Children's social identities



This paper provides a brief overview of recent developmental research on themes related to children's social identities. Initially, consideration is given to the capacity for social categorization, following which attention is given to children's developing conceptions of social identities, their identification with social groups, and the consequences of identification upon various phenomena such as ingroup favouritism and well-being. Finally, some personal thoughts on the wider political implications of this research are offered. Copyright © 2011 John Wiley & Sons, Ltd.

Ten years ago, just prior to becoming an editor of ICD, I edited a Special Issue of the journal entitled Self and Identity. In the editorial I referred to ‘the possibilities for significant and exciting work’ on the topic of children's social identities. Up until that time, developmental research on the self had focused largely upon the personal self; that is, the self defined by idiosyncratic features such as personality traits (e.g. Damon & Hart, 1988; Livesley & Bromley, 1973). By contrast, rather little research had been directed at the social self; that is, the self defined by our group memberships, such as ethnicity, religion, gender, and subcultural groups. However, during the past decade a great deal of significant and exciting work on the social self has indeed been conducted. (For summaries, see for example Barrett, 2007; Barrett & Buchanan-Barrow, 2006; Bennett & Sani, 2004; Dunham & Degner, 2010; Levy & Killen, 2008).

Why should we be interested in social identities? A key insight of the social identity perspective (Tajfel, 1972, 1981; Turner, 1975, 1982), which has guided much research in this area, is that groups become internalized and, as such, constitutive of our sense of self. To the extent that we identify ourselves, for example, as Spanish, an Inter Milan supporter, a socialist, a psychologist, etc., particular social norms become relevant to action and cognition when those identities are salient in specific contexts. And as we shall see, subjective identification with groups has been shown to have a variety of important consequences, both cognitive and behavioural. Even more basically, as Haslam (2004, p. 17) has noted, 'the capacity to think in terms of ‘we’ and ‘us’, and not just ‘I’ and ‘me’, is central to the capacity for meaningful group behaviour.' More generally, the importance of social identity is reflected in the fact that it ‘creates and defines the individual's place in society’ (Tajfel & Turner, 1979, pp. 40–41). Given such considerations, it is perhaps surprising that the developmental study of social identity has come to prominence so recently.

In what follows, I will outline key issues in the field and will provide a sketch of major research themes. Initially attention is given to the fundamental capacity for social categorization, following which consideration will be directed at children's developing conceptions of social identities, their identification with social groups, and the consequences of identification upon various phenomena such as ingroup favouritism and well-being. Finally, I will offer some personal thoughts on the wider political implications of this research for children's place within society.


Human groups, such as gender, ethnic and religious groups, are fundamental categories and constitute an important part of our social world. When do children start to construe others in categorical terms? For some social categories, such as gender, race and age, it is clear that this happens very early indeed. For example, using habituation techniques, Katz and Kofkin (1997) found that by 6 months infants discriminate between individuals at the level of racial categories. (Intriguingly, African-American infants evinced greater sensitivity to racial cues than did European-American infants.) With respect to gender, too, Quinn et al. (2002) established that 3-month-old infants distinguish between males and females, even in the absence of typical hair and clothing cues. Infants also show preferences for faces of peers over those of adults (Bahrick, Netto, & Hernandez-Reif, 1998; Sanefuji, Ohgami, & Hashiya, 2006), indicating that age too is important as a basis for categorization. Recently, Kinzler, Shutts and Correll (2010) have found that a fourth form of categorization is apparent in infancy: native-language speakers versus foreign-language speakers. Although we have known for some time of newborns' preference for their mother tongue over other languages (Mehler et al., 1988; Nazzi, Bertoncini, & Mehler, 1998), recent work has revealed that pre-verbal infants even prefer to interact with and accept toys from speakers of their native language (Kinzler, Dupoux, & Spelke, 2007; Shutts, Kinzler, McKee, & Spelke, 2009).

