Age and Sex Role Differences in Intimate
Friendships During Childhood and Adolescence

Intimacy as experienced in best friendships was studied with respect to sex differences and sex roles, and the development of components of intimacy in childhood and adolescence. Results from a sample of 218 individuals aged 8 to 14 showed that best friendship intimacy is relatively low at age 8, but increases in late childhood; that some components of intimacy develop earlier than other components; and that traditional sex differences, which typically find females higher than males in intimacy, are better explained as sex role differences in which females and androgynous males form a homogeneous high intimacy group, while sex-typed males score significantly lower.

Recent literature in the United States has reflected a growing interest in the nature and quality of intimate friendship, with more of this interest centering on the relationships among children and adolescents (Berndt, 1982). Erikson (1963, 1968) specifically postulated that the central crisis of the developing human was identity vs. diffusion, and that adolescence was the time for this process to take place. After the development of identity it was possible to share oneself with another in the formation of intimate relationships, including, but not limited to, genital sexuality. Before the formation of identity, however, the person is not able to share and commit a self which is not fully differentiated, and not fully understood. Sullivan (1953ab), on the other hand, saw the beginnings of intimate relationship in preadolescence, roughly ages 8 to 10, in the same-sex relationship he called a "chumship." He maintains that there appears in what he terms "preadolescence" (around the age eight-and-a-half, and "sometimes decidedly later") a definite "need for interpersonal intimacy" (1953b, p. 246) which is qualitatively different from all friendships and associations prior to that age, and is related to later love and friendship relationships throughout life.

Perhaps because of Erikson's emphasis on the psychology of males and his apparent view of heterosexual orientation as normative, actual studies of intimacy traditionally have centered on Erikson's view that intimacy is possible only after the establishment of identity, therefore intimacy is the "task" of the young adult (Boyd & Koskela, 1970; Kacerguis & Adams, 1980; Orlofsky, 1978; Orlofsky, Marcia & Lesser, 1973). In fact, until the work of Bigelow and his associates (1975, 1977, 1980), Mannarino (1976, 1978ab, 1979, 1980) and Sharabany and her associates (1974, 1981), most researchers seemed to view childhood as a time when friendships were casual and relatively devoid of deeper meaning, and adolescence as a transitional period in which the characteristics of adult intimacy were first imitated and then, gradually, experienced in later adolescence.

Now it is becoming clear that intimate friendships can become a part of a person's experience beginning somewhere around middle childhood. It seems likely that intimacy, like other personality features, develops over a long period of time, and that it is not a single dynamic, but a collection of components that can be broken down and evaluated separately (La Gaipa, 1977; 1979; Sharabany, 1974; Sharabany, Gershoni & Hofman, 1981). Such components of intimacy, then, might be observable in the behavior of people long before adulthood, and possibly even before the formation of an integrated identity, as required in Erikson's scheme. The reasoning necessary for such a view may be present in the formulations of Sullivan, whose description of early intimacy (chumship) as a characteristic of preadolescence and as a prerequisite or facilitator of identity formation seems to explain the apparent contradiction.

Intimacy in the present study is defined as closeness to a best friend, as reflected in an individual's self-described knowledge of that friend's feelings, honesty with the friend, loyalty, willingness to share, enjoyment of companionship, trust, and attachment. Clear distinctions between intimate and non-intimate behavior are elusive, and are likely to remain so, due to the subjective nature of the experience and a sort of "relativity" problem. It is rather like trying to define the point at which a growing person stops being "short" and begins to be "tall." For this reason, it is assumed that best friendships are intimate, and that a continuum from "low intimate" to "high intimate" is the appropriate distinction for the hypotheses to be tested.

