Friendship

Robert B. Hays
George Washington University, Washington DC, USA

This chapter reviews the properties associated with the friendship bond, then discusses the development, evolution, and effects of friendship. Friendship is conceptualized as a highly flexible, dynamic, multidimensional process, the structure and functioning of which will vary depending on characteristics of the individuals involved, the environmental context, and the developmental stage of the friendship.

The term 'friend' is used very loosely and idiosyncratically, by both the general public and social scientists, to describe a diverse range of relationships. A 'friend' may be a casual companion with whom we play racquetball once a week, an intimate confidant with whom our most private thoughts and feelings are shared, someone we interact with every day, someone who lives across the country and we only exchange letters with several times a year, someone we just met a few days ago, or someone we've known all our lives. Further, the nature and structure of friend relationships varies greatly for individuals at different points in the life cycle (Bigelow and La Gaipa, 1975; Dickens and Perlman, 1981), of different sociocultural backgrounds (Allan, 1979; Brain, 1976), and, to some extent, of different sexes (Winstead, 1986), as well as at different developmental stages of the friendship (Hays, 1985; LaGaipa, 1977). The protean quality of friendship presents a definitional problem for investigators that has impeded the development of a coherent body of knowledge on friendship. On the one hand, researchers often ask respondents to identify friends without specifying or examining the criteria employed in deciding whom to include, making generalizations across studies difficult. On the other hand, when researchers do impose their own definitions of friendship, several problems arise. First, the definition may not match the respondent's phenomenological meaning of friendship or be appropriate to the particular respondents or contexts. Second, one investigator's definition of friendship may differ from that of another. Third, researchers' definitions often impose requirements, such as frequent contact, same sex, or 'closeness', that exclude particular types of friends (e.g. Iong-distance, cross-sex, nonintimate) and so friendship's full spectrum is nor represented in the literature.

Friendship can be considered a social institution in that it is a symbolically recognized system that provides a dependable source of rewards for a large number of people (Suttles, 1970; see also McCall's chapter in this volume), yet clearly friendship is a peculiar institution. Described as the 'least programmed of all social relations' in Western society (Suttles, 1970), there are no formalized duties, obligations, or legal bases for the friendship bond (Hess, 1972; Kurth, 1970). Nor are there any public ceremonies or rituals to mark the initiation or progression of friendships, or honor and celebrate a friendship's existence. As anthropologist Robert Brain (1976) discusses, many non-Western cultures treat friendship very differently; some take it as seriously as marriage in terms of formal role requirements and public legitimacy. For example, in the African Bangwa society, community elders assign children best friends and that relationship carries with it well-understood rights and rituals, for 'having a best friend was as important as having a wife or a brother-possibly more important' (Brain, 1977, p. 94). In Western cultures, however, friendship remains a 'noninstitutionalized institution'.

In order to achieve a valid understanding of friendship, the task of the social scientist is therefore to develop theoretical models and research methodologies that take into account the flexible and idiosyncratic quality of friendship. An essential first step is establishing a comprehensive definition of friendship.

The basis of friendship, as with all interpersonal relationships, lies in an interdependence between individuals, meaning that the behavior of each participant is to some degree coordinated with and influenced by the behavior of the other (Kelley, 1979). Hinde (1979) further states that for a relationship to truly be said to exist, the individuals' interactions must occur over an extended period of time and there must be some degree of continuity between successive interactions. In other words, a relationship between two people is a dynamic process, with each interaction affected by the dyad's past interactions. It is the nature of the interdependence between friends that distinguishes friendship from other interpersonal relationships. In contrast to kin or work relationships, the interdependence between friends is voluntary. As Allan (1979) points out, friendship is never a necessary consequence of role positions. This aspect of friendship has been noted in virtually all theoretical conceptualizations of friendship (Kurth, 1970; Reisman, 1981) and has been most clearly articulated by Wright (1974), who proposes 'voluntary interdependence' as the key behavioral criterion of friendship: 'Two people are friends to the degree that the plans, activities and decisions of one of them are contingent upon those of the other in the absence of constraints toward interaction that are external to the relationship itself' (p. 93). Likewise, Hartup (1975) defines friends as 'people who spontaneously seek the company of one another; furthermore, they seek proximity in the absence of strong social pressure to do so' (p. 11).

The primary goal of the interdependence between friends is social-emotional rather than instrumental meaning that friends derive satisfaction from their interactions themselves (e.g. in the form of companionship, stimulation, belongingness, emotional support) rather than engaging in interaction primarily to achieve an ulterior motive. In a survey of over 1000 adults, Fischer (1982b) found that sociability (i.e. interactions such as discussing pastimes and engaging in recreational activities) was the main criterion used in labeling someone a 'friend'. Likewise, the enjoyment of each other's company consistently emerges as a key property of friendship in studies that asked respondents to define friendship (Crawford, 1977; Davis and Todd, 1982; Weiss and Lowenthal, 1975). Similarly, in a study comparing the preferred activities of friends, kin, and colleagues, Argyle and Furnham (1982) found that friends' interactions were the least task-oriented and tended to revolve around leisure activities such as eating, drinking, and talking. The social-emotional basis of friendship certainly does not exclude the possibility of friendship serving instrumental purposes (e.g. friends helping each other study, get a job, repair a car, etc.); the critical factor lies in the primary motivation for the individuals' involvement. In Allan's (1979) words, 'individuals can be useful because they are friends, but not friends because they are useful' (p. 43).