These are impressive findings that demonstrate an early capacity for various forms of social categorization. Nonetheless it is important to note that many social categories (such as nationality) lack the perceptual discriminability that is associated with gender and ethnicity, and as such, being linguistically based constructions, necessarily lie beyond infants' comprehension. Furthermore, although the capacity for social categorization is a necessary condition for the development of children's social identities, it is a long way from being a sufficient condition. Thus, children must also be able reflexively to conceive of the self in group categorical terms, for example, ‘I am a boy’. Katz and Kofkin (1997) have shown that by 3 years, a majority of European-American children (77%) accurately label themselves in racial terms (however only 32% of African-American appear to do so). Gender self-labelling was found to occur somewhat earlier, at around 30 months. Accurate identification of nationality tends to come somewhat later, typically at around 5 years (Barrett, 2007; Lambert & Klineberg, 1967).

While even young children may be able to label themselves in terms of some of the major social categories, it is clear that conceptions of social identities undergo substantial change during childhood and beyond. Quintana's (1998) work suggests that from 3 to 6 years, children's conceptions of social identities are based merely upon physical features such as skin colour, clothing, hair length, etc. Following this, between 6 and 10 years comes an appreciation of group-related practices such as food and activity preferences, etc. Subsequent to this, between 10 and 14 years, children acknowledge wider implications of identities such as status differences between groups and widely held stereotypes about them. In our own work, Fabio Sani and I have identified a similar sort of progression in children's conceptions of social identities (Sani & Bennett, 2004). The youngest children we investigated, 5-year-olds, conceived of identities primarily in behavioural terms (e.g. ‘Boys play football’). Such a conception was typical through to middle childhood. Only by late-childhood and adolescence were group identities conceived also with reference to beliefs (‘British people believe in fair play’). Despite such sophistication by late childhood, many further developments are likely to remain. For example, recent research by Svirydzenka, Sani, and Bennett (2010) has shown that 10-year-old children's conceptions of groups' ‘groupness’ (or entitativity) emphasize perceptually salient features such as the level of interaction among group members, whereas adults emphasize more abstract features like the subjective importance of the group to its members.

Interestingly, and as predicted by self-categorization theory (Oakes, Haslam, & Turner, 1994), the particular content of children's beliefs about specific groups has been found to be subject to contextual variability. Briefly, self-categorization theory predicts that, within a given context, the attributes that are highlighted as typifying a group are those which maximally differentiate it from another group (or groups) under consideration. Thus, it has been found, for example, that children's stereotyping of boys differs depending upon whether they have just been thinking of opposite sex children (girls) or same-sex adults (men) (Sani & Bennett, 2001; Sani, Bennett, Mullally, & MacPherson, 2003). Boys were more likely to be characterized as strong, brave and big in the former context, but as talkative in the latter. Similarly, girls were viewed as clever and hardworking in the context of boys, but as greedy and loud in the context of adult women. Thus, inspired by recent social psychological theorizing, these studies challenge the traditional orthodoxy that group stereotypes are rigid and inflexible constructs.

At least since Martin and Halverson's (1981) landmark paper on children's schema-based processing of social information, much research has examined the impact of social category schemas (primarily gender) upon attention, memory and preferences (Martin, 2000; Ruble & Martin, 1998). This body of research has shown that children's social category schemas influence their attention to, and memory for, information about target persons. It is now clear that categorical distinctions have important cognitive consequences with respect to representations of persons and behaviour. Thus, photographs of children engaged in gender-consistent activities are better remembered than gender-inconsistent ones (Martin & Halverson, 1981). Moreover, memory for gender-inconsistent depictions is often distorted, such that children's recollections often involve changing the sex of target (Carter & Levy, 1988; Signorella, Bigler, & Liben, 1997; Stangor & Ruble, 1987). Even for information that is neutral with respect to the categories under consideration (e.g. reference to food preferences), it is clear that categories like gender and race play an important role in children's initial encoding and recall of others' behaviour (i.e. so that children can recall that it was a boy who expressed a preference for food x, even they cannot say which boy; Bennett & Sani, 2003).

Unsurprisingly, given the foregoing, a substantial body of evidence attests to young children's willingness to categorize themselves in terms of social categories. However, as we have discussed at length elsewhere (see Bennett & Sani, 2008, 2011), for the most part this research has demonstrated only a capacity for self-categorization. It does not reveal whether children subjectively identify with groups. Ruble et al. (2004) capture this distinction well when they make the point that children's self-categorizations may serve as no more than ‘empty labels.’ Categorizing oneself in terms of a social category is importantly different from identifying with a social group, particularly since it is subjective identification with groups that mobilizes specific forms of group-related cognition and action.