The central purpose of this study is to clarify and extend the existing work on intimate friendships in childhood and adolescence, and to investigate at least one alternative direction for future research into sex differences in the experience of intimate behavior. Existing research includes studies by Bigelow and La Gaipa (Bigelow, 1977; Bigelow & La Gaipa, 1975; La Gaipa, 1979) which demonstrated developmental trends in children's cognitive development of generalized friendship ideas and expectations. Mannarino focused more on children's experience of their own best friendships, drawing on Sullivan's (1953ab) theory to demonstrate that friendship in late childhood is more than mere association. His approach, however, was to describe the relationship of outward behavioral aspects of friendship to other personality traits such as altruism (1976, 1979) and self concept (1978a). He tended to deal with friendship attributes such as stability and mutuality (1978b), rather than an analysis of the child's inner experience. Sharabany has provided the deepest look into children's intimate friendships with the development of her intimacy instrument (1974) and its use in a limited developmental model (Sharabany, Gershoni & Hofman, 1981).

In addition to the need for a broader and more complete developmental approach which includes younger children and which looks at every age level through mid-adolescence, none of the studies of children's friendships has considered sex role differences in intimacy development. The present study is designed to meet these needs by testing two hypotheses.

Hypothesis 1 predicts that each component of intimacy -- and hence overall intimacy -- will increase from initial (age group 9) levels. The test of this hypothesis expands Sharabany's work (1974; Sharabany, Gershoni & Hofman, 1981) by measuring intimacy over a developmental span, beginning at grade 3 (age 8), rather than grade 5, and filling in the ages represented by grades 6 and 8, which were not included in the Sharabany et al. (1981) developmental design.

Hypothesis 2 is a test of the relative importance of sex differences as compared with gender role differences in intimacy. Specifically, Hypothesis 2 predicts that sex-typed males will score lower in intimacy than all females, but androgynous males will show no such difference. Most studies of intimacy, including those of intimate friendships in childhood and adolescence, find marked sex differences (Coleman, 1974; Douvan & Adelson, 1966; Eder & Hallinan, 1978; Hodgson & Fischer, 1979; Lever, 1978; Mark & Alper, 1980). In general, studies have shown that at all levels of development females rate higher than males in intimacy and have more friendships characterized as intimate (Maccoby & Jacklin, 1974). It appears that females are consistently socialized to be more intimate, while males are socialized to compete and maintain individuality at the expense of intimacy.

It is important to note that the roles and socialization processes which lead to such behavior are not necessarily inherent in human nature. Mead (1935/1963) is but one of many anthropologists who have noted that sex roles are not so much innate as they are cultural, with some cultures such as the Tchambuli fostering sex roles that are diametrically opposed to typical Western roles. Because socialization patterns seem to be the basis for sex differences in intimacy, it seems the logical direction in further understanding such differences is to determine whether it is simply gender, or gender-role which affects the friendship behavior of children and adolescents.

The work of Sandra Bem (1974, 1977, 1981) has provided theoretical ideas which can guide such an inquiry. Her formulation of psychological androgyny has been useful in identifying culturally-prescribed masculine-typed or feminine-typed behaviors and attitudes and the presence (or absence) of these in varying degrees in each individual.

Subjects

The sample consists of 116 female and 102 male volunteers from 10 classrooms in a suburban unified school district near Los Angeles. Age of these subjects ranged from 8.5 years to 15.3 years, with a mean of 11.96; mean age for females was 12.05 and for males was 11.85. For purposes of age-based analyses, subjects were assigned to six age groups from 9 to 14, with individuals between age 8.5 and 9.4 assigned to age group 9, those between 9.5 and 10.4 to age group 10, and so forth. Due to low n among the older subjects, age group 14 was extended to include ages 13.5 to 14.9, thus allowing all but one individual (age 15.3) to be included in the sample. These additional subjects (n = 6) were in the same grade (8) as the others in agegroup 14. Among age groups 9 through 14, the largest was age 12 (28 females, 20 males), and the smallest was age 9 (8 females, 10 males). Proportions of females to males in each group were approximately balanced, the largest imbalance being age 13, with 62% females (n = 21) and 38% males (n = 13).

