toGrazyna Kochanska. We are particularly grateful to Juli Fratzke and Kim Vandegeest for theirimportant contributions to coding and data collection, and to Kathy DeVet, Margi Goldman,Kathy Murray, Eric Poole, and numerous graduate and undergraduate students for their helpwith all aspects of this research. We also appreciate the enthusiastic commitment ofthe motliersand children in the Toddler Study. All correspondence should be addressed to Grazyna Kochan-ska, Department of Psychology, University of Iowa, Iowa City, IA 52242.

[Child Decelopment, 1995,66,1752-1769. © 1995 by the Society for Research in Child Development, Inc,All rights reserved, 000&-.'J920/95/6606-0018$01.00]

{Kochanska & Aksan, 1995). We argued that,whereas intemalization may indeed be con-trasted from some forms of child complianceto parent, it also may be seen as rooted in,or unfolding out of, some other forms of earlycompliance. Drawing on earlier theoreticalwork (Kuczynski, 1992; Lepper, 1981; Mac-coby & Martin, 1983; Perry & Perry, 1983),we distinguished internally driven, self-regulated, committed compliance, when thechild appears to endorse, embrace, and ac-cept parental agenda as his or her own, andsituational compliance, when the child,though essentially cooperative, neverthelesslacks sincere commitment and requires pa-rental sustained control. Consequently, weargued that internalization emerges devel-opmentally out of committed, but not situa-tional, compliance.

This differentiation was strongly sup-ported in our earlier study of 103 toddlersand their mothers (Kochanska & Aksan,1995), and it has fostered our understandingof early roots of conscience. In that study,committed and situational compliance ap-peared to be quite distinct in their develop-mental trajectories and relations to internal-ization. Committed compliance increased(cross-sectionally, between 26 and 41months) with the age of the child, whereassituational compliance decreased. Commit-ted, but not situational, compliance relatedstrongly to internalization, both observedand mother-rated.

In the original study, we also exploredthe qualities of mother-child dyads that wereconducive to children's committed compli-ance. Several authors have emphasized theimportance of positive affect, cooperation,and good feelings between parent and childfor child receptiveness to parental socializa-tion (Emde et al., 1991; Lay, Waters, & Park,1989; Londerville & Main, 1981; Maccoby,1983,1992; Maccoby & Martin, 1983; Parpal& Maccoby, 1985; Stayton et al., 1971),Thus, in that study, we measured the dyadic,shared positive affect experienced by themother and the toddler during observed in-teractions. As expected, committed compli-ance was strongly linked to shared positiveaffect in the dyads, both contemporaneouslyIn the same interaction and across two obser-vations on different days. Factor analysis us-ing the dyad as a unit produced a SharedAffective Positivity/Child Cooperation fac-tor, which included high shared positivemother-child affect, child high committedcompliance, low passive noncompliance,and low refusal to comply. Children from the

Kochanska, Akgan, and Koenig 1753

dyads high on that factor were also high onindependently assessed internalization.

The current work draws from and ex-tends that original study while addressingthree goals. The first goal was to replicatethe provocative toddler-age findings. Usingthe same subjects, now at preschool age, weagain examined the viability of the qualita-tively different forms of compliance and ex-plored their relations with mother-child mu-tual positive affect and child intemalization.Again, we expected committed and situa-tional compliance to be distinct, and onlythe former to relate to intemalization. As inpast work (Crockenberg & Litman, 1990;Kuczynski & Kochanska, 1990), we retainedthe distinctions among forms of child non-compliance (passive noncompliance andovert protest) at both toddler and preschool

The first study, cross-sectional in de-sign, allowed for some developmental con-clusions, but it could not inform us if com-mitted compliance increases and situationalcompliance decreases in the longitudinalsense, and if early differences in committedcompliance and early mother-child mutualpositive affect are longitudinally stable overtime. We conceptualized committed compli-ance as the child's receptive, enthusiasticstance that is an outgrowth ofthe enduring,positive mother-child relationship, and thnswe expected it to have longitudinal continu-ity. Additionally, because the measures ofcompliance and conscience in the first studywere contemporaneous, we do not know ifearly committed compliance and mother-child shared positive affect predict thechild's future internalization. If, as we argue,toddler-age committed compliance is thefirst step in developmentally unfolding in-ternalization, it should predict internaliza-tion at preschool age. In addition, we exam-ined whether the important toddler-agedyadic quality of mother-child interaction.Shared Affective Positivity/Child Coopera-tion, has a long-term impact and predictschild intemalization at preschool age. To ex-amine changes and continuities over time,as well as the longitudinal predictions, wasthe second goal ofthe present work.

The third goal was to examine the differ-ences in children's compliance to maternal"do" versus "don't" directives, as in the firststudy. We found that it was more challeng-ing for toddlers to comply to "dos" (requeststo sustain a mundane activity) than to"don'ts" (prohibitions to engage in an attrac-

1754 Child Development

tive activity), that committed complianceacross both contexts was correlated, andthat the developmental increase in commit-ted compliance occurred mostly for the"don'ts." Due to the importance of thispoorly understood developmental issue, weretained the focus on differences betweenthe two contexts in most analyses. We pre-dicted that preschoolers would respond lessenthusiastically to matemal "dos" than"don'ts," that there would be coherenceacross the two contexts for committed com-pliance, and that there would be greater lon-gitudinal changes in committed complianceto "don'ts" than to "dos."

In the present work, we introduced con-siderable methodological extensions in con-science assessments. In the original study,although we did assess separately child com-pliance to matemal "dos" and "don'ts," wedid not make a similar distinction in theconscience measures, which focused ex-clusively on internalization of maternal"don'ts" (not touching prohibited toys). Incontrast, now expanded internalization as-sessments included observations of childrenin separate situations when they felt theywere without surveillance, intended to cap-ture, respectively, their intemalization ofmatemal requests and prohibitions. In one,we assessed intemalization of the maternalrequest (calling for the maintenance of amundane and tedious activity, "do"). In theother situation, analogous to the toddler-ageparadigm, we assessed their intemalizationof the maternal prohibition (demands for re-straint, "don't").

Furthermore, in addition to observingchild conformity to the rules of conduct con-veyed by the mother, we assessed confor-mity to the rules communicated by anotheradult. To that effect, children were observedon three occasions, again without surveil-lance, playing games to win attractive prizes.Winning, however, was impossible withoutcheating, for which ample opportunities ex-isted. The degree to which children eon-formed to the rules ofthe game, clearly spec-ified at the outset by the examiner, wasconsidered to be the expression of intemal-ization. Finally, as at Time 1, we obtainedmaternal reports of child internalization offamily rules in everyday life.

This article originates from an ongoinglongitudinal study on children's tempera-ment and parents' socialization in early de-velopment of conscience. For more infonna-tion, see Kochanska (1995); Kochanska &

Aksan (1995); Kochanska, Casey, and Fuku-moto (1995).

MethodSAMPLE

At Time 1,103 normally developing tod-dlers (51 girls. 52 boys) aged 26 to 41 months(M = 32.86, SD = 4.09) and their mothersparticipated. The mothers responded to lo-cal ads. They represented a wide range ofeducation, from high school (12%) to post-graduate education (31%), and family in-come, from under $15,000 (13%) to over$45,000 (38%). They were mostly Caucasian(7% minorities). More details about the sam-ple are in Kochansfca and Aksan (19^). AtTime 2, 99 children (49 girls, 50 boys) andmothers were seen when tiie children wereaged 43 to 56 months (M = 46.05, SD =2.62). Two families had moved, one couldnot be located, and one refused due to timeconstraints.