In order to examine children's subjective identification with groups, Fabio Sani and I (Bennett & Sani, 2008b) have drawn upon techniques used in cognitive social psychology. For example, in studies of ‘self-stereotyping’, we have shown that if gender is made salient in a given context, children judge themselves as more similar to same-sex classmates than when those same children are considered in a gender-neutral context. Similarly, in such contexts children will more readily ascribe gender-related attributes to themselves than in ‘control’ contexts—that is, they spontaneously incorporate gender-relevant attributes into their self-concepts when that identity is salient. Thus, in stereotyping themselves as group members, our findings show that even by 5 years children subjectively identify with social groups. Moreover, our findings also confirm a basic claim of the social identity approach that self-conceptions are contextually variable. These results bear striking parallels to Banerjee and Lintern's (2000) finding that 4- to 6-year-old boys' self-descriptions were significantly more gender-stereotypical when before a group of same-sex peers than when alone. Thus, context plays a significant role in self-conception even in young children. (See too Turner and Brown, 2007.)

Convergent evidence for young children's subjective identification with groups has come from studies using other methodologies. For example, drawing upon research in social psychology which has revealed that adults' representations of self and in-groups show a substantial degree of overlap (e.g. Aron & McLaughlin-Volpe, 2001; Mashek, Aron, & Boncimino, 2003; Smith, Coats, & Walling, 1999; Smith & Henry, 1996), we have provided evidence that in a memory task even young children confuse characteristics of the ingroup (but not outgroup) with the characteristics of the individual self (Sani & Bennett, 2009). Elsewhere we have shown that children, like adults, recall more about information that is encoded with reference to the self than when the very same sort of information is encoded with reference to other entities (e.g. ‘does the word [x] describe you’ versus ‘does the word [x] mean the same as [y]?’). Extending the self-reference effect to the level of group identities, we have shown similarly that even 5-year-olds recall more about information that is encoded with reference to their ingroups than other entities (e.g. ‘Does the word [x] describe your family’ versus controls: ‘Does the word [x] mean the same as [y]?’; ‘Does the word [x] describe dogs?’). This study shows that for three ingroups (family, gender and age) children process information relevant to self and ingroups similarly (Bennett & Sani, 2008c). Even among young children, then, ingroups are a key aspect of identity.

It is important to note that although our work on subjective identification has focused primarily on the cognitive aspects of social identity, affective aspects too appear to be in place early on. Warin (2000) for example found 4- to 6-year-old children to be highly resistant to the request to dress in opposite sex clothes, and indicated their distaste at the request. Indeed, 43% of children flatly refused to comply with the request, and of those who did comply, most grimaced and expressed unhappiness at complying. Such findings suggest a very strong subjective commitment to the gender ingroup. Similar affective commitments have been found with respect to other forms of identity (e.g. ethnic identity, Bar-Tal, 1996).

Such research has shown that young children subjectively identify with social groups from an early age. Surprisingly, however, our work on this matter has not revealed age differences over the period 5–11 years. This is almost certainly the result of looking at very particular aspects of social identity, not because developments do not take place. Indeed, research by Abrams and his colleagues demonstrates significant developments in social identity processes following the age of 5 years. Abrams, Rutland, Pelletier, and Ferrell's (2009) work has examined children's ‘group nous’; that is, their knowledge of intragroup and intergroup functioning. Particularly important is children's knowledge of loyalty norms, which has been shown to be affected by various factors, including social perspective-taking and group experience. Thus, whereas younger children show unconditional preference for ingroup members, older children are sensitive to others' adherence to group norms, so that for example an outgroup member who expresses support for one's ingroup is preferred to an ingroup deviant (or ‘Black Sheep’) who does not preferentially favour the ingroup over the outgroup. Abrams' research is extremely valuable in that it seeks to address general issues concerning the development of social identity, but does so in the context of a broader range of cognitive and social variables than is typical in the field. In consequence, the explanatory power of this work is impressive. (See too Nesdale, 2004.)

