Close adult friendships,gender,and the life cycle

Abstract

Drawing on 25,185 responses collected via an online news website, our findings extend and update Fischer and Oliker’s (1983) classic study on gender and life cycle differences in friendship. We found no substantial gender differences in number of friends people can count on to celebrate birthdays, discuss intimate matters like one’s sex life, or depend upon when experiencing trouble late at night (ds = .04–.20); however, number of friendships varied substantially according to marital status, age, and parental status. Residential population size was not associated with number of friendships. We also found that virtually all respondents reported having at least one close friend. Satisfaction with friends was a better predictor of life satisfaction than was number of friends.

Keywords

Close relationships cross-gender friends friendship gender gender differences life cycle same-gender friends social support

In the ironically titled, Just Friends, Lillian Rubin (1985, p. 12) asserts, “Friends count,” in part because, “from early childhood, right on up through old age, friends are crucial to our well-being.” Many contemporary researchers concur with Rubin that friendship is important for both men and women across the life course (Crosnoe, 2000; Marion, Laursen, Zettergren, & Bergman, 2013; Siebert, Mutran, & Reitzes, 1999). In fact, not only do people with a greater number of friends tend to enjoy better physical and mental health (Cable, Bartley, Chandola, & Sacker, 2013; Fiori, Antonucci, & Cortina, 2006) but they also live longer (Perissinotto, Cenzer, & Covinsky, 2012; Sabin, 1993).

There are several demographic factors that contribute to the growing importance of friendship. High rates of geographic mobility result in many individuals spending relatively little time with faraway families of origin (van Diepen & Mulder, 2009). People remain single longer and many become single again after a first or subsequent marriage ends (Muraco, 2012). Trends toward smaller nuclear families (U.S. Census Bureau, 2012c) provide midlife adults with more time to enjoy leisure activities that are not family oriented.

Additionally, an extended life span increases older adults’ opportunities to seek out the company of same-age peers, especially after the loss of a life partner (Carstensen, 2006; Löckenhoff & Carstensen, 2004). Notably, Pinquart and Sörensen (2000) found that for older adults, subjective well-being was more closely associated with contact with friends than was contact with adult children.

Despite the significance of friendship, the topic has garnered less scholarly attention than other social relationships such as romantic relationships and kin relationships (Ryle, 2011).¹ One exception to this was a recent scholarly debate on social isolation among Americans that was put to rest by Paik and Sanchagrin (2013) who concluded, based on methodological flaws in prior research, that Americans are not becoming increasingly isolated.

The amount of engagement people have with other people is particularly consequential because social engagement has been associated with life satisfaction and happiness. For example, Diener and Seligman (2002) found that people who scored highest on life satisfaction were those who spent the most time socializing with friends and the least time alone. Nonetheless, some dissent about the relationship between friendship, life satisfaction, and happiness exists. For example, Gundelach and Kreiner (2004) found only a marginal correlation between spending time with friends and happiness.

But friendship quality is also important (Demir & Weitekamp, 2007). Frequent face-to-face contact with close friends decreases stress and improves health and overall well-being (Fuller-Iglesias, Webster, & Antonucci, 2013; van der horst & Coffe, 2012). Requena (1995, p. 272) maintains:

Compared to acquaintances, close friends are more likely to be responsive to one’s troubles, to sense the nature, degree, and source of one’s distress, and to engage in supportive behavior that is appropriate to one’s needs, even if costly in time or effort.

Thus, close friends are able to offer emotional support as well as instrumental forms of assistance, such as helping with car trouble or with relocation to a new home.

Understanding the number of friends people have and the quality of these friendships can help us determine the extent to which people have close friends they can turn to for companionship, confide in, and count on when they are in trouble. In this study, we sought to investigate both emotionally expressive and instrumental aspects of friendship, drawing upon a large sample of respondents from NBCNews.com (formerly msnbc.com), which is one of the most visited news websites in the U.S. (Quantcast, 2013).

This national data set enabled us to examine the extent to which demographic factors in addition to gender—especially age, level of relationship commitment, and parental status—are linked to the number of same-gender and cross-gender close friends people have as well as their satisfaction with their friendships. These data also revealed how many people had at least one close friend to confide in, revisiting the issue of social isolation in America. Finally, we examined how number of friends and satisfaction with friends related to female and male respondents’ overall self-rating of life satisfaction.

Background

Gender

Three decades ago, Fischer and Oliker (1983) used survey research to bring the study of friendship front and center in an oft-cited research note. They examined the interaction effect between gender and life cycle stage on number of friends and friendship composition (kin and nonkin) by using cross-sectional data with 1,050 adult (18+) respondents from Northern California (who were primarily White and urban).