OVERVIEW OF PROCEDURES

At Time 1, modiers and children wereobserved during a home and a laboratorysession (2-2V2 hours each). At Time 2, therewas only a laboratory session (3-3V2 hours).All sessions were videotaped. The labora-tory consists of two observation rooms, con-nected by a sliding door: a playroom withsome toys and a play table with chairs, anda naturalistic living room, furnished with acouch, a shelf with foodstufTs, a rockingchair, and other household knickknacks.There is also a low shelf with objectsextremely attractive to young children,adapted to their current age, for example, afishing game, a train set, a see-through loco-motive, touch-activated birds, a typewriter,and a music box, all serving as the "prohib-ited" objects.

The same main experimenter ("an-chor") conducted both sessions at Time 1.Pleasant rapport was established aA the out-set of the home session (mother and childreceived small gifts). At Time 2, motfiers andchildren came to the laboratory for a ses-sion conducted by another experimenter. Apleasant interaction preceded die session(taking a picture of the child and mother,framing and decorating it). At both times, weassessed child compliance and noncompli-ance and mother and child affect in disci-pline interactions in the "do" contexts (toycleanups) and in the "don't" contexts (theprohibition regarding the objects on theshelf), observed the child's internaiization.

and collected mothers' ratings ofthe child'sinternalization in daily life.

PARADIGMS AND MEASURES ASSESSINGCHILDREN'S COMPLIANCE/NONCOMPLIANCEAND MOTHERS' AND CHILDREN'S AFFECT INDISCIPLINE INTERACTIONS

ProceduresThe "do" context.—The details are in

Kochanska and Aksan (1995). The tasks wereanalogous at Time 1 and Time 2 (home andlab toy cleanup, each 15 min, at Time 1, andone toy cleanup, in the lab, 10 min, at Time2). Each time, we supplied multiple toys tostandardize the task, all spilled on the floorto assure that there would be plenty to cleanup. Following free play, when mother wasasked not to clean up the toys, she requestedthat the child put them all back in a largebag (Time 1 home), basket (Time 1 lab), ora big bag with many pockets and a largefishing box with many compartments (Time2 lab). Most children did not complete thetask during the allotted time; for those whodid, the paradigm was considered com-pleted earlier.

The "don't" context.—The tasks weresimilar at Time 1 and Time 2. Upon entryto the laboratory "living room," the motherprohibited the child from touching anythingon the low shelf. Subsequently, several pe-riods of mother-child interaction during thelab session (e.g., meal, free time; 50 min atTime 1 and 45 min at Time 2) were codedfor all exchanges involving the prohibitedobjects on the shelf.

CodingThe "do" context.—At Time 1, each 60-

sec segment, and at Time 2, each 30-sec seg-ment was coded for quality of child compli-ance/noncompliance and mother and childaffect (we shortened the segments at Time2, as the cleanup was shorter).

The "don't" context.—The coding, thesame at Time 1 and 2, was event-triggered.An episode began when the child looked at,approached, touched, or talked about theprohibited objects, or mother made a com-ment about them. Once an episode began,quality of child compliance/noncompliance

Kochanska, Aksan, and Koenig 1755

was coded for each 20-sec segment until thechild reoriented to another activity. Motherand child affect was coded for each wholeepisode.

The coding of children's compliance/noncompliance.—At Time 1 and at Time 2,six mutually exclusive codes described thedegree of acceptance of maternal agenda(Kochanska & Aksan, 1995). Defiance in"do" and "don't" contexts at Time 2 was lessthan 1%, and it correlated with refusal/nego-tiation (r = .45, p < .001). Thus, we ag-gregated both codes into "overt protest." Toobtain a corresponding variable for longitu-dinal analyses, we created an analogous ag-gregate for Time 1, when these behaviorsalso correlated (r = .47, p < .001).^

1. Committed compliance: Full en-dorsement of maternal agenda. "Do" con-text: Child worked willingly, setting his owngoals (e.g., moving spontaneously from onepile of toys to the next), having clearly ac-cepted and endorsed the maternal agenda.The work was not contingent on maternalsustained control. On occasion, the child in-dicated his or her commitment ("It has to beclean," "These look neat now," "I'll cleanthis mess"). "Don't" context: Even thoughchild attention was on the prohibited ob-jects, as indicated by looking, pointing, ortalking, he or she appeared to have acceptedand "embraced" the prohibition, did not at-tempt to touch, or expressed internal obliga-tion ("These are the pretties, not to touch";"Let's not play with these").

2. Situational compliance: Acceptanceof maternal agenda. "Do" context: Child wasgenerally cooperative, but required mater-nal sustained or repeated control to stay ontask; otherwise, was likely to get distractedand/or cease working. "Don't" context:Child responded to maternal prohibition ordistraction with at least temporary compli-ance, but compliance seemed shaky, notfully endorsed, and child tended to hoveraround the prohibited toys and to attempt totouch them again.

3. Passive noncompliance: Reluctanceto accept matemal agenda. "Do" context:Child did not stay on task on his or her own;

^ Although in past work (e.g., Kuczynski & Kochanska, 1990) negotiation and refusal wereconsidered to be developmentally more advanced and skillful than defiance and passive noncom-pliance, they nevertheless indicate reluctance to accept matemal agenda. In the current study,they were associated negatively with intemalization, at both Time 1 and 2. Thus, they werecombined to help streamline the analyses, which were focused primarily on the differentiationbetween committed and situational compliance (see also the discussion in Kochanska & Aksan,1995).

1756 Child Development

when^ prompted, ignored the mother."Don't" context: The child ignored maternalprohibition and attempted to play with theobjects.

4. Refusal/Negotiation/Defiance: Overtprotest to maternal agenda. "Do" context:Child did not clean up, and refused overtlyif prompted, either nonaversively or angrily,throwing toys, doing the opposite to direc-tive. "Don't" context: Child tiied to playwith the objects and overtly refused, argued,or angrily resisted the mother.

5. Time out ("Do")/Other ("Don't")."Do" context: Mother and child took a breakfrom the cleanup or focused on other issues."Don't" context: Mother allowed child totouch a prohibited toy. This code was tan-gential and thus was not considered further.Reliability was established on at least 20%of cases at each time. Kappas were (Time 1first. Time 2 second): for the cleanups, .78and .63, for the prohibition context, .81 and.78. Because the number of coded segmentsvaried, relative scores were computed foreach child. The frequency of each categorywas divided by the number ofthe coded seg-ments.

The coding of mother and child af-fect.—Four codes were used for bothmother and child: highly positive, neutral/pleasant, neutial/negative, and highly nega-tive (for details, see Kochanska & Aksan,1995). More than one affect could be codedin one segment. Reliability (kappas. Time 1first. Time 2 second) were for mother affect.76 and .81, for child affect .77 and .84. Wethen computed, at Time 1 and 2, for eachdyad, the number of instances when bothmother and child spent an entire coded unitin highly positive or neutral positive affect,and no other affect was coded for either per-son (mother-child shared positive affect).The number of such instances was dividedby the number of coded units and averagedacross the "do" and "don't" contexts.