Friendships vary greatly in the number and types of goals the participants seek in a particular relationship. Reisman (1981) has distinguished between 'reciprocal friendships' which are emotionally close, committed long-term friendships and more casual 'associative friendships' that are based on shared roles and propinquity. This distinction parallels Hess' (1972) description of diffuse, highly involving, intimate friendships versus more narrowly differentiated and compartmentalized friendships such as bowling friends or drinking buddies. Clear examples of diffuse, reciprocal friendships can be seen in the best-friend relationships of adolescents described by Richey and Richey (1980):

the best friend is more or less a constant companion, a confidant with whom one can share very private information, a critic/advisor whose counsel is acceptable, a standard against which to measure oneself, an ego support whose affection and respect for one are known and reliable, an understanding ally, and a moral support in times of crisis. (p. 537)

In contrast, an excellent case of compartmentalized friendship is provided by Rosecrance's (1986) description of 'racetrack buddies' whose interaction is limited to occasional afternoons enjoying horse races together, yet nonetheless is experienced as a satisfying friendship. The types of goals attained in particular friendships will be influenced by a number of factors. Hess (1972) notes that while preadults can afford the time demands posed by diffuse friendships, middle-aged adults tend to form more compartmentalized friendships. McCall and Simmons (1966) suggest there is a 'strain toward totality' in relationships such that as a friendship develops, more and more areas of the friends' lives are brought into it.

Implicit in the social-emotional nature of friendship is what Wright (1974) refers to as the 'person qua person' component of friendship: friends perceive and respond to each other as unique, genuine, and irreplaceable. In other words, friends interact in such a way that they reveal to each other at least part of their 'real' versus 'role' selves (Kurth, 1970). Both Argyle and Henderson (1984) and Davis and Todd (1982) have found that an expectation of intimacy underlies the friendship bond. With increasing intimacy comes an increasing mutual understanding between friends (Davis and Todd, 1982; LaGaipa, 1977). The degree and types of personal knowledge shared between friends can vary from narrow and superficial to broad and highly intimate (Altman and Taylor, 1973) and will depend on an interplay of individual, dyadic and environmental factors. For example, women's friendships tend to involve higher degrees of intimate disclosure than men's friendships (Winstead, 1986). Close friends are more self-disclosing than casual friends (LaGaipa, 1977) and physical proximity has been found to influence the types of information friends know about each other (Rubin and Shenker, 1976).

Given the voluntary nature of friendship, some degree of positive affective bond is necessary to hold a friendship together. This aspect of friendship is included in virtually all theoretical conceptions of friendship (Kurth, 1970; Hess, 1972; Reisman, 1981; Davis and Todd, 1982; LaGaipa, 1977). Indeed, Webster's dictionary definition of a friend as 'one attached to another by affection or esteem' places its sole emphasis on this dimension. The degree and components of two friends' affection can range from the positive regard one may feel for a casual friend, akin to Rubin's (1973) description of 'liking', to a more intense emotional experience between close friends that can genuinely be called love in the sense that the friends deeply care for and depend on each other. Since friends know each other's 'real selves', a friend's positive regard also communicates a sense of acceptance (Davis and Todd, 1982; LaGaipa, 1977; Rose, 1985) that contributes to feelings of trust and spontaneity within the relationship- the security to relax and 'be completely yourself' (Crawford, 1977; Davis and Todd, 1982).

The emotional bond between friends is not necessarily all positive, however. Analyzing friendship from a psychoanalytic perspective, Rangell (1963) notes that any relationship as intense as friendship also brings with it a residue of ambivalence and negative feeling. Likewise, Wright (1974) notes that 'maintenance difficulty' is a potential by-product of interdependence in friendship. Longitudinal studies of friendship have shown ambivalence and conflict to be predictable consequences of increasing friendship intensity (Eidelson, 1980; Hays, 1985) and Duck and Miell (1986) have found that feelings of uncertainty and doubt often underlie college students' friendships.

Friendship can also entail what Kurth (1970) refers to as a sense of 'mutual obligation' between partners: friends help each other. When asked to describe what friendship means, this characteristic of a friend as a source of support,'someone I can call on for help', consistently emerges (Crawford, 1977; Davis and Todd, 19X2; LaGaipa, 1977; Tesch and Martin, 1983; Weiss and Lowenthal, 1975). For example, in a survey of Soviet adolescents' conceptions of friendship, Kon and Losenkov (1978) found 'mutual aid and loyalty' to be one of the most frequently mentioned expectations of a friend. Similarly, four of the friendship rules identified by Argyle and Henderson (1984) reflect this theme (volunteer help in time of need, show emotional support, strive to make him/her happy, stand up for the other in their absence). Importantly, a fifth rule states that friends should 'repay debts and favors', emphasizing the need for a reciprocal 'give and take' between partners in a friendship (Allan, 1979). A degree of equality between individuals on key dimensions such as age and socioeconomic background is often cited as a precondition for friendship (Brain, 1976; Button, 1979), partly because it insures the potential for equity in the social exchange (Jackson, 1977). Friendship is also an 'equalizer' in the sense that if individuals who are disparate on a significant dimension such as age, sex or status do become friends, it is assumed that they will respond to each other symmetrically, as if the status differential were irrelevant. For this reason, some organizations such as military or work settings often actively discourage friendships between individuals of different statuses so as to not threaten the organization's hierarchy.

Interestingly, even though the respondents in Weiss and Lowenthal's (1975) survey of friendship overwhelmingly regarded mutual supportiveness as the most valued characteristic of an ideal friendship, they also reported that it was not often realized in their actual friendships. The degree to which friends help and support one another varies greatly depending on characteristics of the individuals and level of friendship. For example, mutual support is more frequently mentioned in descriptions of women's friendships than men's (Crawford, 1977; Weiss and Lowenthal, 1975). Bigelow and LaGaipa (1975) have found that the view of one's friend as a receiver of help is not prominently mentioned by children until around grade six (i.e. the age of 11-12). In addition, helping and emotional support are more characteristic of close than of casual friendship (Hays, 1985; LaGaipa, 1977).

