Reviews Sibling Relationships in Early Childhood Judy Dunn University of Cambridge DUNN, JUDY. Sibling Relationships in Early Childhood. CHILD DEVELOPMENT, 1983, 54, 787- 811. This review considers (1) recent evidence on the nature of sibling interaction from observa- tional studies of preschool children and (2) the developmental implications of this evidence. Sibling interaction is discussed in terms of "reciprocal" and "complementary" interaction. Sib- ling influence is most plausibly associated with the reciprocal features of the relationship and with sociocognitive development. "Sibling status" variables, the focus of previous research, are not consistently related to the reciprocal but to the complementary features, which are probably of less developmental significance. Research into the origins of the marked individual differences between siblings must take account of the mutual influence of parental and sibling relationships, not solely sibling status variables. Children grow up within a network of relationships���with parents, grandparents, friends, and, for 80% of children in the United States and Britain, siblings. How does this experience of being a sibling affect their development? It is clear that for first children the arrival of a sibling affects their relationships with their parents dramatically (Dunn & Kendrick, 1980 Taylor & Kogan, 1973 Trause, Voos, Rudd, Klaus, Kennell, & Boslett, 1981). But how far does the re- lationship between the children influence their development more directly? Two themes in recent developmental psychological research suggest that the re- lationship between young siblings may be of developmental importance. First, behavior geneticists emphasize that siblings differ markedly from one another in personality, intellectual development, and psycho- pathology, although they share not only more genetic material than unrelated chil- dren but many aspects of the family en- vironment (Rowe & Plomin, 1981 Scarr & Grajek, 1982). To explain these striking dif- ferences between siblings, the direct in- fluence of siblings upon one another must be taken into account, as well as differential treatment by parents, and selection of par- ticular and differing aspects of the family environment by the different children ("ecological niche-picking") (Scarr & Crajek, 1982 see also Jaspars & de Leeuw, 1980). The suggestion is that siblings may, by their behavior toward each other, create very different environments for one another within the family. Second, it is now widely acknowledged that regarding mother and child as a dyad isolated from the other relationships within the family is extremely misleading. The significance and complexity of mutual in- fluences within the family are now clearly recognized (Bronfenbrenner, 1979 Clarke- Stewart, 1978 Lamb, 1976 Lewis & Rosenblum, 1979 Parke, Power, & Gottman, 1979). On commonsense grounds it seems highly likely that the relationship between young siblings will affect and be affected by the children's relationship with their par- ents. Psychologists from widely differing backgrounds have in fact argued that early sibling relations may be developmentally important. Clinicians, for instance, have seen rivalry between siblings as a major in- fluence on personality (Adler, 1928, 1959 Levy, 1934, 1937 Winnicott, 1977). But di- rect studies of sibling interaction have until recently been comparatively rare, and this represents a major gap in our understanding of family relationships. There is an extensive literature on the relation between family This review was written while the author was on the staff of the Medical Research Council. I am grateful to Robert Hinde for his helpful comments on the manuscript. Requests for reprints should be sent to Judy Dunn, Medical Research Council Unit on the Development and Integra- tion of Behaviour, University of Cambridge, Madingley, Cambridge CB3 8AA, England. [Child Development, 1983,54, 787-811. �� 1983 by the Society for Research in Child Development, Inc. All rights reserved. 0009-3920/83/5404-0001$01.00]

788 Child Development structure variables (birth order, sex, and spacing of siblings) and intelligence, achievement, and personality, which suggests that such "sibling status" variables do influence an individual's interests, pref- erences, style of thinking, self-esteem, con- formity, and eminence and achievement as an adult. (See, e.g.. Brim, 1958 Cicirelli, 1973,1975,1976a, 1976b, 1977,1978 Harris, 1964 Jones, 1931 Koch, 1955a, 1955b, 1956, 1960 Marjoribanks, Walberg, & Bergen, 1975 Sutton-Smith & Rosenberg, 1970 Zajonc & Markus, 1975. For a comprehen- sive review of the findings on the association between sibling constellation variables and intelligence, achievement, and personality, see Wagner, Schubert, & Schubert, 1979.) However, Scarr and Grajek (1982) argue for- cibly that the constructs of sibling spacing, birth order, and sex are simply inadequate to explain the magnitude of the differences between siblings. Moreover, the birth-order literature in particular suffers from methodological shortcomings in a critical review. Schooler (1972) argued that most birth-order associations are in fact spurious. In several studies the variables of age, birth order, and birth interval are confounded. Most importantly, such research cannot take us very far toward understanding the processes involved in the development of differences between siblings or the extent of sibling influences on development. It is clear that the birth-order effects could be the result of differential parent treatment or of more direct influences between the siblings, or indeed that both might be important. Sutton-Smith and Rosenberg (1970) argued that individual differences in aggression, power tactics, sex role preferences, and interests in later-born children are most plausibly attributed to sibling influences���to processes of identification and modeling on the older sibling���whereas the greater achievement, affiliative, and conforming be- havior of only and first-bom children are at- tributable to the special relationship these children have with their parents. However, they pointed out in 1970 that there was an imbalance in the research literature on the early development of these differences studies of children over 5 years had examined sibling influences (chiefly indi- rectly) but not differential treatment of first- and later-born children, while studies of younger children had concentrated on parent-child interaction but not the behavior of siblings together. And Scarr and Grajek (1982) argue that we lack a theory to guide or interpret the "sibling constellation" re- search. They point out that theories of role differentiation such as those of Bossard and Boll (1956) or Parsons and Bales (1955) are difficult to apply to the complex data of the sibling constellation studies in a way that is useful for prediction. More often the notions of identification, deidentifieation, and mod- eling on the sibling are used, post hoc, to explain whatever particular patterns of re- sults the studies reveal. To begin to describe the processes that lead to the differences between siblings growing up within the same family and to assess the relative importance of sibling and parental influence on development, we need as a flrst step to understand the nature of the relationship between young siblings and how it influences and is influenced by the other relationships a child forms. Accord- ingly, the first issue considered in this re- view concerns what we at present know about the interaction and relationship be- tween siblings in early childhood. The evi- dence on the nature of the relationship be- tween siblings is considered within a framework that focuses on the distinction that Piaget (1965) and Sullivan (1953) origi- nally drew between child-child and child- adult interaction, the distinction between di- rect reciprocity and complementarity in interaction (see also Harre, 1974 Hinde, 1979 Youniss, 1980). Sullivan and Piaget ar- gued that children's relations with other children play an important part in the de- velopment of sensitivity and understanding of the self and others. The reciprocal nature of peer relationships was seen as of central importance in these developments, con- trasted with the complementarity of the parent-child relationship, in which both child and adult inevitably face difficulties in understanding the reasoning and perspec- tive of the other. In the context of this argu- ment, the relationship between siblings is of particular interest the familiarity and in- timacy of the children, the extent to which they recognize and share each other's inter- ests, and the emotional intensity of their re- lationship all suggest that the relationship between siblings will include elements of the "direct reciprocity" seen as the key fea- ture of the peer relationship. Yet the differ- ences in age between siblings implies that there may well also be elements on the one hand of the "complementarity" of the parent-child relationship (Hinde's [1979]

