1. Introduction

Universally, human mothers invest more than fathers in offspring (for reviews, see Brown, 1991, Hewlett, 1992) as is typical across mammalian species Krasnegor & Bridges, 1990, Trivers, 1972. Little is known however about sex differences in investment in unrelated individuals (Hill, 2002). This gap likely exists because of the complexity involved in defining and measuring investment between unrelated individuals.

In studies of human beings' closest living genetic relatives, chimpanzees, grooming, proximity, and food sharing are used to measure investment. Results indicate that unrelated adult males are more strongly bonded than unrelated adult females in common chimpanzees, Pan troglodytes, whereas the reverse is true for bonobos, Pan paniscusde Waal, 2001, White & Chapman, 1995, Wrangham, 1986. In studies of humans in traditional societies with few resources, food sharing commonly serves as a measure of investment (for a review, see Gurven, in press). In societies where food is relatively plentiful however, other measures are required.

The current study was designed to measure sex differences in investment between unrelated same-sex children using durability of bonds as the measure of investment. As a measure of investment, durability permits assessment over relatively long time intervals and is less influenced than proximity or food sharing by immediate external forces, such as familial or educational constraints. Whereas long distance relationships between adults undermine the validity of durability as a measure of continued investment (Rodseth & Wrangham, in press), durability is useful as a measure for children whose relationships with unrelated individuals are predominantly local.

Several lines of evidence indicate that beginning in middle childhood in numerous and diverse cultures, males invest more than females in unrelated same-sex individuals. Firstly, boys exhibit greater enjoyment and spend more time with same-sex peers than girls do (Benenson et al., 1998, Bryant, 1985, Whiting & Edwards, 1988; for reviews, see Maccoby, 1998, Schlegel & Barry, 1991). Secondly, boys interact in larger groups of same-sex peers than girls who prefer to form a close bond with only one peer at a time (for reviews, see Belle, 1989, Cairns et al., 1998, Schlegel & Barry, 1991), a pattern that continues into adulthood Baumeister & Sommers, 1997, Brown, 1991, Gabriel & Gardner, 1999, Seeley et al., 2003, Tiger, 1969. The larger number of same-sex partners with whom males interact suggests that the cost–benefit ratio of associating with others of the same sex is lower for males than females (Wrangham, 2000).

Finally, two prior studies with preadolescents and adolescents demonstrate that the durability of close same-sex friendships is greater for males than females (Benenson & Christakos, 2003, Kon & Losenkov, 1978), despite females' greater preference for this form of relationship (Belle, 1989). Analyzing this form of relationship thus constitutes a conservative test of the hypothesis that males' relationships with unrelated same-sex peers are more durable than those of females.

Nevertheless, these latter studies suffer from several limitations. First, they were based on self-reports. Cross-cultural evidence indicates that females are more likely than males to provide negative self-evaluations (for a review, see Kling, Hyde, Showers, & Bushwell, 1999), which may have contributed to the sex difference. Second, both studies included only preadolescents and adolescents. During this period, sex differences in rates of maturation translate into females' confronting the challenges introduced by sexual maturation before males do (Tanner, 1962). Consequently, competition between same-sex individuals for mating partners occurs earlier for females than males (Campbell, 1995), which may have weakened females' friendships. Third, the prior studies ignored the availability of resources (Kon & Losenkov, 1978) or included only individuals with few resources (Benenson & Christakos, 2003). Research with chimpanzees indicates that females invest more heavily in same-sex relationships when resources are plentiful (Boesch & Boesch-Achermann, 2000, de Waal, 1982; see Wrangham, 2000, for a theoretical discussion of feeding competition in higher primates and its influence on female bonding). Belle (1987) similarly finds that human females with few resources experience difficulties sustaining relationships with unrelated females. Belle concludes that economically stressed mothers must invest in offspring and cannot afford to share resources with unrelated females.

In order to overcome these limitations, in this study external observers assessed the durability of same-sex friendships in children who had not yet entered puberty. In addition, participants were selected because they came from economically advantaged backgrounds to ensure that they had access to plentiful resources. The goal was to test whether under these conditions relationships between males still would be more durable than those between females. As a second goal, the dynamics of conflicts between current close same-sex friends were compared for males and females. If sex differences were found in the number or quality of conflicts with same-sex friends, these could illuminate the proximate mechanisms that produce sex differences in investment in unrelated same-sex individuals.