PARADIGMS AND MEASURES ASSESSING THECHILD s INTERNALIZATION AT TIME 2

Internalization of Maternal "Dos" and"Don'ts"

Internalization of the "Do" request.—After 10 min of the child's work on the toycleanup, the experimenter asked the motherto move to the other room and to request thechild to complete the task by him/herself.The child was then given 5 more minutes tocontinue the cleanup.

g E v e r y 10-sec segment wascoded. Reliability (kappa), on 20% of cases,was .86. Ultimately, only one code was usedfurther, internalized cleanup, when thechild was engaged in the sustained cleaningup ofthe toys, and there was no play or otheractivity in the segment. Other codes (transi-tory behavior, addressing mother/experi-menter, playing with the toys or other ob-jects, leaving the room) all negativelycorrelated with internalized cleanup (rs- .27 to - .43), and were dropped. For eachchild, the frequency of internalized cleanupwas divided by the number of segmentswhen any toys remained on the Hoor. Typi-cally, many toys remained scattered aroundat the beginning of this period, but 14children had completed the task beforemother's departure, and thus iV for this vari-able is 85.

Intemalization of the "don't" prohibi-tion.—The I2-min paradigm was identicalat Time 1 and 2 (details are in Kochanska &Aksan, 1995). The child was left alone in theliving room with a dull sorting task. Motherreminded him or her about not touching theobjects on the shelf, left the room, andclosed the door. First, the child was alone(3 min); then an unfamiliar female entered,played with the most attractive toys, woundup the music box, and lefl (1 min); then thechild was again alone (8 min).

Coding.—For each 5-sec interval, one ofthe six mutually exclusive codes was used.(1) Looking: looks at objects, but does nottouch (M = 23.40, SD - 15.44). (2) Selfcorrection: touches an object briefly (up to2 sec), or extends hand but then withdraws(M = ,49, SD = .81). (3) Gentle touch:touches objects gently/tentatively (M =4.78, SD = 6.73). (4) Deviation: plays withobjects in an unrestrained manner (M ~15.84, SD = 30.53). (5) Competing activity:wanders around, snacks, plays with "legal"toys (M = 43.67, SD = 33.32). (6) Busy withsorting task (M = 52.88, SD = 41.52). Seg-ments when child addressed the motherwere marked, but not used in the analyses.Reliability, kappa, on 15% of cases (2,444coded segments), was .92, at Time 2. Sepa-rate percent agreements ranged from 61% to95%.

Data reduction.—We submitted thecodes to the principal components analysis(Varimax rotation) to reduce the number ofvariables. It produced two factors, both in-temalization-relevant when reversed. Thefirst one, Task-Oriented (eigenvalue 2.10,

Kochanska, Aksan, and Koenig 1757

35% of variance), described a focus on thesorting task: high sorting (.93), and lowcompeting activity ( - .71), self-correction(— .56), and gentle touch (— .46). The secondone, resembling an analogous factor at Time1 (Kochanska & Aksan, 1995), Behavioral In-ternalization (eigenvalue 1.70, 28% of vari-ance), described low deviation (-.81), highcompeting activity (.52) and looking (.67),and low gentle touch ( - .60).

Internalization of the Experimenter'sRules of Conduct ("Cheating Games")

At three points during the session, theexperimenter invited the child to play "spe-cial games to win great prizes." Prior to eachgame, she brought in a large bin filled withattractively wrapped packages (the prizes).Each game took place in the playroom,while the mother remained in the livingroom.

The animal game?-—The experimenterbrought in a basket with several animals fora "guessing game," and showed the childthree characteristically shaped animals for apractice trial. When the child turned around,she covered one of them with cloth, and ex-plained that the task was to guess the animalwithout lifting the fabric or peeking underit. The only form of touch allowed was usingthe tip of one finger of one hand. Due tothe telltale shape (a large Ninja Turtle), eachchild succeeded. The experimenter praisedthe child, said they would now "play forreal," and covered three other animals andthe basket with separate pieces of cloth. Be-cause odd-shaped animals were used (e.g., asmall bunny placed on a large jar), the taskwas now practically unsolvable. The childwas promised a prize for each animalguessed. After the first two unsuccessfulguesses, the experimenter "remembered"that she had to leave to go to the other room,and asked the child to play in her absence.

Prior to leaving, she stressed the impor-tance of not cheating. Because the pilot workhad found that not all preschoolers under-stood the word "cheating," she then ex-plained, to every child, that "cheating wasnot following the rules, doing what is notfair, doing what we are not supposed to do."This standard statement, the purpose ofwhich was to qualify negatively potentialrule violations, was delivered in a friendlybut serious manner. Finally, the experi-menter reviewed the rules with the child to

make sure the child understood them, en-couraged him or her to play, and left, closingthe door.

The child was alone for 3 min. After 2min, the experimenter called from the otherroom that she would be coming back shortly.Just before her return, she again called thatshe was coming back. Upon return, she sud-denly "noticed she had used the wrong ani-mals," apologized, and they played the gameover. This time, because a characteristicallyshaped caterpillar was used, every child suc-ceeded and won a prize.

The bird game.—Later into the session,the experimenter invited the child to playanother game with prizes in the playroom.The game involved a wooden meadow, 20inches x 30 inches, upholstered with greenfelt, with 30 colorful feathery birds perchedon round wooden disks. The game wasplaced on the fioor. The experimenter ex-plained that some ("magic") birds had happyfaces on the disks, and some, sad faces (stick-ers). To win a prize, one needed to find a"magic bird." She stressed that the rule al-lowed touching only two birds. Once a birdwas touched, it was to be put aside in one oftwo small baskets. She showed how to play:pretending to select thoughtfully, shepicked the first bird, acted disappointed todiscover its sad face, and put it in the basket.Then she picked another bird, with a happyface, and acted excited about winning.

Again, the experimenter "remembered"she had to leave for a while, asked the childto play the game in her absence, reviewedthe rule, and stressed "not cheating." Childwas then alone for 3 min (now, there wereno more "magic" birds on the meadow).Upon return, the experimenter "discovered"she had "messed up" by forgetting to putthe magic bird there. When the child cov-ered his or her eyes, the experimenter addeda visibly attractive bird (with a happy face).Every child selected it and won a prize.

The dart game.—Toward the end of thesession, the experimenter invited the childto play a game of darts. With narrow boards,she outlined a small space on the fioor as aspot where the child had to remain through-out the game, placed a plastic ring close tothe child, and explained that the goal was tothrow five rubber darts into the ring whileremaining fully within the confined space.

This game was adapted, with considerahle modifications regarding its rules and the coding,from Asendorpf and Nunner-Winkler (1992).

1758 Child Development

She demonstrated a successful throw, aswell as an instance when she broke the ruleby putting one foot outside of the space.After the child's successful trial, she empha-sized that the child should try hard to aimwell, because he or she could not leave thespace after all the darts had been thrown—itwas illegal to retrieve the darts and try again,as well as to come closer while throwing.The child was promised a prize for each dartin the ring. The experimenter "had to leavefor a while," reviewed the rules, stressed"not cheating," and asked the child to play.Just prior to leaving, she moved the ringmuch farther away, so it became practicallyimpossible to throw the darts into it. Thechild was left alone for 3 min; upon return,the experimenter "discovered" she had setup the game wrong, apologized, and let thechild play again, now with the ring widenedand placed closer. Each child won a prize.

Coding.—In every game, the 60 3-secsegments when the child was alone werecoded. To create meaningful compositesacross all games, we focused on three maincategories coded in similar manner in allthree tasks: the latencies to transgress, theextent of transgression, and rule-compatibleconduct.