term denoting interaction in which behavior of 6ach partner differs from but comple- ments that of the other). Drawing a distinction between re- ciprocal and complementary interaction has both advantages and disadvantages. Any such classification is of course artificial it represents an attempt to draw boundaries where no clear discontinuities in fact exist. The value of the classification in thinking about sibling interaction lies in the attention it draws to how sibling interaction differs distinctively from that between child and adult or child and peer, and the perspective it gives on the nature of sibling influence. The distinction between reciprocity and complementarity in interaction focuses on the processes of interaction between indi- viduals, but it also carries strong im- plications about the consequences of a re- lationship characterized by one or the other type of interaction. The second issue considered in the re- view concerns the developmental im- plications of this evidence on the nature of the early sibling relationship. Does the work on siblings provide support for the view that children are influenced by their siblings��� either in the quality of the relationship that develops between them or their behavior and personality more generally? Does it il- luminate the processes involved in the de- veloping relationship between the children and in the patterns of influence within the family? Gan it help us to explain the differ- ences between siblings to which behavioral geneticists have drawn attention? It is argued that sibling influence is most plausibly demonstrated in the reciprocal features of the interaction and that the "sibling status" variables of age, birth inter- val, and sex do not appear to be consistently related to these reciprocal features. Rather, such family structure variables relate more clearly to the complementary features of the sibling interaction, and these are develop- mentally of less significance, at least for Western families. This may explain in part the failure of such constructs to account for sibling differences (Scarr & Grajek, 1982). To understand the origins of the very marked individual differences in sibling be- havior and to assess sibling influence in de- 'C'elopment, we must examine differences in the quality of parental relationships with each child, in the mutual influence of these family relationships, and in the tempera- ment of the children and not concentrate Judy Dunn 789 solely on family structure variables. Finally, it is argued that the close relations between reciprocal interaction and sociocognitive development means that the sibling re- lationship provides a particularly illuminat- ing perspective on the development of social understanding. The review focuses primarily on obser- vational studies of preschool aged children in unstructured or semistructured set- tings���either at home or in laboratory play- rooms diary studies and anthropologi- cal references are also included. Work on older children, such as the extensive series of experimental studies of siblings carried out by Gicirelli (1967, 1972, 1973, 1975, 1976a, 1976b, 1977, 1978), the interview studies by Koch (1955a, 1955b, 1956, 1960), and Patterson's studies on aggression (Pat- terson, 1975 Patterson & Gobb, 1971) will be referred to only briefly. A number of limitations in the nature and range of studies which provide the ob- servational material (shown in Table 1) should be noted. First, all of the studies have concentrated on sibling dyads interaction between more than two siblings has not been studied. Second, with the exception of the studies by Harkness (1977) and the Whitings and Pope-Edwards (Whiting & Pope-Edwards, 1977 Whiting & Whiting, 1975), the focus has been on the interaction between first- and second-born children. Third, in all of the studies the age gap be- tween the siblings is confounded with the age of one of the siblings even in the Gana- dian study of Abramovitch, Gorter, and Pepler (1980), which was designed to com- pare two different age gaps, the age of the older sibling differed in the two groups. Fourth, no studies focus on the behavior of infants under 6 months with their siblings. And finally, the range of cultures studied is comparatively narrow. With the notable ex- ceptions of Whiting and Whiting's cross- cultural material and the language study of Harkness (1977), systematic information comes from studies of U.S., Ganadian, and British families. It is plainly unwise to generalize from these studies to the character of sibling re- lationships in large families or in different national cultures. Nevertheless, the evi- dence from these studies does raise general questions about the developmental process- es involved in the relationship between young siblings, and these questions are con- sidered in the second section of the review.

on u. 5 3 9 5 I ���5S U u. u O 3 o S E 0. o S i g " be A l l l l l l l l il e t o ul 1 stat e fee I 1/S 3 "(3 fe m [ina l itu c Jpat-5t c u c �� M .2 0. .2 I I I ^ 9 D ffl X lil l si a. a 2 C g ��� ? A I ii s a HI i n .5 790

td rJ 3 9 5J 3 C o O z S z ���50. c o Is z ��. Z �� I Z �� .1 Z �� Z �� �� z �� z Ilillll O 0) 2 .�� 2 S ���o 3 5 ���= i^ o A ��i '.c tyj 2Q f J= �����������= -O .5 = CA S 5 J 5 U SO e r t o ���V 0 nger : 0 , 'Ul i 1A "5 0 c nge r 3 j i eek s E1st st-bo : irls oy s tsCX Z �� J I . 2 3 I S E 6c ^ 2 E S �� S 5 -i O 2 E 2 o Is ���3 2 c^ a II 791