They found that young men were likely to have more friends than young women but that this pattern reversed among older adults. Fischer and Oliker (1983) interpreted women’s greater number of friends in later years as evidence of “women’s greater disposition toward building friendships” (p. 131). Their supposition dovetails with much of the friendship literature.

In addition to focusing on gender differences in number of friends, many studies conclude that women share greater levels of emotional expressiveness and self-disclosure with their friends, while men engage in activities centered on common interests like sports. These purported gender differences are typically summarized thus: women’s friendships are conducted “face to face” while men’s are “side by side” (Caldwell & Peplau, 1982; Ryle, 2011).

These studies have sometimes been used to assert that women’s friendships are more intimate, deeper, and more meaningful than men’s. But as Cancian (1987, p. 701) cautions, “part of the reason that men seem so much less loving than women is that their behavior is measured with a feminine ruler.” Biases noted by Cancian include “measures based on feminine styles of loving, such as verbal self-disclosure, emotional expression, and willingness to report that one has close relationships” (p. 699).²

Cancian (1987) proposes that researchers use measures of closeness that include not only emotional expressiveness and intimacy but also the types of instrumental practices that are more typically associated with men’s friendships, such as providing pragmatic help and engaging in shared activities, especially physical ones. Walker’s (1994) in-depth interview-based study supports Cancian’s assertion that gender differences are relatively small when assessments of friendship are more inclusive. Although the men and women she interviewed recognized gendered friendship ideals (i.e., expecting women friends to share feelings and men to engage in activities together), many of these same participants actually departed from these gendered ideals in practice within their own same-gender friendships (e.g., female friends attended sporting events together and male friends shared personal feelings with each other). Other researchers have also suggested that reported gender differences in friendship may be exaggerated (Greif, 2009; Wright, 1988).

Studies of men’s and women’s assessments of their friendships find both differences and similarities. Jones (1991) found that women reported higher satisfaction with their friendships than did men; yet higher levels of self-disclosure were associated with friendship satisfaction for both men and women. Similarly, Reisman (1990) found that men as well as women who rate their friendships as low in disclosure also rate them as low in closeness and find their friendships to be less satisfying.

Whereas much of the literature on friendship and gender examines same-gender friendships, cross-gender friendships have nonetheless garnered some scholarly attention (Lenton & Webber, 2006). In two samples of college students, Bleske-Rechek and Buss (2001) found that both men and women frequently initiate cross-gender friendships for the same reasons that they initiate same-gender friendships such as companionship, the sharing of good times, conversation, and laughter. Men and women also seek similar traits such as honesty, intelligence, sensitivity, humor, and dependability in their cross-gender friends.

Among heterosexuals, especially men, romantic potential can also serve as a key factor motivating interest in cross-gender friendships (Bleske-Rechek & Buss, 2001; Bleske-Rechek et al., 2012; Fehr, 1995; Halatsis & Christakis, 2009). In fact, O’Meara (1989) identified the management of sexual attraction as one of the key challenges faced by cross-gender friends; this likely explains why unmarried people have more cross-gender friends than do married people (Booth & Hess, 1974; Rose, 1985). Still, Monsour, Harris, Kurzveil, and Beard (1994) found that only a minority of their participants reported that concern about sexual tension caused significant difficulties with their cross-gender friendships. Lenton and Webber (2006) indicate that people who perceive that cross-gender friendships have benefits beyond those offered by same-gender friendships have more cross-gender friends but suggest that people generally do not avoid cross-gender friendships because of their perceived detriments.

Other demographic variables

Most studies of friendship rely on small college samples (e.g., Bowman, 2008; Caldwell & Peplau, 1982) and community samples with limited generalizability (Felmlee & Muraco, 2009; Walker, 1994). In terms of scope, most studies forgo the use of age as a variable, instead recruiting participants from a single stage of the life course, such as young adulthood (Burleson, Holmstrom, & Gilstrap, 2005), middle age (Liebler & Sandefur, 2002), or later life (Felmlee & Muraco, 2009); such studies frequently focus on gender differences.

Other demographic variables like differences in marital or relationship commitment status (Cohen, 1992; Cornwell, Laumann, & Schumm, 2008), parental status, and social class (Walker, 1994) are less commonly examined. Only a small number of studies examine the effect of residential population size on friendship, with mixed findings of small magnitude (Fischer, 1982; Marsden, 1987).

Many of these demographic variables are widely used in social network research. For example, network research consistently demonstrates that higher education and income are associated with larger social networks (Brashears, 2011). Although these significant relationships are consistent within the research on broad social network size (which may include neighbors, acquaintances, and coworkers), we expect less variation by education and income since our investigation is confined to close personal friendships, which arguably serve a more essential social purpose that may transcend the effects of respondents’ levels of income and education.