The latencies to transgress.—All laten-cies were coded in seconds. If the childnever transgressed, the score was 180. Themeasures were: in the animal game, the la-tency to the first instance of lifting the cloth(M = 130.96, SD = 73.47); in the bird game,to touch the first illegal (third) bird (Af =42.00, SD = 58.43) and to examine the firstillegal (third) bird (M = 49.59, SD = 65.07);in the dart game, to put one foot outside themarked space (M = 60,12, SD = 70.22), tomove the whole body outside the space butin immediate proximity (M = 84.17, SD =78.65), and to take at least one step awayfrom the space (M = 86.56, SD = 79.06).

The extent of transgression.—Behav-ioral categories, coded for each of 60 3-secintervals when child was alone, included: inthe animal game, all instances when childtouched the animals with both hands (M =.93, SD = 3.48), lifted cloth, including thatover the basket (M = 3.10, SD = 4.16), orpeeked under cloth (M = 1.04, SD = 2.04),and, in addition, the number of animals seen(M = .61, SD = 1.00); in the bird game,all instances of illegal examinations (beyondsecond bird, M = 17.25, SD = 14.56); and in

the dart game, all segments when the childmoved the ring closer (M = .56, SD = 1.26),was outside the space (M = 4.94, SD =8.70), and the number of darts manually putin ring (M = .94, SD = 1.93) or retrieved(M = 2.53, SD = 4.56).

Rule-compatible conduct.—These cate-gories, also coded for every 3-sec interval,included: in the anim^vl game, all instancesof looking at the animals without touching(M - 13.05, SD = 13.11) and of touchingcorrectly with a tip of one finger (M = 7.80,SD = 13.29); in the bird game, all instancesof looking at the birds without touching (M= 17.84, SD == 13.50); and in the dart game,all instances when the child was fiiUy withinthe designated space (M = 35.97, SD =21.17).

Reliability.—Reliability was estab-lished on 10 cases, using all three games foreach child (1,800 coded segments). For thejudgments of latency, 85% were identical,14% differed by 1 sec, and 1% diflPered by 2sec. For the behavioral categories coded forevery 3-sec segment, the kappas^ were .85for the animal game, .89 for the bird game,and .92 for the dart game. The number ofanimals seen and the number of darts re-trieved and manually put in the ring werejudged identically in every instance. Thenumber of illegally examined birds waswithin the difference of 1 for eight cases, andwithin 10% of the score for the other twocases.

Data aggregation.—First, we standard-ized the variables. Then, all latencies wereaggregated into the latency to transgress,alpha = .78, all transgressions into the ex-tent of transgression, alpha = .76, and all"legal' behaviors into rule-compatible con-duct, alpha = .55.

Matemal Reports of Children'sInternalization

As at Time 1, we used a 20-item scaleof internalization that describes the child'scompliance widi family rules without super-vision, a part of a larger instrument to assessyoung children's conscience (Kochanska,DeVet, Goldman, Murray, & Putnam, 1994).Each item is rated from 1 (extremely untrue)to 7 (extremely true). Time 2 alpha was .88.

The toy cleanups, the prohibition con-text, child alone widi the forbidden toys, andthe cheating games were coded by indepen-dent teams of coders. Within each team, the

^ Other codes were used but not included in the analyses, for example, child engaged inan unrelated activity rather than playing the game.

Kochanska, Aksan, and Koenig 1759

coders coded several cases together every cused on the replication ofthe original Timefew weeks to prevent drift. The descriptive 1 findings, the first goal ofthe study. We ex-statistics are in Table 1. amined the relations among the qualitatively

different forms of the child's compliance.Results noncompliance, and mother-child sharedOVERVIEW positive affect, and the sex and age effects;

explored the relations of children's compH-There were two major sets of analyses. ance, noncompliance, and mother-child pos-

In the first set, all analyses were contempo- Itive affect with the conscience measures;raneous, involving the Time 2 data. We fo- and compared the rates of different forms of

TABLE 1

CHILD COMPLIANCE/NONCOMPLIANCE, MOTHER-CHILD MUTUALLY POSITIVE AFFECT, AND CHILDINTEBNALIZATION; DESCRIPTIVE DATA

TIME 1 (N = 103) TIME 2 (N = 99)

SD A/ SD

Mother-Child Control InteractionComposites across "Do" and "Don't" Contexts

Child corapliance/noncompliance;Committed compliance 41 .23 .56 .18Situational compliance 33 .14 .22 .11Passive noncompliance 13 .12 .14 .13Overt protest 09 .10 .05 .07

Dyadic affect:Mutually positive affect 75 .20 .77 .19

"Do" Contest (Toy Cleanups)Child compliance/noncompliance:

Committed compliance 22 .21 .28 .25Situational compliance 47 .19 ,36 .18Passive noncompliance 14 .13 .22 .21Overt protest 10 .12 .08 .11

Dyadic affect:Mutually posiHve affect 74 .20 .70 .23

"Don't" Context (Prohibited Toys)Child compliance/noncompliance:

Committed compliance 60 .35 .84 .25Situational compliance 19 .19 .08 .12Passive noncompliance 12 .18 .05 ,12Overt protest 07 .13 .03 ,08

Dyadic affect:Mutually positive affect 76 .28 .83 ,24

Children's Intematization at Time 2Toy cleanup:^

Internalized cleanup .44 .38Alone with prohihited toys:''

Task-oriented .00Behavioral internalization .00

Cheating games:''Latency to transgress .00Extent of transgression .00Rule-compatible conduct .00Intemalization composite .00

Mother report:Internalization scale 3.78

NOTE,—AJl mother-child interaction variables are relative scores (percent of the coded segments).' Percent of coded segments with toys to be cleaned up.*• Factor scores produced by PCA." Following the standardization and aggregation ofthe component variables.

11.00.00

.69

.59

.65

.56

.78

1760 Child Development

child compliance and noncompliance in the"do" and "don't" contexts and coherenceacross the contexts for individual children.

The second set addressed the secondgoal, the longitudinal questions. We exam-ined developmental changes in differentforms of children's compliance and noncom-piiance between Time 1 and Time 2. Weexplored if there was continuity, for the indi-vidual children and dyads, in forms of com-pliance/noncompliance and in dyadic affect.We also focused on longitudinal predictionsof conscience. We examined whether childcommitted compliance and mother-childpositive affect, and the dyadic factor SharedAffective Positivity/Child Cooperation, mea-sured at Time 1, predicted child internal-ization at Time 2. To address the thirdgoal, further examination of the "do" and"don't" contexts, we included their separatemeasures in most analyses (in addition to thecomposite scores for compliance and non-compliance across both contexts). The corre-lations for girls and boys, as at Time 1, wereessentially similar. Thus, they are presentedfoT the entile sample.

Contemporaneous Analyses ofthe Time 2Data: The Replication of Time 1 Findings

Do forms of children's compliance andnoncompliance remain qualitatively differ-ent at preschool age?—We examined the in-tercorrelations among the forms of childcompliance and noncompliance. Not sur-prisingly, and as at Time 1, not only did com-nmitted compliance relate negatively to non-compliance, but more important, it relatednegatively to situational compliance (Table2).