Having reviewed the characteristics most typically associated with friendship in the theoretical and empirical literature, I will now suggest the following definition of friendship: voluntary interdependence between two persons over time, that is intended to facilitate social-emotional goals of the participants, and may involve varying types and degrees of companionship, intimacy, affection, and mutual assistance. Embodied in this definition are several key premises First, individuals are seen as active and goal-directed in pursuing friendships (Miell and Duck, 1986; Ginsburg, 1986). Those goals, however, will vary in kind and number for a particular friendship depending on characteristics of the individuals (e.g. age, sex, personality), the developmental stage of the friendship (e.g. casual versus close friendship), and environmental factors (e.g. existing social networks, environmental opportunities). Thus there is no universally 'right' type of friendship or ideal endpoint that friendships are assumed to lead to (Altman, Vinsel and Brown, 1981). Second, friendship is a multidimensional phenomenon (LaGaipa, 1977; Wright, 1974), expressed simultaneously on behavioral, cognitive and emotional levels, and every friendship will show a unique pattern of interactional properties. Third, friendship is a dynamic process rather than a state; thus the properties that characterize a friendship dyad are continually evolving as the dyad progresses (Duck and Sants, 1983; Hays, 1985). Finally, since the friendship bond lies in the interrelation between individuals, the appropriate unit of analysis is the dyad as a holistic system rather than the individuals themselves (Ginsburg, 1986). Since situational and developmental factors influence the structure and patterning of the dyad's interdependence, the environmental and temporal contexts must be considered integral components of that system. In sum, friendship can best be understood as a dynamic, multidimensional process that unites individuals and contexts.

The initial formation of a friendship represents the convergence of individual, environmental and dyadic factors. At an individual level, personal characteristics such as friendship motivation (McAdams and Losoff, 1984), loneliness (Solano, 1986), marital status (Blau, 1961), and position in the life cycle (Hess, 1972; Shulman, 1975) have all been found to influence an individual's motivation to seek friendship. For example, people who are single, young or elderly direct more energy into cultivating friendships than do married or middleaged adults (Shulman, 1975). In addition, personal characteristics such as social skills (Cook, 1977), friendship schemas (Young, 1986), and shyness (Zimbardo, 1977) will influence an individual's ability to successfully initiate friendships.

Environmental factors create the opportunity for friends to meet and provide situational conditions that can foster or inhibit their developing a friendship. An abundance of research demonstrates that physical proximity contributes to friendship formation (Festinger, Schachter and Back, 1950; Nahemow and Lawton, 1975; Berscheid and Walster, 1978) in that it increases the likelihood two individuals will meet as well as the frequency with which they interact. Equally important, but less investigated, are environmental factors that influence the quality of the interactions between acquaintances. Particular environmental factors such as organizational climates, job designs, time schedules and interior design may promote friendly interaction more so than others. For example, the 'jigsaw classroom' described by Aronson el al. (1978) illustrated one way of structuring elementary school classroom activities so as to promote cooperation and the development of friendship between students of different ethnic backgrounds. In an interview study concerning friendship and work, Parker (1964) found that the nature of the respondents' work was associated with the extent to which they made friends with coworkers. Those employed in service occupations were more likely than business employees to have friends in the same line of work. Further, those who had positive attitudes towards their jobs and those whose jobs most encroached on their free time were most likely to draw friends from their work settings. In a survey of over 900 men living in metropolitan Detroit, the most common sources of the men's closest friendships were found to be work (26 per cent), neighborhood (23 per cent), childhood (20 per cent), voluntary associations (7 per cent), and family involvements (7 per cent).

Some degree of mutual liking is an essential condition for friendship formation. Research on interpersonal attraction suggests that at initial encounter, perceptions of another's attitude similarity, physical attractiveness, competence, positive responsiveness, and apparent liking for us will cause us to be attracted to some individuals over others (Berscheid and Walster, 1978; Huston and Levinger, 1978). Investigations of naturally occurring friendships show that friends tend to be more similar to one another than nonfriends in virtually every examined dimension, including age, sex, marital status, race, religion, attitudes, interests, personality traits, and intelligence (Verbrugge, 1983b; Huston and Levinger, 1978; Jackson, 1977). Interestingly, in contrast to laboratory research, which emphasizes attitude similarity (e.g. Byrne, 1971), studies of real friendships have found similarity in behavioral preferences to be more highly associated with friendship (Kandel, 1978; Werner and Parmelee, 1979). At initial acquaintance, characteristics of the other person serve as stimuli from which to make predictions about the potential rewardingness of the other as a friend (Altman and Taylor, 1973). As a friendship begins to develop, however, assessments of actual rewardingness become critical. Knapp and Harwood (1977) asked college students to rate the importance of a variety of characteristics in the formation of a close same-sex friendship and found attitudinal agreement, intimate accessibility, and reciprocal candor to be the most desired qualities.

The critical determinants of attraction for any friendship will depend on the goals the individuals have for the friendship. For example, shared values may be important for a confidant, but not for a bridge partner. Likewise, the evaluative criteria will vary with the developmental stage of the friendship. LaGaipa (1977) found that authenticity (openness and honesty; being real, genuine and spontaneous) was perceived to be important at all levels of friendship, whereas helping and self-disclosure were considered less important until the development of close friendship. In a longitudinal study of friendship development, Duck and Craig (1978) found that value similarity was correlated with friendship ratings at three months of acquaintance, whereas at eight months a more subtle personality characteristic, personal construct similarity, emerged as more associated with friendship ratings. In addition, the individual's existing social network will influence the types of attributes one finds attractive in a potential friend. For example, Lea and Duck (1982) found a tendency for friends to be similar in those relatively unique personal values for which it was uncommon for them to find similarity with others. Thus, the ability of a new acquaintance to provide rewards not currently available in one's social network may increase one's motivation to initiate a friendship with him or her.