The Self Perception Inventory (SPI). This is a 60-item version of the Bem Sex Role Inventory (BSRI) (Bem, 1974; 1981) which was developed and validated for use with pre-adult individuals by Thomas and her colleagues (Thomas, 1983; Thomas & Robinson, 1981; S.J. Thomas, The Self Perception Inventory, personal communication, February 19, 1985). Instead of the BSRI's adjectives (to which the subject responds on a scale of one to seven as to how appropriate for the subject each adjective seems), the SPI presents children and adolescents with short statements, to which the individual responds never (like me), usually not, sometimes, usually, or always. Forty of the items are scored, with the other twenty functioning as filler or distractor items. Of the 40 scored items, 20 are reduced to a mean femininity score and 20 to a mean masculinity score in a manner identical to the BSRI (see Bem, 1981, for detailed scoring procedures). Each subject, whether male or female, receives both a femininity and a masculinity score. Examples of feminine-scored items are, "I am a shy person", "I am a loving person", or "I act like a girl". Examples of masculine-scored items are "I am able to take care of myself", "I like to find out how and why things happen", or "I act like a boy".

This modification of the BSRI uses the same method specified by Bem (1977, 1981) for scoring and classification, namely a simple median-split technique. Using the sample as a whole (i.e., males and females together), one median is calculated for the femininity scale and one for the masculinity scale. Respondents who are above the median on either scale are classified as high on that scale, and those below the median are classified as low. If a person is high on both scales (masc and fem) that person is classified as androgynous; males who are high masc and low fem are sex-typed, and males who are low masc and high fem are cross-sex-typed; females who are high fem and low masc are sex-typed, and females who are low fem and high masc are cross-sex-typed; individuals low in both categories are classified as undifferentiated.

Since the present study uses a developmental design, consideration was given to using separate medians for each age group, rather than the usual single set of medians for the sample as a whole. Though the medians for individuals in each age group varied from 3.3 to 3.58, ANOVAs showed no significant differences in femininity or masculinity levels across age groups nine to 14. The decision was made to use overall group medians to assign sex role categories. Further research with much larger samples will be needed to determine whether the differences in masculinity and femininity medians by age group justify the use of separate medians in developmental research.

Validity of the SPI instrument was measured (Thomas, 1983) by correlation with the Adolescent Sex Role Inventory (ASRI) (Thomas & Robinson, 1981), resulting in correlations of .875 for the Masculinity scales and .880 for the Femininity scales of each measure. The ASRI, in turn, had been compared favorably with the BSRI, in part by performing factor analyses on each; the factor structures were found to be essentially the same (Thomas & Robinson, 1981). The present study assessed the reliability of the SPI by computing coefficient alphas for the femininity and masculinity scales. Despite a relatively small sample size for reliability purposes, the scales yielded acceptable alphas of .67 (femininity) and .68 (masculinity).

The Sharabany Intimate Friendship Scale. The Sharabany Scale (Sharabany, 1974) is a 64-item Likert-scaled questionnaire consisting of sentences descriptive of friendship. Half of the items are worded from the point of view of the subject (e.g., I like to do things with her), and the other half are the same questions worded from the point of view of the subject's best friend (e.g., She likes to do things with me). This split yields a "Self" and an "Other" subscale, respectively, consisting of 32 items each. The entire scale is further subdivided into eight components with eight items each (4 items worded for "Self" and 4 for "Other"). Components are 1) Frankness and Spontaneity, 2) Sensitivity and Knowing, 3) Attachment, 4) Exclusiveness, 5) Giving and Sharing, 6) Imposition, 7) Common Activities, and 8) Trust and Loyalty.

Reported reliability and validity data (Sharabany, 1974) indicate considerable internal consistency and high item-total and cluster (component)-total correlations. Sharabany's results demonstrate not only that the entire scale measures intimacy as a general factor, but also that there is a distinctive contribution from each cluster of items (i.e., each intimacy component) indicating that the 8 components are measuring relatively independent aspects of intimacy. The data from the present study also show strong reliability, with overall coefficient alphas for each of the 8 scales ranging from .77 to .89. Reliability is strong even in the 9 and 10-year-old age group measured separately; this is the first use of the Sharabany scale with individuals younger than ten years, yet alphas for this age group ranged from .78 to .89.