Are there age and sex effects in theforms of children's compliance and noncom-pliance?—There were no significant corre-

lations with child age at Time 2 (Table 2). AMANOVA, with the four forms of compli-ance and noncompliance as dependent vari-ables (composites), and child sex as the be-tween-subject factor, produced, as at Time 1,significant sex differences, F(4, 94) = 3.29,p < .025. The univariate analyses revealedthat they were significant for three out of thefour behaviors. As at Time 1, girls showedmore committed compliance (M = .62, SD- .17) than boys (M = .50, SD - .18), F(l,97) = 10.76, p < .001, and less passive non-compliance (girls M = .11, SD = .11, boysM = .16, SD = .14), F(l, 97) - 4.21, p <.05. A new finding at Time 2 was lower sitna-tional compliance in girls (M = .19, SD =.11) compared to boys (M = .25, SD - .11),F(l, 97) = 6.40, p < .025.

Are forms of children's committed com-pliance associated with mother-child sharedpositive affect at preschool age?—We rep-licated the Time 1 findings. There was anegative association between mother-childshared positive affect and the child's overtprotest, and a positive one with committedcompliance.

Are the child's committed complianceand mother-child shared positive affect as-sociated with internalization at preschoolage?—The contemporaneous correlationsbetween child compliance and noncompli-ance, mother-child shared positive affect,and internalization measures aFe presentedin Table 3. For the composite scores acrossthe "do" and "don't" contexts, the data pro-vide a powerful replication of the Time 1associations between committed (but not sit-uational) compliance and the now expandedchild intemalization measures. Childrenwho appeared to embrace and endorse thematernal agenda during the control enconn-

TABLE 2

ASSORIATIONH AMONG FORMS OF CHILDREN'S COMPLIANCE AND NONCOMPLIANCE, AGE, ANDMOTHER-CHILD MUTUALLY POSITIVE AFFECI' AT TIME 2

Committed complianceSituational compliancePassive noncomplianceOvert protestMother-child mutually

positive affect

• p < ,05,'•* p < ,025,•*' () < ,001,

Committed Situational PassiveCompliance Compliiince Noncompliance

-.56*** -.68***-,04

.2V - .08 -.01

OvertProtest

-.45***-.00

.12

-.50*

Age

.11

.09-.18-,03

- .26*

l l

^ a0= B

oU

bOi n

(? f̂̂ ^p-Hp pp-Hpp' i' i' r ' I I I I

inooo<N»nOC<l(MCO

J 3

£i i"3J

ess

&

aH

_ o

2 ,a-c P

1 1 1.N

£

<;

£

-ai

£iiE u

( N O I W C O HOOOOO

I I I

I I I I I I

I I I I I

§ o2 u

*; oi« " "

O -D

II § 1S V o fi s =

1762 Child Developmentters, while alone spent more time complet-ing a mundane task, were more internalizedwith prohibited toys, took longer to trans-gress and complied with the rules more inthe cheating games, and were described bythe mothers as more internalized in dailylives. In contrast, children whose compli-ance was sitnatlonal rather than committedin their control interactions with their moth-ers were less internalized on several oftheinternalization measures. Again, childrenwho ignored or actively rejected the mater-nal agenda in the interactive contexts ap-peared less internalized while alone.

Further inspection of Table 3 revealssurprising "specificity" of these associa-tions. The measure of internalization in thetask calling for the maintenance of the pre-scribed mundane activity {cleanup alone)was related to the forms of compliance andnoncompliance in the "do" context. On theother hand, the measures of intemalizationin the tasks that called for restraint (proscrip-tions), such as not touching the prohibitedtoys or not cheating in the animal, bird, anddart games, were strongly related to theforms of compliance and noncompliance inthe "don't" control context. Matemal ratings^that encompassed both "do" and "don't"family rules related to children's complianceand noncompliance in both types of controlcontexts.

In contrast to Time 1, when mother-child shared positive affect during controlinteractions was positively associated withchildren's intemalization, all but one rela-tion at preschool age were not significant.

Are "do" demands more challengingthan "don't" demands for preschool chil-dren?~We conducted a MANOVA with thescores on child behaviors {committed andsituational compliance, passive noncompli-ance, and overt protest) as the dependentvariables. For each variable, we entered thescore for the "do" context and the one forthe "don't" context. Thus, the type of con-text {"do" vs. "don't") and the form of re-sponse (one of the four) were the within-subject factors. Sex of child was thebetween-subject factor (but it did not pro-duce significant findings beyond those de-scribed previously for die composites acrossthe "do" and "don't" contexts).

As at toddler age, we found a significantmultivariate interaction effect of the form ofresponse and the type of context, F{3, 95) =99.13, p < .001. At preschool age, childrenagain responded differently to matemal re-

quests and prohibitions. T tests indicatedthat the "dos" (toy cleanup) were more chal-lenging than the "don'ts" {not touchingtoys). During the cleanup, compared to theprohibition context, children's committedcompliance was lower {t = -16.44, p <.001), whereas situational compliance, pas-sive noncompliance, and overt protest werehigher (ts, respectively, 13.47, 7.59, and4.01, all p s< .001). All means are in Table 1.

Are the forms of children's complianceand noncompliance coherent across the"do" and "don't" contexts?—In contrast tothe original assessment at toddler age, wefailed to find coheience across contexts atpreschool age. The correlations between the"do" and "don't" contexts were .08 for com-mitted compiiance, .08 for situational com-pliance, .13 for passive noncompliance, and.03 for overt protest, indicating a relative in-dependence of children's responses in thetwo contexts.

Longitudinal AnalysesAre there longitudinal changes between

toddler and preschool age in the rates oftheforms of the child's compliance and non-compliance?—We first conducted an overallMANOVA to determine the presence of sig-nificant developmental effects. Then, we fol-lowed up w i ^ a series of MANOVAs, re-peated-measures ANOVAs, and t tests toclarify the specific developmental changes.They will be described in conjunction witheach overall effect.

In the overall MANOVA, child compli-ance and noncompliance scores were the de-pendent variables. Because, for each child,there were four forms of response {commit-ted and situational compliance, passive non-compiiance, and overt protest), the firstwithin-subject factor was "the form oi re-sponse." Because, for each form, we enteredchild scores for the "do" context (cleanup)and for the "don't" context {not touchingtoys), the second within-subject factor was"type of context." And, because, for eachscore, that for Time 1 and that for Time 2were entered, the third within-subject factorwas "time of assessment." Child sex was tiiebetween-subjeet factor, dne to the noted sexdifferences {but it did not produce any sig-nificant interactions involving the time of as-sessment).

The overall MANOVA indicated thatthe rates of different forms of complianceand noncompliance changed significantlybetween toddler and preschool age, as re-fiected in the significant interaction of the

Kochanska, Aksan, and Koenig 1763

form of response and time of assessment,F(3, 95) = 29.66, p < .001. We then exam-ined each form of child response separatelyusing paired t tests. Compared to toddlerage, at preschool age ehildren more oftenshowed committed compliance {( = —6.78,p < .001), but less often situational compli-ance (t = 6.95, p < .001) and overt protest(t = 3.02, p < .005). The rate of passive non-compliance did not change. The means arein Table 1 (see the composite scores).