Assuming two mutually interested individuals meet under circumstances conducive to personal interaction, the stage is set for the development of friendship. The beginning of a friendship is denoted by interaction that goes beyond that required by the individuals' formal roles, for example having lunch together, talking about one's personal life, playing a practical joke on another, etc. Duck (1983) points out that the desire to become closer is usually not directly stated, rather it is indicated by one's behavioral style. Kurth (1970) explains that initial moves towards friendship may be tentative and deliberately ambiguous, given risks of rejection plus uncertainty about the desirability of the other as a friend. Social exchange theorists (Huston and Burgess, 1979) suggest that when individuals find superficial interaction to be unusually rewarding, or when they feel the other has the potential to be rewarding in ways not currently available, they begin to deepen and develop the relationship. Several general theories of relationship development have been proposed (Altman and Taylor, 1973; Levinger and Snoek, 1972; Murstein, 1970) though a number of them are framed more with regard to courtship couples than platonic friendships. Given the qualitative differences between friendships and romantic relationships (Davis and Todd, 1982), one cannot assume the process of development to follow identical patterns. For example, Purdy (1978) asked college students to draw graphs of relationship development for same-sex friendships and heterosexual romantic relationships. Involvement in friendships was reported to increase more gradually, level off at a lower magnitude, and include fewer fluctuations than romantic relationships. Altman and Taylor's (1973) social penetration theory is the most thoroughly elaborated and researched model of relationship development that is applicable to friendship. According to social penetration theory, interpersonal exchange in a developing relationship operates on two dimensions-breadth (the content areas of exchange) and depth (the intimacy level of exchange). Relationship development is seen as proceeding in an orderly sequence from superficial interaction in narrow areas to increasingly intimate interaction in broader areas. Progression along the breadth and depth dimensions is governed by a process of dyadic exploration, evaluation and forecasting: an individual samples the rewards and costs of interacting with a potential friend and then decides whether to increase or decrease the level of involvement based on perceptions of the probable rewardingness of future interactions. Support for social penetration theory has been found in a number of studies (Altman and Taylor, 1973; Hays, 1984, 1985). For example, in two longitudinal studies of friendship development among college students, Hays (1984, 1985) asked first-term university students to complete an extensive behavioral checklist of the types of interactions they had engaged in within newly forming friendships at three-week intervals during their first semester. Analysis of responses during the first three weeks of acquaintance showed that the initial interactions of new friends corresponded to a Guttmanscale progression from superficial interaction (e.g. watching television, discussing local events, telling jokes) to increasingly intimate levels of behavioral exchange (e.g. discussing personal problems, visiting each other's relatives, giving gifts). At early stages of the friendships (three and six weeks), the quantity of behaviors the dyads engaged in accounted for most of the variance in attitudinal ratings of friendship intensity; however, as the friendships progressed, the intimacy level of the interaction accounted for an increasing percentage of the variance. Other investigators have also found increasingly intimate interaction between friends as their relationship progresses. Berg (1984), for example, in a longitudinal study of college roommates, found greater self-disclosure in the spring semester than the fall. Likewise, in a longitudinal diary study, Duck and Miell (1986) found that intimate 'sharing' conversations between new college friends increased in frequency over the course of the school year, and that the settings of their interactions shifted from public to more private locations.

Berg and Clark (1986) have recently questioned the assumption that friendship development follows as gradual a process as is implied by social penetration theory, suggesting instead that some decisions about the nature of the relationship and corresponding behavioral differences may occur quite early following initial acquaintance. For example, in a series of laboratory investigations with college students, Clark (Berg and Clark, 1986) manipulated participants' expectations that a fellow college student was a potential friend or not and found that subjects showed immediate differences in their responses to the target individual, in that they were more likely to follow 'communal' than 'exchange' norms (Clark and Mills, 1979) with a potential friend. In addition, in Hays' (1984, 1985) longitudinal studies, the dyads progressed to close friendship surprisingly quickly. After only six weeks, the dyads had reached their peak in number of intimate behaviors performed and the friendships had begun to stabilize (ratings of friendship closeness at six weeks were correlated 0.78 with friendship ratings five months later). These findings do not contradict the social penetration view that friendship development follows a systematic 'unfolding' of increasing intimacy levels, but do show that the rate of that process can be quite quick in some cases. Individual and situational characteristics will certainly be determining factors. For example, the friendships of college students who are roommates progress more quickly than friends who are not roommates (Hays, 1985). Likewise, the friendships of highly self-disclosing students may develop faster than those of more reticent individuals (Taylor, 1968).

Berg and Clark's (1986) analysis highlights the catalytic role that often overlooked cognitive factors such as friendship scripts, goals and strategies can play in the development of friendship. Through in-depth interviews with university students concerning the communication strategies they used in developing friendships, Miell and Duck (1986) found that self-disclosure is often offered and elicited purposively as a means of gathering information about a potential friend or strategically channeling the flow of communication to intensify or restrict the growth of a friendship. Their work nicely complements Altman and Taylor (1973) by suggesting cognitive strategies that might guide the process of social penetration.

