The definitions of friendship presented in the previous section refer only to the most basic criterion or feature of friendship (i.e., mutual liking). Certainly, there is more to friendship than this mendasar defining characteristic. In this section, we show that the interactions of friends differ substantially from the interactions of nonfriends. The centerpiece of this discussion is well known, specifically, that reciprocity is a fundamental characteristic of friendship.
Beyond the simple reciprocity of liking discussed in the previous section, we show that reciprocity between friends can be seen in several domains, including affect and behavior. “Reciprocity” refers to the tendency of two persons to act in the same way, either simultaneously or in sequence (Hinde, 1979). Typically, reciprocity is directional in the sense that actions of friends are directed toward each other, such as when one friend does something and the other friend responds (Ross, Cheyne, & Lollis, 1988). The responsiveness between friends is often symmetrical, because each friend contributes equally to the interaction and neither of the two partners dominates the other.
Evidence of the presence of affective and behavioral reciprocity as a central feature is seen clearly in the findings of a meta-analysis of studies of friendship. Newcomb and Bagwell (1995) organized the rich set of interactions that had been studied in friendship research into four broadband categories: positive engagement, conflict management, task activity, and relationship properties. In each of these four areas substantial differences were found between the interactions of friends and nonfriends. The overall size of this difference for the category of positive engagement was nearly half of a standard deviation (Cohen’s d+ = 0.472).
Similar differences were found on each of the four types of interaction within this category: a difference of nearly two-thirds of a standard deviation on social contact, one-half a standard deviation on talking, and roughly one-third of a standard deviation on cooperation and positive affect. Smaller differences were seen on conflict management (Cohen’s d+ = 0.128); however, consistent with comments we made earlier, there were large differences between the forms of interaction within this category.
Although there were no differences on measures of conflict instigation, a difference of at least moderate size was found on measures of conflict resolution. For the category of task activities, an overall small effect was observed (Cohen’s d+ = 0.224), but again, the size of the difference between the interactions of friendship varied across the particular measures in this category. A moderate difference (roughly two-fifths of a standard deviation) was found for measures of task performance, whereas a nonsignificant difference was found for measures of basic communication. With regard to indices of relationship properties, moderate to large differences were observed across several dimensions (overall Cohen’s d+ = 0.397). These dimensions included similarity (roughly 0.4 of a standard deviation), equality (0.4), dominance (–0.2), mutual liking (0.75), closeness (0.61) and loyalty (0.54).
Although there were some age differences, Newcomb and Bagwell’s (1995) findings can be generalized across the childhood and adolescent periods. For example, even during the preschool period, children’s behavior with their friends differs from how they behave with children they know but who are not their friends. In addition to differences on features such as supportiveness and exclusivity (Sebanc, 2003), differences can be seen in the interactions of friends and of nonfriends as young as 3½ years on measures such as the directness and frequency of social overtures, the degree of engagement in social interactions, and complexity in play behavior (Dunn & Cutting, 1999; Dunn, Cutting, & Fisher, 2002).
As well, preschool-age friends tend to cooperate and exhibit more positive social behaviors with each other than with nonfriends (e.g., Dunn et al., 2002). It is known also that friendship stability in the preschool years varies across friendships and is related to patterns of interaction. Ladd, Kochenderfer, and Coleman (1996), for example, have shown that, compared with unstable friendships in early childhood, stable friendships involve higher levels of positive friendship qualities (e.g., validation) and lower levels of negative friendship qualities (e.g., low conflict) are most likely to be stable.
Although preschool children have been shown to engage in more conflicts overall with friends than with other peers (Hartup, Laursen, Stewart, & Eastenson, 1988), the way that friends resolve conflicts has been shown to be more cooperative. Specifically, compared with nonfriends, preschool friends engage in more arguments, more assaults and threats, and more reactive hostility, such as refusals and oppositions (Dunn & Cutting, 1999; Laursen & Hartup, 1989). These differences are likely due to the larger amount of time that friends spend interacting with each other. It is the manner in which friends, relative to nonfriends, resolve conflicts that shows the higher level of responsivity that friends have for each other.
For example, Hartup and his colleagues (1988) also reported qualitative differences in how preschool friends and nonfriends resolve conflicts, and in the likely outcomes of these conflicts. Friends, as compared with nonfriends, make more use of negotiation and disengagement, and they make relatively less use of winner-takeall strategies in their resolution of conflicts. In terms of conflict outcomes, friends are more likely than nonfriends to have equal resolutions. Also, following conflict resolution, friends are more likely than nonfriends to stay in physical proximity and continue to engage in interaction.
Two studies conducted with school-age children demonstrate quite clearly the differences shown in Newcomb and Bagwell’s (1995) meta-analysis. In one study, pairs of 7- and 8-year-old children, equally divided between friends and nonfriends, were observed while watching a comedy film (Foot, Chapman, & Smith, 1977). The interactions of the two children in each pair were coded into five categories (laughing, smiling, looking, talking, and touching). In each of these categories, the observed frequencies and durations were higher for the pairs of friends than for the nonfriends. More importantly, a higher level of interdependence was observed between the children in the friend pairs than in the nonfriend pairs. Specifically, children in the friend pairs were more likely than the children in the nonfriend pairs to respond to the behavior of their partner. An example of a response would be when one child would laugh after the other child had laughed. These findings point to the much higher level of coordinated positive affect in friendship.
Similarly, Newcomb and Brady (1982) observed the interaction of pairs of schoolage and early adolescent boys who were either friends or acquaintances as they worked with a “puzzle box,” a brightly colored wooden case about the size of a small footlocker. It included 15 features. Five features could be manipulated by a single child (e.g., a combination lock hooked to a latch on the outside of the box), five needed the coordination of two people (e.g., a flashlight that could be illuminated only if each of the two children pushed buttons located at different places on the box), and five could be manipulated by one child or by two (e.g., playing with two cars kept in a drawer in the box). As in the Foot et al. (1977) study regarding matched affect (e.g., mutual laughter), differences were observed between the friend and acquaintance pairs on all measures of coordinated positive affect (e.g., mutual laughing).
Differences were also observed on measures of coordinated activity. Friends were more likely than acquaintances to share activities and to participate in features that required the action of two people. Most importantly, the friend pairs were more likely to engage in exploration together. These findings point to the overall higher level of coordination and mutual responsivity for friends relative to nonfriends in both affect and behavior.