Moreover, the results indicated that cer-tain forms of child compliance/noncom-pliance changed differently over timedepending on the context {"do" versus"don't"). It was reflected by a significant ef-fect of interaction of the form of response,context, and time of assessment, F(3, 95) =12.04, p < .001. To clarify these findings, weconducted four repeated-measures ANOVAs(separately for committed and situationalcompliance, passive noncompliance, andovert protest), in each entering separatescores for the "do" and "don't" contexts, atTime 1 and at Time 2. Thus, in each analy-sis, the context ("do" vs. "don't") and thetime of assessment were the witbin-subjectfactors, and sex was the between-subject fac-tor. All means are in Table 1 {see the sepa-rate "do" and "don't" scores).

The effect in question (interaction ofcontext and time of assessment) was signifi-cant for two forms of child behavior, commit-

ted compliance, F(l, 97) = 20.17, p < .001,and passive noncompliance, F{1, 97) =33.22, p < .001. Although committed compli-ance increased between toddler and pre-school age for both "do" and "don't" con-texts, the change in the "don't" context wassignificantly greater (7% change in "do"compared to 25% change in "don't"). Pas-sive noncompliance (which showed no de-velopmental change as a composite acrossboth contexts), increased by 8% betweenTime 1 and Time 2 for the "do" conttixt(( = -3.80, p < .001), but decreased by 7%for the "don't" context {* = 3.75, p < .001).

Are there longitudinal continuities inchildren's forms of compliance and noncom-pliance?—The data are in Table 4. The com-posite scores, as well as the scores in the"do" context on committed compliance andpassive noncompliance, were significantlycontinuous between toddler and preschoolage. In the "don't" context, all forms of com-piiance and noncompliance were signifi-cantly related across assessments.

We also examined the continuity of theshared positive affect in individual mother-child dyads. It was somewhat stable; for thetota! scores, r = .26, p < .001; for the "do"context, r = .22, p < .05; and for the "don't"context, r = .20, p < .05.

Do qualitatively different forms ofchild compliance/noncompliance, shared

TABLE 4

LONGITUDINAL CONTINUITY OF THE FORMS OF CHILDREN'S COMPLIANCE AND NONCOMPLIANCE

TIME 1

TIME 2

Committed Situational Passive OvertCompliance Compliance Noncompliance Protest

Composites across "do" and "don't" contexts:Committed compliance 36***Situational compliance - .15Passive noncompliance -.20*Overt protest -.23**

"Do" context (toy cleanups);Committed compliance 26***Situational compliance - .13Passive noncompliance -.09Overt protest -.06

"Don't" context (prohibited toys):Committed compliance 45**»Situational compliance -.30***Passive noncompliance -.28***Overt protest __29***

• p < .05.**p< .025.'"*p< .01.

-.22*.12.03.14

-.14.17

-.17.05

-.35***.21*.24**.23**

-.29***.13.25**.11

-.27***.02.33***.01

-.37***.24**.22*.22*

-.15.12.06.12

-.04.13

- .01.00

-.29***23**.16.25**

1764 Child Development

positive affect, and Shared Affective Posi-tivity/Child Cooperation dyadic scores attoddler age predict intemalization at pre-school age?—The data in Table 5 stronglysupport the prediction that children who astoddlers had eagerly embraced the motheragenda would show higher intemalization13-15 months later, for the host of measures.Toddlers who had been high on committedcompliance, as preschoolers spent moretime completing a mnndane task withoutsurveillance, showed more task orientationand restraint while alone with the pro-hibited toys, were more reluctant to trans-gress and more likely to engage in rule-compatible condnct in the cheating games,and were rated higher by mothers on inter-nalization of the family rules. Toddler-agesitnational compliance had either no or neg-ative relations with preschooKage intemal-ization. Toddler-age passive noncomplianceand overt protest, for the most part, relatednegatively to preschool-age intemalization.

There was some indication that mother-child shared positive affect at toddler agepredicted child intemalization at preschoolage. In particular, children who as toddlerswere likely to share positive affect with theirmothers in the toy cleanups, as preschoolerswere more task-oriented when alone withprohibited toys, had longer latencies totransgress in the cheating games, and wererated by the mothers as more internalized inthe daily lives.

The dyadic scores, compnted at toddlerage, of Shared Affective Positivity/Child Co-operation continued to predict intemaliza-tion at preschool age, just as they did at tod-dler age. Children from the dyads who attoddler age were characterized as mntuallypositive and cooperative, at preschool agespent more time completing the cleanuptask alone, were more internalized with theprohibited toys, had longer latencies totransgress in cheating tasks, and were per-ceived by their mothers as more internalizedin their daily lives.

Does committed compliance at toddlerand preschool age make unique contribu-tions to intemalization?—In the last seriesof analyses, we asked abont unique contri-butions of Time 1 and Time 2 committedcompliance to intemalization at Time 2.Does contemporaneous committed compli-ance at preschool age predict intemaliza-tion beyond the contribution of committedcompliance at toddler age? In hierarchicalmultiple regressions, Time 2 intemaliza-

tion measures were the dependent vari-ables (intemalized cleanup, two intemahza-tion factors while alone with prohibited toys,and matemal rating). To reduce the num-ber of equations, we aggregated latency, re-versed extent of transgressions, and rule-compatible conduct from the cheating gamesinto one intemalization composite (alpha =.83). Because girls scored higher on someintemalization measures, we entered chiidsex at Step 1. Then, following the develop-mental seqnence, we entered committedcompliance at Time 1 at Step 2, and that atTime 2 at Step 3.

For intemalized cleanup, child sex wasnot a significant contributor. Corrimittedcompliance at toddler age was a significantpredictor, Feh{2, 82) = 4.89, p < .05, adding6% of explained variance, and so was com-mitted compliance at preschool age, adding6% of explained variance, F ,̂h(3, 81) = 5.71,p < .025.

For both intemalization scores whilealone with prohibited toys, sex was signifi-cant (girls scored higher). For the Task-Oriented score, committed compliance attoddler age was marginally significant, Fch(2,96) = 3.46, p = .07, adding 3% of variance,and committed compliance at preschool agewas significant, F,h(3, 95) - 4.48, p < .05,adding 4% of variance. For the BehavioralIntemalization score, toddler age commit-ted compliance was significant, Fch(2, 96) =4.89, p < .05, adding 5% of variance, and thatat preschool age was marginally significant,F^h(3, 95) - 2.92, p < .10, adding 3% ofvariance.

For the cheating games intemalizationcomposite, sex was not a significant pre-dictor. Both committed compliance scoreswere significant; that at toddler age added6% of variance, F,h(2, 96) = 5.84, p < .025,and that at preschool age, 4% of explainedvariance, F,,h(3, 95) = 4.11, p < .05.

Finally, for the matemal rating ofinter-nalization, sex was significant, wim girlsseen as more intemalized. Toddler age com-mitted compliance contributed an additional3% of variance, marginally significant, F .̂h(2,96) ^ 3.30, p < .10, and preschool age com-mitted compliance added 11% of explainedvariance, highly significant, F^^(3, 95) =12.63, p < .001.

DieeussionThe results of this study provide further

powerful confirmation of our model of thedevelopmental dynamics of compliance and

^I I I 1 1

3OS

"SaBc

U

O <M - - ' O IO(N O -H (M -H

I I I I I

SnObe.£13u

o

*^w

o

!ssi'

1 - .

bi)a;cfl

H

oH

o beg fl

E- 'CC

dj 3C £=

' ^

m;

Inte

aM

u

O -H O CDCO O CO —'

I I I I

I I I I I

I I I

I I I

-aG

c c S

PI2 E-"

a >

-•S s

,1^

•5

D

> O

1766 Child Development

intemalization. The view of compliance as acomplex phenomenon that encompassesqualitatively different forms has gained fur-ther support by the longitudinal data.