In addition to increases in the depth and breadth of interaction, a number of other changes characterize the transition from initial acquaintance to close friendship. One assumption of social exchange approaches to relationship development (Huston and Burgess, 1979) is that the value of the resources exchanged increases as a relationship progresses. Evidence for this with regard to friendship was found in Hays' (1985) previously described longitudinal study: ratings of the amount of benefits received from interacting with one's friend were highly correlated with ratings of friendship closeness at all assessment times. Moreover, the benefit ratings of dyads which successfully progressed to close friendship by the end of the school term showed a linear increase over time, in contrast to the declining ratings of the dyads which did not progress. Content analyses of free-response questions which asked participants to list the types of benefits received from their friendships identified the following benefits of friendship: companionship, having a confidant, emotional support, information exchange, material or task assistance, self-esteem, and general value of having a friend. Interestingly, whereas companionship was the benefit of friendship most frequently cited by all dyads, the greatest differences between close and nonclose friends appeared in reports that the other person was a confidant and that one received emotional support from the friend, suggesting that those are the critical ingredients that make a friendship 'close'. Similarly, in a diary study in which college students kept records of their interactions with close friends and casual friends for one week (Hays, 1988), close and casual friendships did not differ in the amounts of fun, task assistance or intellectual stimulation experienced but close friends provided more emotional and informational support than did casual friends. Likewise, LaGaipa's (1977) examination of friendship expectations for varying levels of friends found that helping and selfdisclosure distinguished close friendship from more casual friendships.

The increasing emphasis on mutual support and confiding as a friendship evolves from a casual to a close friendship highlights the qualitative changes that occur at different stages of friendship development. Kelley (1979) has described the development of a close relationship as marked by a transformation in the partners' motivations, shifting from a focus on one's immediate reward-cost outcomes towards an increasing responsiveness to the joint outcomes of oneself and one's partner. The norms of exchange operating within a friendship are thus altered as the friendship increases in closeness. O'Connell (1984), for example, interviewed individuals whose friends had helped them build their own homes. Though the friends were typically given no payment for their help, this 'one-sided giving' tended not to result in dissatisfaction or a weakening of the friendship ties; indeed many interviewees felt the experience served to strengthen the friendships. Respondents felt the 'friendship licence' bestowed a right to request help without imposing an obligation to reciprocate. In a 'communal' relationship (Clark and Mills, 1979), friends respond to each other's needs and helping the other is in itself personally satisfying. Further, close friends do not expect immediate 'tit for tat' reciprocation, but assume an eventual balance will be reached in the long term (Levinger, 1980). Similar patterns have been found in self-disclosure between close friends (Derlega, Wilson and Chaikin, 1976).

The emergence of unique pair norms in the dyads' communication and interactional styles also characterizes the development of close friendship. For example, in an analysis of naturally occurring telephone conversations, Hornstein (1985) found that when speaking with close friends, females used a highly implicit conversational style compared to when speaking with casual acquaintances or strangers. Similarly, Rands and Levinger (1979) found that 'norm regulation' (e.g. dropping by unannounced, using other's things without permission) was perceived as more likely to occur between close friends than more casual friends. Knapp (1978) suggests that as a relationship successfully progresses, the partners' communication patterns become increasingly personalized, synchronized and efficient. In support of this, Baxter and Wilmot (1985) found in a diary study of college students that dyads in growing friendships reported more satisfaction, effectiveness, and personalized communication than nonprogressing dyads. In a similar vein, McCall (1970) suggests that a 'private culture' emerges within a friendship and becomes increasingly elaborated as the friendship progresses. This very interesting aspect of friendship development (see McCall's chapter, this volume) has received very little empirical examination, however.

A number of investigators have noted that ambivalence and conflict tend to accompany increases in dyadic involvement (Braiker and Kelley, 1979; Eidelson, 1980; Hays, 1985). In Thibaut and Kelley's (1959) words, 'To enter a new relationship is to abandon an old adaptation . . . some degree of conflict may be inevitable' (p. 66). In two longitudinal studies, Eidelson (1980) asked college freshmen to evaluate newly developing friendships at two-week intervals over the course of their first semester. Reports of interpersonal satisfaction increased during the early stages of friendship, then declined at approximately six weeks when the friendships had reached intermediate levels of involvement, and subsequently rose again with greater involvement. The drop in satisfaction was interpreted as the friends' reactions to the accumulation of restrictive costs that inevitably arise in a growing friendship. In Hays' (1985) longitudinal study, the most frequently mentioned costs in college students' friendships were: time expenditure, added responsibilities, emotional aggravation, loss of independence, negative influence of friend on self, and negative effects of friendship on other relationships. The consistent positive correlations between ratings of costs and attitudinal ratings of friendship closeness suggested that increasing costs accompanied friendship progression. In particular, reports of emotional aggravation experienced in the friendships, which were fairly infrequent at early stages of friendship, sharply increased at nine and twelve weeks of acquaintance, with one-third of all dyads mentioning instances of aggravation or conflict. The participants' comments suggested that the increase in emotional aggravation resulted from increased knowledge and familiarity between friends. Thus, increasing self-disclosure and interdependence not only are the mechanisms by which friendships are built, but also open up the potential for increased disagreement and disenchantment within friendships.

Braiker and Kelley (1979) point out that conflict and negativity can play a potentially positive role in interpersonal relationships by offering the dyad anopportunity to grow by 'thrashing out their differences' and thus achieving a more satisfying interpersonal adaptation. Insight into how the negotiation of potentially negative aspects of interaction can serve to help a friendship grow is provided by Rawlins' (1983a, b) dialectical analyses of friendship communication. Through in-depth interviews with ten pairs of close friends, Rawlins (1983a) identified the dialectic of expressiveness-protectiveness that is inherent in the development of friendship. Through their expressiveness or selfdisclosure, two friends open up their areas of vulnerability to each other. In achieving this openness, however, the dyad also creates the conditions for closedness (protectiveness), for the friends must thereafter strategically manage their communications so as to protect the other's discovered vulnerabilities. According to Rawlins, the appropriate management of this dialectical tension contributes to the development of trust between the friends. In a similar manner, by successfully negotiating the freedom to be simultaneously independent and dependent within the friendship, a friendship becomes more secure and valued by the participants (Rawlins, 1983b). Investigations into the processes by which friends handle relationship dissatisfactions and tensions would greatly increase our understanding of friendship.