Methodological challenges of studying friendship

Studying friendship quantitatively poses some particular methodological challenges, most notably, what counts as “friendship” varies tremendously. Depending upon context, the term “friend” can imply many types of relationships ranging from lifelong confidants to acquaintances linked only via social networking websites. The few nationally representative samples capable of examining variations across different demographic factors rely on only one or two items regarding overall number of friends (Marks, 1994; McPherson, Smith-Lovin, & Brashears, 2006). A careful look at the large-scale studies that have assessed friendship reveals that the primary focus is usually on social network size or broad sources of social support rather than on specific aspects of close friendship, including the distinction between cross-gender and same-gender friends (Cornwell et al., 2008; van der Poel, 1993).

We chose to focus on close friendships in particular because there is greater agreement about the reciprocal benefits and obligations associated with such relationships (Rubin, 1985) and because close friends have a greater impact on an individual’s everyday life (Muraco, 2012). Additionally, recent research suggests that the social support obtained through close friendships confers a number of psychological benefits that contribute to people’s sense of life satisfaction (Rook & Ituarte, 1999). This social support contributes to life satisfaction in a number of ways. First, friends provide emotional support, characterized by talking about personal problems and sharing information, guidance, and advice. Close friends also provide instrumental support, which entails tangible and practical social support (Taylor, 2010). Friendships also yield companionship, which is based on spending time with friends and mutually engaging in social activities (Demir & Weitekamp, 2007; Fischer, 1982). Each of these types of friendship support map onto the measures employed in this study; we tied our definitions of friendship to specific interactional contexts in which close friends could be relied upon for expressive, instrumental, and/or companionate support. Using more restrictive parameters not only enables clearer operational definitions but provides a better test of some previous researchers’ assertions that women’s friendships are more meaningful and intimate than are men’s.

To explore the different aspects of friendship, we included measures likely to assess expressive aspects of friendship, such as personal sharing, and instrumental aspects of friendship, such as ability to rely upon a person to provide tangible help. As noted above, previous research indicates that women’s friendships revolve around emotional expressiveness (Veniegas & Peplau, 1997; Wright, 2006) and the disclosure of intimate information (Braithwaite & Kellas, 2006), leading us to ask respondents the number of people with whom they could discuss a highly personal subject, namely their sex life. Researchers have categorized discussion of personal matters with friends, specifically sex-related topics, as being characteristic of close and emotionally intimate friendships (Aries & Johnson, 1983; Lefkowitz, Boone, & Shearer, 2004).

Some previous research indicates that the commemoration of birthdays is especially important among female office workers (Kanter, 1977) and that women are more likely than men to celebrate good news publicly with a close friend (Roy, Benensen, & Lilly, 2000), leading us to ask respondents how many people they would “expect to do something” with them to celebrate their birthday. Given that celebrating a friend’s birthday serves as a show of support and caring as well as an occasion for partaking in shared activities, we conceived of this measure as a hybrid, combining both expressive and instrumental aspects of friendship. Researchers have found that sharing positive events is an important component of companionate close friendship that enhances the emotional bond and is related to perceived mattering, the psychological tendency for an individual to feel that they matter to others (Gable, Gonzaga, & Strachman, 2006; Demir & Davidson, 2013).

We then sought to include a measure that might favor men’s presumed orientation toward instrumentality in friendships. Studies suggest that men’s friendships are largely associated with doing activities together (Auckett, Ritchie, & Mill, 1988), as well as doing things for people they care about (Cancian, 1987), and that older men provide more tangible help to nonkin than do older women (e.g., transportation and errands; Kahn, McGill, & Bianchi, 2011). Presuming that our respondents were capable of estimating their close friends’ willingness to provide these types of support, we asked respondents how many people they could call or text if they found themselves in trouble late at night. Research has highlighted availability of practical assistance as a defining feature of friendship (Fehr, 1995; Sias & Bartoo, 2007). Further, van der Poel (1993) found that many respondents are able to correctly judge whether associates would provide them with support.

Research purpose and questions

This research aims to make several important contributions to the literature on friendship. First, in addition to examining gender differences in the number of overall friends, we also examine number of same-gender versus cross-gender friendships, a question rarely investigated despite the importance of cross-gender friendships to many men and women in the contemporary U.S. (Bleske-Rechek & Buss, 2001). We can also determine whether women report greater satisfaction with their friendships, insofar as prior studies had mixed findings regarding this gender difference.