Committed and situational complianceat preschool age, as at toddler age, appearedto be distinct. They were strongly negativelycorrelated, and they had different devel-opmental trajectories. Between toddler andpreschool age, the committed, self-^regulatedform increased, and the situational form de-creased, as expected. Perhaps that explainswhy in past work {Kuczynski & Kochanska,1990), in a comparable longitudinal study,no developmental changes were found incompliance, then considered as an undiffer-entiated construct.

Perhaps most important, however, werethe links with intemalization, also assessedat preschool age, that support our argumentthat committed compliance is strongly asso-ciated with intemalization, and indeed maylegitimately be viewed as a form of earlyconscience. Children who appeared to em-brace, endorse, and accept as their own theirmothers' agendas also showed higher inter-nalization on several measures. Childrenwho were cooperative in response to mater-nal immediate control bnt lacked sincerecommitment or dedication showed evidenceof much poorer intemalization.

Committed compliance had significantlongitudinal continuity between toddler andpreschool age, which supports our view thatit is rooted in long-term qualities ofthe rela-tionship with the parent. Maccoby (1983;Maccoby & Martin, 1983; Parpal & Mac-coby, 1985) discussed conditions conduciveto the emergence of the child's receptivestance toward parental socialization and ea-gemess to accept parental influence. Shepointed out that such receptive compliance("committed," in our terms) emerges in thecontext of a mutually responsive and recip-rocal cooperative set between the parent andchild.

Perhaps the most powerful evidenceabout the developmental implications ofcommitted and situational compliancecomes from our longitudinal findings. Com-mitted compliance was not only contem-poraneously associated with child intemal-ization, but it also predicted futureintemalization. We also found that, for mostpreschool intemalization measures, boththe toddler-age and preschool-age commit-ted compliance made unique contributions.These findings are remarkably consistent

with the recent emphasis on the toddler pe-riod as critical for the early development ofconscience (Emde et al., 1991; Kochanska,1994), and in stark contrast to the cognitive-developmental approaches that disregardedthe role of early childhood in moral devel-opment.

As at Time 1, we explored the impor-tance ofthe shared positive affective set be-tween the mother and child at preschool age.Interestingly, in contrast to toddler age, wedid not find strong associations betweenmother-child positive affect and children'sintemalization at preschool age, and theonly significant findings pertained to longi-tudinal predictions. Children who sharedpositive affective set wi& their mothers astoddlers (or were in dyads who scored highon Shared Positive Affectivity/Child Coop-eration) were more intemalized on somemeasures as preschoolers. The failure to rep-licate the contemporaneous associations inthe preschool age data may again indicate aparticular importance of mother-child rela-tionship in the toddler period. Perhaps thequ^ity of mother-child affeetive climate inthe first 3 years, at least regarding compli-ance and intemalization, is more significantand has more lasting developmental impli-cations than at preschool age. It is also possi-ble that affect is the central currency of ex-change in toddlerhood, and dius it may bethe best indicator of mutual cooperation andresponsiveness for young children. As chil-dren grow older, nonverbal affect expressionmay give way to more verbal forms. Thus,for older children, we may need to find age-appropriate measures ofthe mutually coop-erative dyadic climate, such as qualities ofthe discourse between parent and child.

This study confirms the importance ofdifferentiating between two contexts of de-mands, dos and don'ts. With a few excep-tions (Kochanska & AJcsan, 1995; Schneider-Rosen 6c Wenz-Gross, 1990; Vaughn, Kopp,& Krakow, 1984), this issue has rarely beensystematically studied, although is seen asimportant (Emde, Johnson, & Easterbrooks,1987). It appears from our toddler and pre-school-age data that its developmental sig-nificance may be considerable.

At preschool age, the links betweenforms of compliance and noncompliance inthe "do" and "don't" discipline contexts andchildren's intemalization appear to be quitecontext-specific. They suggest that perhapsintemai systems of standards for restraint

Kochanska, Aksan, and Koenig 1767

(proscriptions) and standards for production(prescriptions) either have relatively inde-pendent developmental origins, or that inearly childhood a marked differentiation oc-curs between the two systems. Recall thatthe forms of compliance/noncompliance inthe "do" contexts were linked to production-type intemalization (completing a mundanetask), whereas the forms of compliance/non-compliance in the "don't" contexts werelinked to intemalization in paradigms re-quiring restraint.

Similarly, for individual children, therewas little continuity in the forms of compli-ance across the two contexts, despite the factthat some coherence had been found at tod-dler age (Kochanska & Aksan, 1995). It sug-gests that over the course of development,the two systems may indeed diverge ratherthan converge. The "do" system appears,just as at toddler age, to be more challengingand difficult for children and parents. Chil-dren showed significantly less committedcompliance and more sitnational complianceand noncompliance when requested to sus-tain the on-task behavior (cleanup) thanwhen asked to refrain from touching thetoys. We also explored whether this diffi-culty was reflected in the affect within thedyads, and found that, indeed, mothers andchildren spent a lower proportion of thecleanup-related control episodes in mutu-ally positive affect (M = .70, SD = .23) com-pared to the control episodes triggered by areference to the prohibited toys (M = .83,SD = .24; t = -4.44, p < .001).

This study occasions broader reflectionon parental control and child compliance. Insome respects, parental control is not unlikeother processes of social influence. Analysesmay focus on the amount of power requiredfor a successful intervention, or the clarity oreffectiveness of different forms of influencetechniques. The effectiveness of parentalcontrol is reflected in child compliance, of-ten defined in terms of close temporal prox-imity to the directive (often as short as 5-10sec [Forehand, 1977], or "reasonable time"[Crockenberg & Litman, 1990]). Such an ap-proach does not require qualitative differen-tiations within compliant responses (al-though different forms of noncompliancehave been distinguished; Crockenberg &Litman, 1990).

We argne that a closer analysis revealsmotivational complexities within compli-ance, and that committed compliance cannot

be fully explained in terms of qualities ofthesocial influence process. Rather, it is betterconceived in the context ofthe history oftheparent-child relationship. In terms of conse-quences for socialization, while situationalcompliance seems limited to the "here andnow" ofthe control context, committed com-pliance appears to involve a deeper andmore long-term attitude change or intemal-ization.

Because ofthe important consequencesof committed compliance for socialization,more research is needed on factors that pro-mote it. On the parents' side, future workshould focus on responsiveness to the child,which likely contributes to the mutually co-operative dyadic orientation, either long-term, like secure attachment, or situational(Parpal & Maccoby, 1985). Perhaps we canidentify personality factors that make someparents more effective in establishing suchmutually binding sets. Parental gender maybe one such variable (Maccoby, 1990), withwomen being particularly apt to enter suchrelationships (perhaps particularly withtheir daughters, consistent with higher com-mitted compliance in girls at both times). Weshould also, however, address fathers' abi]-ity to form systems of mutual cooperationwith their children. Other qualities of par-ents, such as empathy, may also surface asimportant contributors. Additionally, a long-term pattern of parent discipline, based onnegotiation rather than power (perhaps, intum, made possible by the already existingmutual cooperative set), may increase thelikelihood of committed compliance (seeanalysis of "mature compliance"; Crocken-berg, 1991). On the children's side, we em-phasize the role of temperament, in particu-lar, inhibitory control and fearfulness, asqualities that may be important underpin-nings of committed compliance and intemal-ization (Kochanska, 1993). We also describethe links between temperament, committedcompliance, and parental socialization (Ko-chanska, 1995). Additionally, cognitive func-tions such as memory or attention may alsocontribute to the committed or situationalquality of child compliance, and they de-serve research attention.