Two sessions were scheduled for each group of individuals to whom the two questionnaires were to be administered. All sessions were conducted personally by the investigator, without assistants, and all items were read out loud by the investigator in all sessions, regardless of the age level of the subjects. Participants were instructed to read along with the investigator, and circle the answer that was best for them after hearing each item. In addition to the formal instruments, subjects provided demographic information (name, age, sex, school and grade), and answered questions about their best friendships.

Since the items in the Sharabany instrument referred to the respondent's best friendship, respondents were asked to identify for themselves their one best friend; booklets given to females contained female-specific pronouns and booklets given to males contained male-specific pronouns. Variations in sex of pronouns were discussed only with those individuals who named an opposite sex best friend (9 out of the 218 subjects) and wanted to know how to mark their booklets. Such individuals were told to ignore the sex of the pronouns and mark each item as if it referred to their best friend, regardless of sex. In every other respect the questions and other information requested of the subjects were identical. Each time the subjects were asked to identify their best friend, they first were given a definition of best friendship by the investigator, and were asked if they had any questions. After each testing session data were committed to computer and after all testing was completed, analysis was performed via SPSS-X.

Results and Discussion

Development of Components of Intimacy

The data showed partial confirmation of the first hypothesis of the present research which predicted that each intimacy component in the Sharabany Intimate Friendship scale (and hence overall intimacy) would increase from beginning (age group 9) levels. Overall intimacy means in the six age groups from 9 to 14, in order, were: 3.71, 4.01, 4.36, 4.22, 4.25 and 4.20. (These calculations represent the 201 subjects with nonmissing intimacy scores, while some later tests are performed only on the 144 androgynous and sex-typed individuals. It should also be noted that, except for androgynous males, the cells for each sex role group across age groups contained ns approximately proportional to their relative total group size. Androgynous males [n = 19] were overrepresented in age groups 9 and 10 [n = 9] and underrepresented in age groups 13 and 14 [n = 3]). While a main effect for age in this six-level age grouping only approached significance (p<.08) in the Intimacy X Sex X Age ANOVA, a similar ANOVA of the data collapsed across three age groups groups (ages 9/10, 11/12 and 13/14) found a significant main effect of age (F[2, 195] = 3.497, p<.05), and a significant deviation from linearity as well (F[1, 193] = 4.386, p<.05). Post-hoc cell contrasts (Scheffe=.05) showed that 11/12 was significantly higher than 9/10. Means, in chronological order, were 3.89 (n = 40), 4.29 (83) and 4.22 (78). This progression shows the general pattern of rising to a peak at age group 11, followed by a leveling trend, a pattern which was evident also in many of the separate components.

All eight components showed overall increases in intimacy from beginning levels, but oneway ANOVAs performed on the intimacy means of each component by age group (9-14) reached significance only for Frankness and Spontaneity (component 1), F(5, 195) = 3.834, p<.005; Sensitivity and Knowing (component 2), F(5, 195) = 5.392, p<.0005; and Exclusiveness (component 4), F(5, 195) = 3.302, p<.01. Means and results of cell contrasts for each component by age group are found in Table 1. It is believed that a larger sample size would be necessary for confirmation of the trends in the remaining components.

Table 1
Intimacy Component Means by Age Group
_______________________________________________________________
     Component     Age Groups
       Name          9     10     11       12      13       14

                n=  16     24     42       41      32       46
 1. Frank/Spontan  3.45   3.90   4.27     4.24    4.36     4.50(ab)
 2. Sens/Knowing   3.30   3.87   4.43(b)  4.48(b) 4.55(ab) 4.52(b)
 3. Attachment     4.20   4.23   4.64(a)  4.18    4.45     4.31
 4. Exclusiveness  3.63   3.74   4.15(ac) 3.87    3.80     3.47
 5. Giving/Sharing 3.87   4.24   4.47(a)  4.34    4.34     4.36
 6. Impos/Taking   3.52   3.87   4.15(a)  4.00    3.98     3.91
 7. Common Activ   3.46   3.59   3.90     3.92(a) 3.68     3.75
 8. Trust/Loyalty  4.25   4.60   4.86(a)  4.70    4.83     4.73
_______________________________________________________________
   (a).Highest (peak) level for each component
   (b).Significantly higher (p<.05, Scheffe) than age group 9
   (c).Significantly higher (p<.05, Scheffe) than age group 14