Second, the current study harnesses some of the strengths of a large sample that is not restricted to one limited community and examines variations in numbers of different types of friendship support across key demographic variables, including age, relationship, and marital status as well as parental status. With this examination, we can see the impact of gender, holding these other variables constant, and also see which other variables contribute to one’s number of close friendships and how these close friendships affect life satisfaction.

Structuring our analyses to be comparable to those of Fischer and Oliker (1983), we used our large and demographically diverse survey data set to examine these links. In contrast to Fischer and Oliker, we relied on more recent data obtained using the increasingly common mode of data collected via the Internet, examined a more geographically diverse sample, created more fine-grained age-based categories, examined a greater variety of demographic factors, and relied on specific interaction-based measures of friendship rather than subcategories of friendship (kin, nonkin, and coworkers). In particular, our large sample size enabled us to assess friendships among more specific groups excluded from Fischer and Oliker’s analysis (namely, single parents of all ages and unmarried adults aged 36–64). Further, we assessed not only number of friends but also satisfaction with friendships.

Research Question 1 (RQ1): Are there gender differences in number of friends, expressive versus instrumental forms of support, and satisfaction with friend relationships?

We examined how many close same-gender and cross-gender friends men and women have and how many total friends they have. We tested whether women would report having more same-gender friends for expressive support whereas men would report having more same-gender friends available to provide instrumental support.

Research Question 2 (RQ2): What are the other demographic predictors of number of friends and friendship satisfaction?

We examined the links between demographic factors (i.e., age, marital status, parental status, age of children, education, income, and residential population size), number of friends, and satisfaction with friends.

Research Question 3 (RQ3): What is the link between number of friends, satisfaction with friendships, and overall life satisfaction?

We examined whether people with more friends were more satisfied with their friendships. Further, we examined whether number of same-gender or cross-gender friends was a better predictor of this satisfaction for men and women. Finally, we examined the extent to which number of cross-gender and same-gender friends and satisfaction with friendships predicted how satisfied people are in general with their lives.

Method

Respondent recruitment

The present study is based on secondary analyses of anonymous data collected via a survey posted on a news website, formerly msnbc.com (now NBCNews.com), for 10 days in April of 2010. An invitation to participate in a survey appeared continually on the front page of the health section and periodically on the website home page. All respondents were provided with an opportunity to view the privacy agreement and asked for their birth year; those under age 18 were dismissed as too young to participate. Respondents were volunteers who received no compensation; to prevent the same individual from responding to the survey more than once, a software program denied multiple responses from any given computer. All respondents were asked a series of questions to assess their perceptions of life satisfaction and success, life’s stressors, and sources of social support and happiness.

Given that access to the Internet has grown remarkably in the last decade, the opportunity to participate in surveys such as this one is available to 95% of those between ages 18 and 29, 87% of those between 30 and 49, 78% of those between 50 and 64, and 42% of those 65 and older (U.S. Census Bureau, 2012a). Market research on NBCNews.com³ shows it ranks as one of the most popular nonpornographic websites in the U.S.; its 49 million unique monthly visitors are diverse in terms of age, income, and educational attainment (NBC News, 2012).

As consultants on this survey, as well as previous ones posted on this news website, the authors have access to the anonymous data sets, which have been used for nine peer-reviewed publications on a variety of topics—including examinations of close relationships (Galperin et al., 2013), workers’ attitudes toward female bosses (Elsesser & Lever, 2011), use of online dating sites and adult websites (Grov, Gillespie, Royce, & Lever, 2011), and elements of body dissatisfaction (Frederick, Peplau, & Lever, 2008; Peplau et al., 2009). For a detailed discussion of the advantages and disadvantages of Internet research, see Fraley (2007).

Variables and sample characteristics

The 25,185 respondents consisted of 12,654 (50%) men and 12,531 (50%) women. Respondent gender was coded as 0 = male and 1 = female for regression analyses. The average age for the sample was 42.3 (SD = 13.1; male M = 46.1, SD = 12.9; female M = 38.4, SD = 12.3). For the sake of comparison with Fischer’s and Oliker’s findings, preliminary analysis explored data using their age categories (under 36, 36–64, and 65+); however, in our tables we present more discreet subcategories: 18–24, 25–35, 36–49, 50–64, and 65+. An interval-level measure for age was used for the regression analysis.⁴ Based on the zip codes provided by respondents, at least 95% of our sample came from each of the 50 U.S. states (the rest left the item blank or lived abroad). The sample included 1,493 gay, lesbian, and bisexual respondents. Due to space constraints, we address the friendship patterns of nonheterosexual respondents in a separate paper.⁵