The return of intemalization and con-science to the developmental stage has re-opened exciting classic questions in socialdevelopment. Perhaps the near fnture willbring real progress in our understanding ofthe fundamental processes in parent-childrelationship and socialization.

1768 Child Development

References

Asendorpf, J. B., & Nunner-Winker, H. (1992).Children's moral motive strength and temper-amental inhibition reduce their egoistic be-havior in real moral conflicts. Child Develop-ment, 63, 1223-1235.

Crockenberg, S. (1991, April). Conceptual issuesin the study of child compliance, noncompli-ance, and parental control. Paper presentedat the biennial meeting of Society for Re-search in Child Development, Seattle.

Crockenberg, S., & Litman, C. (1990). Autonomyas competence in two-year-olds: Maternalcorrelates of child defiance, compliance, andself-assertion. Developmental Psychology, 26,961-971,

Emde, R. N., Biringen, Z., Clyman, R. B., & Op-penheim, D. (1991). The moral self of infancy:Affective core and procedural knowledge. De-velopmental Review, 11, 251-270.

Emde, R. N., Johnson, W. F., & Easterbrooks, A.(1987). The do's and don'ts of early moral de-velopment: Psychoanalytic tradition and cur-rent research. In J. Kagan & S. Lamb (Eds.),The emergence of morality in young children(pp. 245-276). Chicago: University of Chi-cago Press.

Forehand, R. (1977). Child noncompliance to pa-rental requests: Behavioral analysis and treat-ment. In M. Hersen, R. M. Eisler, & P. M.Miller (Eds.), Progress in behavior modifica-tion (Vol, 5), New York: Academic Press.

Gralinski, H. ]., & Kopp, C. B. (1993). Everydayrules for behavior: Mothers' requests to youngchildren. Developmental Psychology, 29,573-584.

Crusec, J. E., & Goodnow, J, J. (1994). The impactof parental discipline methods on tbe child'sintemalization of values: A reconceptualiza-tion of current points of view. DevelopmentalPsychology, 30, 4-19.

Kagan j . , & Lamb, S. (Eds.). (1987). The emer-gence of morality in young children. Chicago:University of Chicago Press,

Kochanska, G. (1991), Socialization and tempera-ment in the development of guilt and con-science. Child Development, 62, 1379-1392,

Kocbanska, C. (1993), Toward a synthesis of pa-rental socialization and child temperament inearly development of conscience. Child De-velopment, 64, 325-347,

Kochanska, G. (1994). Beyond cognition; Ex-panding the search for the early roots of inter-nalization and conscience. DevelopmentalPsychology, 30, 20-22.

Kochanska. G. (1995). Children's temperament,mothers' discipline, and security of attach-ment: Multiple pathways to emerging inter-nalization. Child Development, 66, 597-615,

Kochanska, G., & Aksan, N. (1995). Mother-childmutually positive affect, the quality of childcompliance to requests and prohibitions, andmaternal control as correlates of early inter-nalization. Child Development, 66, 236-254.

Kochanska. G., Casey, R. J., & Fukumoto, A.(1995). Toddlers' sensitivity to standard viola-tions. Child Development, 66, 643-656.

Kochanska, G., DeVet, K,, Goldman, M., Murray,K., &: Putnam, S. (1994). Maternal reports ofconscience development and temperament inyoung children. Child Development, 65,852-868.

Kohlberg, L. (1969). Stage and sequence: Tbe cog-nitive-developmental approach to socializa-tion. In D, A. Goslin (Ed.), Handbook of so-cialization theory and research (pp.347-480). Chicago: Rand McNally.

Kopp, C. (1982). Antecedents of self-regulation: Adevelopmental perspective. DevelopmentalPsychology, 18, 199-214.

Kuczynski, L, (1992). The concept of compliancein child rearing interaction. In J. M. A. M.Janssens and J. R. M. Gerris (Eds.), Childrearing: Influence on prosocial and moral de-velopment (pp, 125-144). Amsterdam: Swets& Zeitlinger,

Kuczynski, L., & Kochanska, G. (1990). Develop-ment of children's noncompliant strategiesfrom toddlerhood to age five. DevelopmentalPsychology, 26, 398-408.

Lay, K. L,, Waters, E., & Park, K. A. (1989), Mater-nal responsiveness and cbild compliance:The role of mood as mediator. Child Develop-ment, 60, 1405-1411.

Lepper, M. R, (1981). Intrinsic and extrinsic moti-vation in cbildren; Detrimental effects of su-perfluous social controls. In N. A. Collins(Ed.), Minnesota Symposia on Child Psychol-ogy (pp. 155-214), Minneapolis: Universityof Minnesota Press.

Londerville, S., 6c Main, M. (1981). Security ofattachment, compliance, and maternal train-ing methods in the second year of life. Devel-opmental Psychology, 17, 289-299.

Lytton, H, (1980). Parent-child interaction: Thesocialization processes observed in twin andsingleton families. New York; Plenum.

Maccohy, E. E. (1983). Let's not overattribute totbe attribution process: Comments on socialcognition and bebavior. In E, T. Higgins, D.Ruble, & W. Hartup (Eds.), Social cognitionand social development: A sociocultural per-spective (pp. 356-3TO). Gambridge: Gam-bridge University Press,

Maccoby, E. E. (1990). Gender and relationships:A developmental account, American Psychol-ogist, 45, 513-520,

Maccoby, E. E. (1992). The role of parents in tbesocialization of children: An historical over-

Kochanska, Aksan, and Koenig 1769

view. Developmental Psychology, 28, 1006--1017.

Maceoby, E, E., & Martin, J. (1983). Socializationin the context of the family: Parent-child in-teraction. In E. M. Hetherington (Ed.), P. H.Mussen (Series Ed.), Handbook of child psy-chology: Vol. 4. Socialization, personality,and social development (pp, 1-101). NewYork: Wiley.

Parpal, M., & Maccoby, E. E. (1985), Maternal re-sponsiveness and subsequent cbild compli-ance. Child Development, 56, 1326-1334.

Perry, D. G., & Perry, L. C. (1983). Soeial learning,causal attributions, and moral intemalization.In J. Bisanz, G. L. Bisanz, & R. Kail (Eds,),C, J. Brainerd (Series Ed.), Learning in chil-dren: Progress in cognitive development re-

search (pp. 105-136). New York: Springer-Verlag.

Piaget, ]. (1932). The moral judgment ofthe child.London: Kegan Paul.

Schneider-Rosen, K., & Wenz-Cross, M. (1990).Patterns of compliance from eighteen to thirtymonths of age. Child Development, 61,104-122,

Stayton, D. J., Hogan, R., & Ainsworth, M. D. S,(1971). Infant obedience and maternal behav-ior: Tbe origins of socialization reconsidered.Child Development, 42, 1057-1069.

Vaugbn, B. E., Kopp, C, & Krakow, J. B. (1984).Emergence and consolidation of self-controlfrom eighteen to thirty months of age: Norma-tive trends and individual differences. ChildDevelopment, 55, 990-1004,