2. Method

2.1. Participants

Seventeen classes from grades 1 to 6 in a private school in Montreal, Canada, participated in the study. On average, there were significantly more boys (M=14.65, S.D.=1.77, range 10–17) than girls (M=11.94, S.D.=1.39, range 10–15), t(32)=4.96, p<.001, in a class. Given that class sizes were large however, each sex was believed to have had ample opportunity to select compatible same-sex friends. Within each class, three to four children of each sex were interviewed about the friendships in their classes following the procedure of Lagerspetz, Bjorkqvist, and Peltonen (1988). Participants were chosen randomly. Consent forms were sent home with students in a class until at least three children of each sex had obtained parental consent to participate and were willing to be interviewed. Participants included 107 children (54 boys and 53 girls) ranging in age from 6 to 11 years. Ages for each sex and grade level are presented in Table 1. Over 95% of children who participated were Caucasian and from upper–middle class socioeconomic class backgrounds.

Table 1. Mean ages of children

	Grade level	Females			Males
		M	S.D.	n	M	S.D.	n
	1	6.67	0.52	6	6.83	0.41	6
	2	7.90	0.32	10	7.80	0.42	10
	3	8.67	0.50	9	8.33	0.50	9
	4	9.89	0.33	9	9.44	0.53	9
	5	10.33	0.50	9	10.40	0.53	10
	6	11.40	0.52	10	11.20	0.44	10

3. Results

To ensure statistical independence, the class constituted the unit of analysis. Within each class consequently, data from the three to four participants of each sex were averaged to generate one boy's score and one girl's score per class. To ensure an adequate sample size, data were collapsed from grades 1 to 2, 3 to 4, and 5 to 6 producing three grade levels. There were five classes from grades 1 to 2 (two from grade 1 and three from grade 2), six classes from grades 3 to 4 (three from each grade level), and six classes from grades 5 to 6 (three from each grade level). An analysis of variance (ANOVA) was conducted for each variable, with sex and grade as the independent variables.

No sex differences were found in reports of either frequency (M=1.94 for males; M=1.76 for females, F<1), nor duration (M=2.43 for males; M=2.04 for females, F(1,28)=1.93) of conflicts between current same-sex friends. Neither the effects of grade (F(2,28)=2.09, n.s. for frequency and F<1 for duration) nor the Sex × Grade interactions (F<1 for frequency and F(2,28)=1.39, n.s. for duration) were significant.

Likewise, no sex differences were obtained in reports of levels of immediate emotional distress (M=2.60 for males; M=2.83 for females, F(1,28)=1.09, n.s.) nor long-term emotional distress (M=1.44 for males; M=1.49 for females, F<1) following conflicts between current same-sex friends. There was a significant effect of grade for level of immediate distress, F(2,28)=4.47, p<.025. Tukey's test, p<.05, revealed that compared to children in both older grade levels (M=2.57 for grades 3–4; M=2.44 for grades 5–6), the youngest children (M=3.20) reported that conflicts generated greater emotional distress. The remaining effects were all nonsignificant, Fs<1.

Nonetheless, participants reported significantly more same-sex friendships had ended between girls than between boys, F(1,28)=8.88, p=.006. Neither the main effects of grade, F<1, nor the interaction between sex and grade, F(2,34)=1.48, n.s., was significant. Table 2 displays the means and standard deviations for past friendships within the three grade levels.

Table 2. Mean number of same-sex friendships that had ended

	Grade level	Number of classes	Females		Males
		n	M	S.D.	M	S.D.
	1–2	5	1.02	0.96	0.22	0.33
	3–4	6	0.61	0.25	0.50	0.28
	5–6	6	0.85	0.45	0.26	0.39

A more stringent analysis then was conducted in which only the total number of different same-sex friendships that had ended in each class was included. That is, if two or more children of the same sex in a class named the same friendship that had ended, that friendship was included only once in calculating the total score for the children of that sex in that class. Using a matched-pair t test, the results remained unchanged, t(15)=3.11, p=.007 (females: M=2.47, S.D.=1.80; range 1–7 past friendships per class; males: M=1.06, S.D.=1.08; range 0–3 past friendships per class).

Finally, the number of males and females who named at least one past same-sex friendship in the class that had ended was compared. Over twice as many females (56.60% or 30/53) as males (27.77% or 15/54) named at least one same-sex friendship in their classes that had ended, χ2=9.12, p=.003.