In general, the results shown in Table 1 are consistent with previous research, taking into account that most previous studies cannot be compared rigorously due to differences in terminology, target populations, definitions of friendship or intimacy, and similar inconsistencies. (See Table 4 in Sharabany, Gershoni & Hofman, 1981, p. 806, for a comparison of their work with studies by Bigelow & La Gaipa, 1975; Bigelow, 1977; and Reisman & Shorr, 1978.) For example, Sensitivity/ Knowing begins low in the present study, and increases with age, as it did in Sharabany, Gershoni and Hofman (1981). This seems reasonable in light of the concurrent cognitive development of this age range. Frankness/ Spontaneity in this study begins low, but increases immediately to become one of the highest components at age 14. La Gaipa (1979) similarly noticed a sharp increase in his variable "authenticity" between age 11 and 13, with the increase continuing into adolescence. The same increase was noted by Sharabany, Gershoni and Hofman (1981) between grades 5 and 11. Exclusiveness here is found at a moderate level to begin with (grade 3, age group 9), then becomes quite important in early adolescence with fluctuation thereafter. This fluctuation after age 11 parallels the findings of Sharabany, Gershoni and Hofman (1981), and would seem to fit with the observations of Douvan and Adelson (1966), among others, that the intensity of adolescent friendships is often not equalled thereafter.

Sex differences in intimacy across age groups. As noted above, males in general score lower than females in intimacy throughout the literature. Unlike previous studies, however, the design of the present study also permits a developmental analysis of how this sex difference may change across age groups 9 through 14. In the same ANOVA used in analyzing overall sex differences (Intimacy X Sex X Age), there was a significant interaction between sex and age, F(5,189) = 2.432, p<.05. Figure 1 displays this progression graphically.

In order to determine which age levels contributed to the significant interaction, a t-test was performed on each set of means. No differences were significant until age 11, at which point females were suddenly higher than males (21 females [M = 4.62], 21 males [M = 4.09], t[40] = 2.46, p<.05). The difference at age 12 approached significance (25 females, [M = 4.41], 16 males [M = 3.92], t[39] = 1.93, p = .06). At age 13, however, female levels regressed slightly and male levels increased substantially, closing the gap again. Finally, at age 14, females were at their highest level and males were at their lowest, the difference once again reaching significance (25 females [M = 4.62], 21 males [M = 3.63], t[44] = 4.92, p<.001).

With regard to comparisons of means across age groups within each sex, the present data revealed no deviation from linearity in ANOVAs using six age groups, though similar ANOVAs using the three age groupings showed significant differences among females in an age progression that deviated from linearity (F[1, 106] = 6.041, p<.05). Means for females in the three age groupings were 3.82 (9/10), 4.5 (11/12), and 4.49 (13/14); the mean for ages 9/10 was significantly lower than the other two. Males still showed no such significant differences over time.

Gender role and intimacy level differences.

Virtually all studies which report sex differences in friendship and intimacy indicate that males score lower on these variables than females. Results of the present study, when analyzed in the traditional female vs. male comparison, do confirm previous research with females reporting more intimacy than males (main effect of sex in an ANOVA of intimacy X sex X age, F[1,189] = 18.040, p<.001). But Hypothesis 2 in the present investigation predicted that sex-typed males -- not males in general -- would be "responsible" for such apparent sex inequality, while androgynous males would score equally high with all females, androgynous as well as sex-typed. This prediction was confirmed.

To investigate Hypothesis 2 and others involving questions of sex role, an Intimacy X Sex X Sex Role (androgynous vs sextyped) X Age (collapsed into three groups) ANOVA was calculated. Main effects were found for sex (F[1, 132] = 4.015, p<.05) sex role (F[1, 132] = 8.030, p<.01), and age (F[2, 132] = 4.020, p<.05), and there were significant two-way interactions between sex and sex role (F[1, 132] = 6.022, p<.05) and between sex role and age (F[2, 132] = 3.779, p<.05). These results, however, only confirm intuitive expectations that sextyped individuals differ from androgynous individuals in intimacy, and that females differ from males. The question remains whether one group (expected to be sex-typed males) differs significantly from the other three, as hypothesized.