Friendships are continually evolving. Specific friendships may stabilize at a particular level of intimacy or within a circumscribed range of activities, but a friendship cannot be static. To continue to exist, a friendship requires ongoing interaction between the partners. Each interaction between friends can be seen as not only expressing the degree and type of friendship the two individuals have (cf. Watzlawick, Beavin and Jackson, 1967) but also impacting their friendship- serving to maintain, strengthen or weaken the dyad's bond. Baxter and Wilmot (1986) have also discussed the need for conceptualizing friendships as 'in process' (i.e. growing closer or more distant). There has been very little research, however, on the processes by which established friendships are maintained and evolve (see Duck and Sants, 1983).

The day-to-day functioning of friendships will obviously vary greatly depending on individual, dyadic and situational factors. In terms of the frequency of contact between friends, sociological surveys of social participation (Jackson, 1977; Verbrugge, 1983b) show that the young and elderly interact with their closest friends more often than do the middle-aged. Single people see their friends more than do married people; students more than professionals and self-employed; and neighbours more than friends who live further apart. As Verbrugge (1983b) concludes, convenience may be the primary determinant of interaction frequency between friends. For example, in a study of black urban wives, Feagin (1970) found that friendship contact varied inversely with the distance between friends' residences. Consequently, Jackson (1977) found the somewhat paradoxical result that the urban men in his survey tended to have the most frequent contact with their least intimate friends (neighbors and coworkers), while the most intimate friends were seen the least often. If situational factors permit it, close friends tend to interact more frequently than casual friends (Hays, 1985; Verbrugge, 1983b). For example, in Hays' (1988) diary study of college students, close friends reported interacting about six times during a week whereas casual friends averaged between three and four interactions. Once a friendship has reached a certain degree of closeness, however, friends may not need to interact very frequently to maintain their friendship. As Hays (1985) found, at advanced stages of friendship development, the quality of a dyad's interaction becomes more important than simply the quantity. Thus, the friendship bond between friends with very limited interaction, as in the case of long-distance friends, can endure if the rewards of their periodic contacts are sufficiently valued. An examination of the processes by which long-distance friendships are maintained would be a valuable area of future research.

When friends do get together, Argyle and Furnham (1982) have found their interactions typically center around joint leisure and recreational activities, personal conversations, eating and drinking. Psychology Today's (Parlee, 1979) friendship survey of over 40 000 respondents found that the activities most frequently engaged in with friends were: having an intimate talk, helping each other, eating a meal together, attending a movie or sports event, shopping, and playing sports. In contrast to notions of friendship as a highly intimate relationship, Duck and Miell (1986) found that superficial conversation was the most frequently occurring interaction between college friends. Similarly, casual communication and superficial companionship were the behavior categories most highly endorsed by college friends in Hays' (1984, 1985) longitudinal studies. Several investigators have examined in depth the content of conversation between close friends. In a questionnaire study of middle-aged adults, Aries and Johnson (1983) found the most common topics of conversation to be daily activities, community affairs, family activities, family problems, reminiscences, and work. Interestingly, a similar study with college students (Aries and Johnson, 1983) showed that college friends' conversations were much more often personally focused, dealing with their intimate relationships and family issues. In both studies, female friends discussed personal topics more often and in greater depth than did male friends, who more frequently discussed sports. Davidson and Duberman (1980) examined reports of conversations between best friends and found that the women's conversations included a broader range of topics, particularly with regard to personal and relationship-oriented issues, than did the men's conversations, which emphasized external topics.

The findings cited above illustrate the consistent gender differences which emerge in studies of friendship interaction (see Hendrick, this volume). Females tend to emphasize emotional sharing, trust and confiding in their interactions with friends more so than males, whose friendship interactions often revolve around shared activities (Caldwell and Peplau, 1982; Crawford, 1977; Weiss and Lowenthal, 1975). Females are more likely than males to get together 'just to talk' (Caldwell and Peplau, 1982; Baxter and Wilmot, 1986). Personality characteristics can also be expected to influence the patterning of friendship, yet these have received little examination by researchers. One exception is the finding that high self-monitoring individuals are more likely to choose activity partners on the basis of a friend's skill in the particular activity, whereas the low self-monitor's choice is made more on the basis of general liking for one's friends (Snyder, Gangestad and Simpson, 1983).