4. Discussion

Results demonstrate that beginning as early as 6 years of age, even amongst young children with access to abundant resources, external observers report that fewer males' than females' same-sex friendships had ended. Given that the study examined females' preferred form of relationship, dyadic friendships, the findings represent a stringent test of the hypothesis that bonds between males are more durable than those between females. Interestingly, although participants were not asked about other-sex friendships, five boys across the grade levels spontaneously reported that one or more friendships between girls in their class that had ended. By contrast, only one girl in the first grade reported that a pair of boys were no longer friends.

It has been well established that human mothers maintain more durable ties than fathers to offspring (for reviews, see Brown, 1991, Hewlett, 1992). It is extremely rare for a mother to terminate a relationship with her child, whereas fathers vary widely in their degree of parental investment. Researchers therefore often assume that the same biological mechanisms that ensure greater investment by mothers than fathers in kin extend to other forms of relationships (Taylor et al., 2000, but see Geary & Flinn, 2002). Results from the current study suggest that sex differences in degree of parental investment do not necessarily translate into sex differences in investment in unrelated same-sex individuals.

Knowledge of the ecological costs and benefits that accrue from mothers' investing in unrelated same-sex individuals may provide clues as to reasons for this sex difference. To illustrate, research with chimpanzees indicates that sociability for mothers with dependent offspring relates to degree of scramble competition over resources, particularly food (Wrangham, 2000). When food is abundant, females can afford to associate with other unrelated females who may provide benefits such as protection from males and other predators (Boesch & Boesch-Achermann, 2000) or opportunities for infants and juveniles to play (Goodall, 1986). In common chimpanzees, high status females who are better able to obtain food (Pusey, Williams, & Goodall, 1997) also are more likely to affiliate with one another Boesch & Boesch-Achermann, 2000, de Waal, 1982. When food is scarce however or when females are low status, they may avoid other unrelated females to conserve resources for themselves and their offspring. Whether the same algorithms evolved in human females or the same ecological factors apply to modern humans however remains unknown. Nevertheless, it is reasonable to hypothesize that the uniquely human costs of simultaneously provisioning multiple dependent offspring (Lancaster, 1991) may discourage females from investing too heavily in another individual who confronts the same costs and as such cannot supply surplus resources.

Relative to females, males may benefit more at present or have gained more in the ancestral past from maintaining bonds with other males. de Waal (1989) suggests that chimpanzee males who require alliances in order to increase their dominance rank have evolved to reconcile conflicts rapidly. In contrast, chimpanzee females who depend far less on other chimpanzees have not evolved such mechanisms for reconciliation mechanisms. Rodseth and Wrangham (2004) further propose that alliances are more important for males than females because territorial politics and ultimately warfare are predominantly male domains.

The current study fails to provide behavioral evidence for sex differences in number or duration of conflict or levels of short- or long-term distress generated by conflict in current same-sex friendships. Possibly external observers were unaware of conflicts between children who currently were close friends. Suggestive findings from past studies of children's and adolescents' peer relationships indicate that compared to males, females report that they have been more hurt by their current closest same-sex friends (Benenson & Christakos, 2003) and that they are more likely to remain angry after a conflict (e.g., Lagerspetz et al., 1988). Further, observers of young children find that females are less likely than males to utilize joking to repair relationships (Butovskaya, Verbeek, Ljungvery, & Lunardini, 2000). These findings merit replication.

The current research cannot be generalized to adults with offspring. No research has compared the durability of same-sex bonds between adult males versus females. One difficulty is that in modern societies, adult relationships can be maintained at large distances with infrequent contact (Rodseth & Wrangham, 2004), so that it is unclear how to define and measure investment in a relationship. Self-sacrifice or cooperation constitute potentially valuable definitions of investment. Analogous data on adults must be collected before conclusions can be drawn regarding the evolutionary origins of sex differences in investment in unrelated individuals.

In summary, the current findings indicate that as early as 6 years of age, males form more durable bonds than females with unrelated same-sex individuals. If the same results are replicated with additional measures of investment and the findings extended to adults, this suggests that in human beings, investment in same-sex relationships exacts a higher cost for females relative to males or alternatively provides a larger benefit to males. Knowledge of ancestral and current ecological conditions will be necessary to provide a full evolutionary account of the basis for this sex difference.