A oneway ANOVA was calculated on the intimacy scores of 44 sex-typed males (M = 3.95), 42 sex-typed females (M = 4.39), 41 androgynous females (M = 4.51), and 17 androgynous males (M = 4.56), and significant differences were found, F(3, 140) = 5.773, p<.001. Post-hoc Scheffe comparisons showed that sex-typed males were significantly lower than all three other groups at the .05 level. Hypothesis 2 was confirmed. Because of extremely low ns, cross-sex-typed females and males were excluded from these analyses. Undifferentiated individuals also were excluded because of the lack of consistent findings in the literature regarding these subjects (Bem, 1984).

Gender role analysis by intimacy component.

A breakdown of the overall intimacy score into the eight components allows a further analysis of relative intimacy levels of the gender role categories. Means by gender role category for each component are found in Table 2. Intimacy X sex role ANOVAs revealed significant differences among sex roles on seven of the eight components (all but Component 7, Common Activities) with sex-typed males showing the lowest measured intimacy levels in every case. Summaries of these ANOVAs, with cell contrasts, also are reported in Table 2.

Table 2
Means and ANOVA (Intimacy X Sex Role)
Results by Intimacy Component
_____________________________________________________________
     Component   1     2     3     4     5     6     7     8
                 M     M     M     M     M     M     M     M
Androgynous
 41 Females    4.58  4.74  4.60  4.01  4.71  4.31  4.06  5.09

   17 Males    4.45  4.64  4.90  4.33  4.63  4.22  4.15  5.11

Sex-typed
   42 Females  4.46  4.54  4.71  3.87  4.54  4.14  3.85  5.02

   44 Males    3.91  4.05  4.02  3.59  4.13  3.79  3.57  4.57

          F =  4.90  4.15  6.80  3.72  4.12  3.08  3.18  4.20
       d.f. = 3,140 3,140 3,140 3,140 3,140 3,140 3,140 3,140
          p = <.005 <.01  <.001 <.05  <.01  <.05   n.s. <.01
_____________________________________________________________

Notes.
Italicized means are significantly higher (Scheffe=.05) than
means for sex-typed males.
Components are: 1) Frankness and Spontaneity, 2)
Sensitivity and Knowing, 3) Attachment, 4) Exclusiveness,
5) Giving and Sharing, 6) Imposition and Taking, 7) Common
Activities, and 8) Trust and Loyalty.
 ____________________________________________________________
Note. Differences between females and males were significant only
in age groups 11, 12 and 14.

Gender role analysis by age group.

The significant interactions of Sex by Sex Role and Sex Role by Age reported above (Intimacy X Sex X Sex Role X Age ANOVA) suggest the probability that patterns of change over time are different for different sex role groups. Scheffe comparisons of cell means within each of the four sex role groups across age categories reached significance at the .05 level only for androgynous females, whose mean of 3.67 (age group 9/10) was significantly lower than their means of 4.7 (11/12) and 4.67 (13/14) in a progression that deviated significantly from linear (F[1, 38] = 7.98, p<.005). Means (non-significant differences) of the other two sex roles across age groups were as follows: androgynous males, 4.16 (9/10), 4.98 (11/12), and 4.67 (13/14); sex typed females, 4.02 (9/10), 4.48 (11/12), and 4.49 (13/14); and sex-typed males, 4.22 (9/10), 3.89 (11/12), and 3.92 (13/14). As can be seen, the sex-typed males were the only group that did not increase in intimacy from initial levels.