Platonic friendships between men and women have received very little attention by researchers. As Booth and Hess (1974) discuss, social and occupational structures tend to be sex-typed, making it less likely that individuals of different sexes will meet, especially under conditions of equal status. Further, social norms often discourage cross-sex friendship, considering it a threat to marital relationships. In a survey of 800 middle-aged Midwesterners, close cross-sex friendships were reported by only 28 per cent of the respondents (Babchuk, 1965). Interestingly, the cross-sex friendships that did form were structurally similar to same-sex friendships in that they tended to be homogeneous with respect to age, education and social class. Qualitatively, however, cross-sex friendships appear to differ from same-sex friendships. Seventy-three per cent of the respondents in Psychology Today's friendship survey (Parlee, 1979) agreed cross-sex friendships were unlike same-sex friendships, citing as reasons sexual tensions, social discouragement, and the fact that opposite-sex friends have less in common. K. E. Davis (1985) found that cross-sex friendships were viewed as including less sharing of confidences and assistance and less tendency to 'give the utmost' than same-sex friendships, and were also perceived as less stable. Similarly, in an interview study comparing the functions of same-sex and cross-sex friendships (Rose, 1985), both women and men described their crosssex friendships as including less communication and providing less help than same-sex friendships. Further, less active strategies were reported in the formation of cross-sex friendships. Whereas same-sex friendship formation was seen as instigated by expressions of mutual affection, communication and acceptance, cross-sex friendships were more often characterized as being initiated by 'no strategy', developing gradually through the passage of time, or stimulated by an initial sexual attraction that eventually settled into being 'just friends'. With the societal relaxation in sex roles and increase of women in the workforce, the incidence of cross-sex friendships will surely increase, and their functioning perhaps change (cf. Dillard and Miller, this volume). The unique value of crosssex friendship as a significant contributor to greater understanding and equality between the sexes should not be underestimated, and merits investigation.

The end of a friendship can be precipitated by factors at an individual level (e.g. death, change in marital status), environmental level (e.g. moving to separate cities, changing jobs) or dyadic level (e.g. change in reward-cost ratio of the friendship). Factors at each level may be interrelated and may in combination bring about a friendship's dissolution (cf. McCall's chapter, this volume). Rose (1984) points out that one factor (e.g. change in reward level) may weaken the friendship bond, making it more vulnerable to other factors (e.g. moving). Kurth (1970) notes that a sense of relief may be felt when situational changes 'gracefully' end a less than satisfying friendship. As Duck (1982a) emphasizes, the dissolution of a personal relationship is not an 'event' but a process and cannot be viewed as simply a reversal of the developmental process since the partners' relationship experience (e.g. intimate knowledge of the other, shared memories) cannot be withdrawn. In contrast to romantic relationships and marriages, which often end with an emotional, protracted 'break-up', the dissolution of a friendship is more likely to be characterized as a gradual 'fading away'. Since there are no formal ties to be severed nor public announcements to be made, there is little need for negotiation in ending a friendship. It may also be difficult to distinguish temporary downswings in a friendship from absolute declines. Consequently, the dissolution of a friendship many times is only realized in hindsight and often regretted (Rose, 1984).

Rose (1984) asked 155 undergraduates to write essays describing reasons for the decline of a close, same-sex friendship. The main causes identified were: physical separation, new friends replacing old ones, growing to dislike characteristics of the friend's behavior or personality, and interference from dating or marriage relationships. Few friendships ended due to an internal 'breakdown', but were more likely disrupted by external factors.

Argyle and Henderson's (1984) research on friendship rules provides insight into dyadic factors that can contribute to friendship dissolution. They asked individuals to think of a specific friendship that had lapsed because of something about the relationship and to rate the extent to which failure to keep various rules had contributed to the breakdown. The rule violations found to be most critical included jealousy or criticism; lack of tolerance for a third-party relationship; disclosing confidences; not volunteering help when needed; nagging or criticizing the person publicly; and not showing trusting, confiding, positive regard or emotional support. In her conceptualization of relationship dissolution, Rodin (1982) refers to such violations of one's expectations of what a friend should be like as one friend meeting the other's 'dislike criteria'. Alternatively, Rodin (1982) suggests that a friendship can end because either partner's 'like criteria' change, that is, one begins to look for different things in a friend. What an individual expects, desires or needs from a friendship will vary with experience and lifespan changes (Bigelow and LaGaipa, 1975; Dickens and Perlman, 1981), therefore some friends may no longer provide the satisfactions they once did. Likewise, as individuals enter new situations, their comparison level for alternatives changes and they may encounter new acquaintances who are more rewarding than their current friends. From this perspective, letting go of particular friendships is a necessary and potentially growth-enhancing aspect of the individual's own developmental process. Hays and Oxley (1986), for example, found that the most adaptive social networks for first-term university students were those which were permeable, in that they brought in new friends who were also college students, rather than holding on completely to one's old high school or neighborhood friends who did not also enter college. Thus individuals must continually restructure their friendship networks to fit their changing identities and life situations. Enduring friendships may be those in which both partners grow in ways that continue to be mutually stimulating and satisfying.

In an interview study of college students, Rose and Serafica (1986) found that the processes responsible for keeping and ending friendships varied with friendship level. Casual friendships were described as likely to end due to decreased proximity, whereas close and best friends could withstand decreases in contact, assuming degrees of affection within the friendships did not change. Jackson, Fischer and Jones (1977) have made the distinction between 'friends of convenience' and 'friends of commitment'. Friends of convenience result primarily from situational factors that hold the friends together and are easily replaced when one moves on to a new situation. One's attachment to friends of commitment, however, is such that one holds on to them despite changes in external factors. A valuable direction for future research is the identification of characteristics of the friendship bond which contribute to a friendship's durability throughout the individuals' respective life changes.

An individual's friendships can have a profound impact-both facilitative and disruptive-on other aspects of the individual's life. The burgeoning literature on social support documents the potentially beneficial consequences of friendship, demonstrating that friends can significantly contribute to one's happiness and life satisfaction, ability to cope with stress, and vulnerability to illness (Fleming and Baum, 1986; Duck, 1983; Gottlieb, this volume).

Friends provide a multitude of functions for individuals, serving both as preventive resources (contributing to positive personal adaptation, thereby making stress and adjustment problems less likely to occur) and as buffers (mediating the effects of stresses that do occur). As Cicero wisely observed, 'A friend multiplies our joys and divides our sorrows'. Although the mechanisms by which friendships produce their positive outcomes are not fully understood (Fleming and Baum, 1986), Solano (1986) discusses three types of important resources that friendships can provide: (1) emotional (e.g. intimacy, support, acceptance, belongingness, self-esteem, status), (2) cognitive (e.g. stimulation, social comparison, information), and (3) material (task assistance, tangible support).