Other researchers investigating social identity development have adopted a rather different strategy from that implicit in much of the research so far described. Rather than tackling relatively circumscribed hypotheses, with small numbers of measures, some researchers have focused in depth upon the development of specific social identities and have used a broad range of measures. Barrett (2004, 2007), for example, has sought to chart the development of national identity. Importantly, this research does not yield a neat developmental picture and instead reveals immense cross-national variability in national identification, and knowledge, feelings, beliefs, etc. about the nation. Verkuyten (2004), too, summarizing a substantial and complex body of findings concerning ethnic identity, notes the difficulties associated with the attempt to specify a clear developmental trajectory. Theoretically, the major challenge that such research poses to the field is the need to look not merely at psychological factors such as cognitive-developmental status, but also at the impact of social factors, such as group norms and values, status differences between groups, and situational conditions.

Important too is the need to look beyond the much-studied categories of gender, ethnicity and nationality. For example, very little research has looked at children's conceptions of, and identification with, age-based identities such as infancy, childhood, teenagerhood, and the various phases of adulthood. It would be extremely interesting to address children's conceptions of these identities, particularly in the context of action theory approaches (Brandstädter & Lerner, 1999) that conceive of identity, at least in part, as a self-directed project: thus, rather than viewing identity development in terms of particular causes, action theory approaches draw attention to development as a reflexive process in which children, increasingly with age, play a role in their own development. In the context of age-based identities this seems particularly relevant since there is plenty of anecdotal evidence that children are highly motivated to achieve the statuses associated with age identities that lie ahead of them. Similarly, more attention should be given to the identities that children and especially adolescents choose (such as sub-cultural identities) and not just those, like gender and ethnicity, which are imposed upon them. During adolescence in particular, chosen identities become especially important and reflect increased concerns with social face and with peer group acceptance. Where earlier in childhood ethnic, religious, gender and other ascribed identities are likely to have been scaffolded by adults, it seems reasonable to speculate that the sorts of processes involved in the construction of achieved identities are those addressed by action theory approaches (Brandstädter & Lerner, 1999).

Other significant avenues for inquiry lie in the fact that social identities have consequences. Perhaps the most widely demonstrated consequence, which is predicted by social identity theory, is ingroup favouritism. Many studies show that the extent of ingroup favouritism is greater in children with higher levels of ingroup identification (Bennett, Lyons, Sani, & Barrett, 1998; Pfeifer et al., 2007; Verkuyten & Thijs, 2001). While much research shows that children favour the ingroup over outgroups, they do not routinely derogate outgroups. Nonetheless, clearly there are circumstances under which children's ingroup identification can give rise to prejudice (see Nesdale, 2004; Rutland, Killen, & Abrams, 2010).

Identification with ingroups can have consequences not just for intergroup cognitions and behaviour but also for individual well-being. For example, Egan and Perry (2001) found that children's perceptions of their gender typicality and gender contentedness related positively to self-worth even when self-efficacy for gender-typed activities had been controlled. Mental health too appears to be impacted by social identification: in a 1-year longitudinal study of 12- to 14-year-old African-American children, Mandara, Gaylord-Harden, Richards, and Ragsdale (2009) found that increases in racial identification were related to decreases in depressive symptoms even when self-esteem had been controlled. Kiang, Yip, Gonzales-Backen, Witkow, and Fuligni (2006) report similar findings for young Mexican-Americans and Chinese-Americans. Such work emphasizes the importance of children's identification with social groups. Many possibilities remain for future research. For example, my guess is that a rich seam of investigation could be revealed by looking at identification with a group of special significance to children: the family. Sani (2011) reports data showing that, among adults, higher levels of family identification predict greater satisfaction with life, happiness, psychological well-being, physical mobility, and physical health. Key here is that—in adults at least—group identification promotes health by facilitating social support, among other factors. It remains for future research to explore this matter developmentally.