One further set of cell contrasts was made, in order to see at what point the sex-typed males began to "look" different from the other groups in intimacy levels. Each part of the three-level age grouping was viewed separately, showing that in the 9/10 age group there were no differences between the four sex role group means. In the 11/12 age group, however, sex typed males and females were significantly lower than androgynous males and females (F[3, 59] = 6.653, p<.001,Scheffe=.05), while in the 13/14 age group, sex typed males were significantly lower than androgynous females (F[3, 47] = 4.708, p<.001, Scheffe=.05). Obviously, at least in the present sample, sex-typed males began to differ from their age mates around age 11, remaining lowest in intimacy through age 14.

Two important findings have resulted from the present investigation. First, it was found that overall intimacy, as well as separate component means, increase from initial levels found at age group 9, though some components may follow non-linear trends and differences in only three components reached significance. It was noted that larger sample sizes in future research may confirm apparent trends in the other components. Second, traditional sex differences in intimacy, with girls showing more intimacy than boys, were confirmed in the present study, but a further analysis on the basis of sex roles revealed that only sex-typed boys were significantly lower than girls in intimacy; androgynous boys scored equally high with sex-typed and androgynous girls.

The study has confirmed and extended previous findings that intimacy develops in the same sex best friendships of childhood and stabilizes somewhat as subjects approach middle adolescence. Some components of intimacy are apparently more salient than others. Future research should investigate the possibility that other components of friendship, such as the physical contact involved in putting arms around a friend, or accuracy of one's knowledge about a best friend's feelings, might be added to the eight components now in the Sharabany scale. Also, it is time for a longitudinal design in studies of the development of intimacy and its components, using consistent and replicable terminology, definitions and criteria.

Finally, and most importantly, the present investigation clearly has shown that sex-typed males are at a disadvantage with regard to intimacy development, when compared to androgynous males and all females (i.e., both sex-typed and androgynous). This disadvantage, however, does not seem to arise until age 11, when males in general for the first time show significantly lower intimacy levels than females.

Perhaps the most obvious follow-up to the present study is a replication with late adolescent and young adult subjects of the portion in which intimate friendship levels are analyzed according to gender role differences, rather than simple sex differences. If the finding holds true, that sex-typed males are lower than females in intimacy but androgynous males are not, then the results of all previous studies in which females have been found to be higher than males on measures of intimacy must be questioned and, perhaps, reinterpreted. Whether or not the finding holds true, it would be useful to initiate longitudinal research which begins in childhood with subjects in gender role categories and traces both their gender role development (some individuals change from one category to another over time) and their intimacy development.

In such research particular attention should be paid to androgynous males and to how they compare over time with sex-typed males and with females. If some friendships are less functional than others, as appears to be the case with sex-typed males perhaps because of rigid and unrealistic social expectations, it would seem the time has come for further research which identifies and challenges the negative aspects of male socialization and supports those behaviors that lead to more constructive relationships. Friendships are valuable enterprises which both derive from and contribute to a strong, healthy society. When they are mutual and constructive, intimate friendships become an integral part of individual and group functioning, and should be encouraged in whatever form, in whatever configuration they naturally emerge.

Bem, S.L. (1974). The measurement of psychological androgyny. Journal of Consulting and Clinical Psychology, 42, 155-162.

Bem, S.L. (1977). On the utility of alternative procedures of assessing psychological androgyny. Journal of Consulting and Clinical Psychology, 45, 196-205.

Bem, S.L. (1981). The Bem Sex-Role Inventory Professional Manual. Palo Alto, CA: Consulting Psychologists Press.

Bem, S.L. (1984) Androgyny and gender schema theory: A conceptual and empirical integration. In T.B. Sonderegger (Ed.), Nebraska Symposium on Motivation: Psychology and Gender. Lincoln: University of Nebraska Press.

Berndt, T.J. (1982). The features and effects of friendship in early adolescence. Child Development, 53, 1447-1460.

Bigelow, B.J. (1977). Children's friendship expectations: A cognitive-developmental study. Child Development, 48, 246-253.

Bigelow, B.J., & LaGaipa, J.J. (1975). Children's written descriptions of friendship: A multidimensional analysis. Developmental Psychology, 11, 857-858.

Bigelow, B.J, & LaGaipa, J.J. (1980). The development of children's friendships. In H.C. Foot, A.J. Chapman & J.R. Smith, Friendship and Social Relations in Children (pp. 45-64). New York: John Wiley and Sons.