Friendship has long been recognized as an important contributor to an individual's personality development and socialization throughout the lifespan (Berndt, 1982; Hess, 1972; Sullivan, 1953). Sullivan (1953) postulated that the experience of intimacy and acceptance with a close, same-sex 'chum' during preadolescence provided a 'consensual validation' critical for individuals' developing a sense of personal worth. There are a number of aspects of the friendship process that can contribute to an individual's sense of self-esteem, including the recognition that one's friend knows, accepts and likes your 'real self' (Kurth, 1970); perceives the world the same way you do (Duck, 1973); and supports your view of yourself (Bailey, Finney and Helm, 1975). In addition, one's friend provides an opportunity to help, nurture or influence someone else (Z. Rubin, 1980; Candy, Troll and Levy, 1981).

Friendship plays an important role in an individual's socialization process by providing sources of social comparison and a safe setting for trying out new roles and gaining feedback and support (Hess, 1972; Weiss and Lowenthal, 1975). As Rangell (1963) says, 'friends serve as a giant projection screen or sounding board against which one can measure and check [one]self' (p. 20). For this reason, friends have been found to be especially critical during periods of role transition such as adolescence, young adulthood, and retirement (Dickens and Perlman, 1981; Hess, 1972). As peers, friends are able to furnish invaluable information and validation in a way family members are rarely able to. For example, Tokuno (1983) found that college females preferred to discuss developmental issues such as personal values and relationships with friends rather than family members. Similarly, Millen and Rolls (1977) found that adolescent males felt their closest friends were the people who understood them the best; fathers and teachers were seen as the least understanding.

Friendship has often been placed in a sort of 'competition' with other personal relationships, particularly kin and romantic relations, with the expectation that all an individual's needs could or should be met within the nuclear family. Thus friendships have often been viewed as unnecessary or interfering (Bell, 1981a; Brain, 1976). Brain (1977) emphasizes that such unrealistic assumptions-aside from their disservice to friendship-place an unfair burden on marital and family relationships. Weiss' (1973) research on loneliness demonstrated that a spouse provides a sense of emotional attachment whereas friends provide a sense of belongingness and social integration, thus each type of relationship fills unique and complementary needs for the individual. Kon and Losenkov (1978) found that adolescents regarded their best friends as those who understood them best and with whom they would most likely share intimate, confidential information. However, for advice on complicated life situations, both boys and girls reported they would turn to their parents first. Litwak and Szelenyi (1969) suggest that individuals tend to turn to their friends for help on issues in which a degree of similarity and affection is important, whereas family are turned to for problems of a more serious or long-term nature.

In a similar vein, different types of friends will serve different functions for individuals. In a study of foreign students' adaptation to the University of Hawaii, foreigners reported having two kinds of friends: Hawaiians and 'compatriots'. The Hawaiian friends were useful in helping them get settled and 'learn the ropes' in the new culture, while with their compatriots they could share familiar activities and personal conversation (Bochner, McLeod and Lin, 1977). In a comparison of the functions of cross-sex and same-sex friends (Rose, 1985), women reported that their women friends provided them with more acceptance and intimacy than did men friends. Similarly, Hays and Oxley (1986) found that both male and female college students reported receiving more emotional support from female friends than from male friends.

The level of a friendship will greatly influence the resources it offers. The intimacy, mutual concern and ease of communication that characterize close friendships may enable close friends to provide more emotional and information support than casual friends (Hays, 1988), thus close friends are perceived as more therapeutic than casual friends (Davidson and Packard, 1981). On the other hand, the high degree of interdependence and familiarity associated with close friends may interfere with their ability or motivation to provide effective support in some instances. Several studies have shown that close-knit friendship networks are not likely to encourage change or provide the resources needed for coping with situations that are unfamiliar (Hirsch, 1980; Walker, MacBride and Vachon, 1977). As Granovetter (1973) has emphasized, the 'strength of weak ties' such as casual friends and acquaintances lies in the fact that they are more distant, and thus able to provide access to new information and perspectives one may not normally encounter.

Clearly, however, there are limits to friends' abilities to help in some situations. An important focus for future research is in identifying the types of problems for which friends can be effective sources of support, and more fully explicating the interactional processes and contextual factors that contribute to their effectiveness. Also in need of investigation are the potential negative consequences of friendship. For example, the desire to be liked and accepted by one's friends or to help one's friends can cause individuals to behave in ways that may go against personal or societal standards and thus work against their best interests. Friendship groups can be particularly vulnerable to social psychological dynamics such as 'groupthink' (Janis, 1972) or deindividuation (Zimbardo, 1970), in which group involvement contributes to members engaging in actions that are more risky, impulsive or ill advised than would the individuals by themselves.

In terms of a scientific understanding of friendship, the surface has only been scratched. Investigators have described properties associated with the friendship bond, identified factors that precipitate the formation of a friendship, outlined at a general level some of the changes that occur as a friendship develops, and suggested a number of beneficial consequences friendship can have. Very little is known about the actual process of friendship: how friendship is maintained, why and how it grows and evolves, what mechanisms underlie the positive and negative functions friendship can serve, and how a friendship interrelates with the social and physical context in which it is embedded. Most research has focused on the casual and close same-sex friendships of college students. More attention needs to be given to different varieties of friendship (e.g. crosssex friends, long-distance friends, friendship cliques, 'best' friends, specialized friends, cross-generational friends) among diverse populations and environmental settings. Only then will a valid understanding of friendship be attained.