In sum, then, although a prodigious amount of valuable work has been conducted in recent years, much remains to be done. As I noted above, one of the major challenges is to move beyond the widespread practice of delineating ‘the’ developmental course of some capacity or other, to detailing a more nuanced and complex account of development that acknowledges the role of social factors, such as group norms and values, status differences between groups, and also more transient contextual factors that trigger particular social identities. Moreover, as I have indicated above, looking at the effects of group identities is likely to be a fruitful line of inquiry. Relatedly, we should extend cognitive analyses by investigating what the sociologist Bourdieu (1986) referred to as the capital that is associated with social identities. That is, identities should be seen as more than cognitive constructions; they should also be seen as associated with differential resources (e.g. material, social, symbolic, etc.) that can play an important role in shaping individuals' opportunities, obligations, actions, etc. To take an obvious example, in the majority of the world's cultures, to be male is to be afforded many resources (sometimes enshrined in law) that impact materially upon one's life. Not only are there differences between social groups in this respect, clearly there are differences within social groups. For example, as Abrams et al. (2009) note, even among those who share a social group identity there are significant variations in ‘group nous’, which is clearly one of many forms of capital that can be deployed with significant effects. In short, group identities are ‘for real’ and associated with resource-related properties that impact upon individuals' experience of the social world. This potentially represents an important theme for future research.


As will be clear by now, developmental research in the area of social identity is flourishing. This, of course, is to be welcomed. But, looked at in a broader, societal context, there's an uncomfortable irony here. As I have noted elsewhere, 'the position of children in contemporary Western societies is distinctly odd in socio-historical terms: seldom have children been quite as literally ‘useless’, confined to relatively passive roles in the narrowly domestic arena, and largely excluded from participation in broader social and economic arenas' (Bennett, 2006, p. 342). Despite children's participation rights being enshrined in the United Nations Convention on the Rights of the Child, their participation in Western societies is very modest indeed. Indeed, some have suggested that the ‘insularization’ of children is now endemic (Zeiher, 2000). Thus, what makes me uncomfortable is that the rise of research demonstrating a capacity for collective life has coincided with a period in history during which children's experience of collective life is at an all-time nadir.

For some time, sociologists have noted important changes in modern childhood, such as a considerable reduction in children's autonomy to organize their own free time. Nearly 20 years ago, for example, Ennew (1994) noted that 'the sociability of the street has given way to the privacy of the family, the ‘filling up’ of children's time, and the scheduling of their lives' (p.143; see too Prout, 2005). Although these changes are due in part to (largely misplaced) concerns about child safety (Furedi, 2008), they have the consequence of greatly limiting children's independent mobility and exposure to collective life. In recognition of the costs both to children and society of this marginalization of children, some have proposed, radically, that we should re-think citizenship to include children (Roche, 1999). Although we are far from being a society that routinely recognizes children's and adolescents' rights to societal participation (such as contributing to decision-making about urban planning, transport provision, leisure, healthcare, etc.), the case for children's citizenship, even if partial, has been made articulately and powerfully (Cockburn, 1998; De Winter, 1997; Held, 1991; King, 1997). Nonetheless, there is clearly a long way to go—for example, under UK legislation children lack the right even to form a pupil council! (More remarkable still is that the US is the only country in the world other than Somalia not to ratify the United Nations Convention on the Rights of the Child.)

Some work has been conducted in both the US and Europe to examine the impact of citizenship programmes on young people (e.g. Holden & Clough, 1998; Yates & Youniss, 1999). Participation in such programmes appears to have a variety of very favourable consequences. For example, not only do participants come to understand more about societal issues, such as inequality, homelessness, etc., but also they show greater moral engagement with such issues, which in turn has beneficial motivational consequences—typically, a desire to contribute to the well-being of others and the wider community. Given the research discussed earlier in this paper, an important empirical question is whether the benefits of such programmes are due, at least in part, to changes in children's self-conception, such that the boundaries of the self are expanded to include those sharing category membership at the level of one's community and society. If this were to be the case then the implications for children's nascent sense of citizenship would be significant indeed.

In my view the exciting prospects for academics working in the broad field of social identity development are broadly twofold. On the one hand there is the theoretical challenge of going beyond traditional approaches, which have sought to chart developmental progress within a domain, typically with reference to cognitive-developmental variables. Specifically, if we are to capture the subtlety of social identity development we need to formalize our understanding of the relationship between cognition, context and action. Clearly, this is no easy task, but greater collaborations between developmental and social psychologists will be important here. On the other hand it is important that we be alert to the real-world implications of such research. In particular, we should embrace the political challenge of making an empirically founded case for children's inclusion and participation in the lives of their communities. Psychologists have frequently fallen short in terms of realizing the societal implications of their work, but in these financially straitened times, careful consideration of the wider implications of work in this domain can potentially serve both narrow grant-getting ambitions and profoundly important social goals bearing upon human welfare.