Boyd, R.D., & Koskela, R.N. (1970). A test of Erikson's theory of ego-stage development by means of a self-report instrument. Journal of Experimental Education, 38, (3), 1-14.

Coleman, J.C. (1974). Relationships in Adolescence. London: Routledge and Kegan Paul.

Douvan, E., & Adelson, J. (1966). The Adolescent Experience. New York: Wiley.

Eder, D., & Hallinan, M.T. (1978). Sex differences in children's friendships. American Sociological Review, 43, 237-250.

Erikson, E.H. (1963). Childhood and Society. New York: W.W. Norton.

Erikson, E.H. (1968). Identity: Youth and Crisis. New York: W.W. Norton.

Hodgson, J.W., & Fischer, J.L. (1979). Sex differences in identity and intimacy development in college youth. Journal of Youth and Adolescence, 8, 37-50.

Kacerguis, M.A., & Adams, G.R. (1980). Erikson stage resolution: The relationship between identity and intimacy. Journal of Youth and Adolescence, 9, (2), 117-126.

La Gaipa, J.J. (1977). Testing a multidimensional approach to friendship. In S. Duck (Ed.), Theory and Practice in Interpersonal Attraction. London: Academic Press.

La Gaipa, J.J. (1979). A developmental study of the meaning of friendship in adolescence. Journal of Adolescence, 2, 201-213.

Maccoby, E.E., & Jacklin, C.N. (1974). The Psychology of Sex Differences. Stanford, CA: Stanford University Press.

Mannarino, A.P. (1976). Friendship patterns and altruistic behavior in preadolescent males. Developmental Psychology, 12, 555-556.

Mannarino, A.P. (1978a). Friendship patterns and self-concept development in preadolescent males. Journal of Genetic Psychology, 133, 105-110.

Mannarino, A.P. (1978b). The interactional process in preadolescent friendships. Psychiatry, 41, 308-312.

Mannarino, A.P. (1979). The relationship between friendship and altruism in adolescent girls. Psychiatry, 42, 280-284.

Mannarino, A.P. (1980). The development of children's friendships. In H.C. Foot, A.J. Chapman & J.R. Smith (Eds.), Friendship and Social Relations in Children (pp. 45-64). New York: John Wiley and Sons.

Mark, E.W., & Alper, T.G. (1980). Sex differences in intimacy motivation. Psychology of Women Quarterly, 5, 164-169.

Mead, M. (1963). Sex and Temperament in Three Primitive Societies. New York: William Morrow and Co. (Original work published 1935, Second edition, 1950).

Orlofsky, J.L. (1978). The relationship between intimacy status and antecedent personality components. Adolescence, 13, 419-441.

Orlofsky, J.L., Marcia, J.E., & Lesser, I.M. (1973). Ego identity status and the intimacy versus isolation crisis of young adulthood. Journal of Personality and Social Psychology, 27, 211-219.

Reisman, J.M., & Shorr, S.I. (1978). Friendship claims and expectations among children and adults. Child Development, 49, 913-916.

Sharabany, R. (1974). Intimate friendship among kibbutz and city children and its measurement. Dissertation Abstracts International, 35, 1028B-1029B. (University Microfilms No. 74-17,682).

Sharabany, R., Gershoni, R. & Hofman, J.E. (1981). Girlfriend, boyfriend: Age and sex differences in intimate friendship. Developmental Psychology, 17, 800-808.

Sullivan, H.S. (1953a). Conceptions of Modern Psychiatry. New York: W.W. Norton.

Sullivan, H.S. (1953b). The Interpersonal Theory of Psychiatry. New York: W.W. Norton.

Thomas, S.J. (1983, April). Providing a concrete context for sex role endosement (sic) in adolescents: Some measurement considerations. Paper presented at the annual meeting of the American Educational Research Association, Montreal.

Thomas, S., & Robinson, M. (1981). Development of a measure of androgyny for young adolescents. Journal of Early Adolescence, 1, 195-209.