Satisfaction measures

Satisfaction with various aspects of people’s lives was assessed early in the questionnaire (Q3). Participants were presented with the following question stem “On a scale from 1 to 7, how dissatisfied or satisfied are you with your…” and then a series of items to which they could respond, including [Life overall] [Current job situation] [Health] [Relationships with your friends] [Relationships with your partner] [Relationships with your kids] among other sources of stress or support not included in our analyses. The scale ranged from 1 = very dissatisfied to 4 = neutral to 7 = very satisfied, with the option to mark “not applicable.” We note that the order of all answer options was randomized. So for some participants, partners and “kids” preceded friendship, while for other participants these items appeared in the reverse order. It is important to note, too, that two other items with randomized answer options preceded our core item on friendship support that included family members beyond one’s children.

In order to avoid overstating the role of friends in life satisfaction, in our model we also included measures for satisfaction with one’s health and current job situation (whereas we could not include partner and children because so many respondents lacked one, the other, or both). Therefore, we were able to determine whether friendship mattered above and beyond other important sources of life satisfaction. Including these additional measures also helped avoid inflated scores resulting from common method variance (Lucas, Dyrenforth, & Diener, 2008).

Number of friends and expressive versus instrumental forms of friendship

Same-gender and cross-gender friendships were assessed through three items. Respondents were presented with the question stem, “Thinking of your female and male friends, how many do you have that…” and then three items to which they could respond: [You could talk to about your sex life? (Expressive)], [You expect to do something with you to celebrate your birthday? (Expressive/Instrumental)], [You could call/text if you were in trouble late at night? (Instrumental)].

For each of these three items, respondents were presented with a drop-down menu for “number of female friends” and “number of male friends” separately with the following responses for each: 0, 1, 2, 3–5, 6–10, 11–15, 16–20, 21+. For the options with ranges, the respondent was given the midpoint score (and a score of 23 for the 21+ category).

For each variable, we also calculated the total number of friends by summing the maximum number of same-gender friends and cross-gender friends referred to by each respondent. Thus, we created variables for overall total number of friends, total number of same-gender friends, and total number of cross-gender friends. These variables were meant to represent a person’s highest total number of same- and cross-gender friends in any of the three variables assessed. For example, the following respondent would be classified as having 12 total close friends, 8 total same-gender friends, and 4 total cross-gender friends: friends they could talk to about sex life = 10 (8 same-gender, 2 cross-gender); friends they could celebrate birthday with = 8 (4 same-gender, 4 cross-gender); friends they could call if in trouble at night = 9 (8 same-gender, 1 cross-gender).

Control variables

Educational status was assessed using an ordinal measure. The percentage of the overall sample, men, and women, respectively, who fell in each category were high school or less (8%, 8%, 8%), some college or associate’s degree/trade school (31%, 29%, 33%), college graduate (31%, 31%, 32%), and advanced degree (30%, 32%, 27%). Overall, the sample has more highly educated people compared with the U.S. population, where only 20% have a bachelor’s degree and 11% have an advanced degree (U.S. Census Bureau, 2012b); however, there was sufficient diversity in educational status to give us the statistical power needed to take into account the impact of education at all levels. For regression analyses, the ordinal categories were coded 0–3 to indicate lower versus higher levels of education (high school through advanced degree).

Income was based on an ordinal measure of the respondent’s reported level of personal income. A total of 18 response categories ranged from 0 to 4,999 at the lower end through US$1 million or more at the upper end. The variable was recoded using the midpoint value for each income category (the amount of US$1,000,001 was used as the highest income). The overall median income was US$55,000, with the median for men being US$75,000 and the median for women being US$45,000. This is somewhat higher than national statistics, which place the median income for full-time working men at just under US$50,000 and for women at just under US$40,000 (U.S. Census Bureau, 2012d), which is not surprising given that the current sample includes relatively more individuals with an advanced degree compared with the general population.

Children were reported by 63% of the sample (men = 74% and women = 53%). Respondents were also asked to specify the ages of their children. The percentage of the overall sample, men, and women, respectively, who fell into each category were: has no children (37%, 27%, 47%), has children under the age of 6 (8%, 7%, 9%), has children over 6 (48%, 59%, 38%), and children under and over 6 years old (7%, 8%, 6%). For regression analyses, this variable was coded so that has no children = 0, allowing it to serve as the reference group.

Relationship status and level of relationship commitment were assessed. The percentages of the overall sample, men, and women, respectively, who fell in each category were single and not dating (14%, 11%, 17%), casually dating one or more people (7%, 5%, 8%), in a relationship but not cohabiting (12%, 8%, 15%), in a cohabiting relationship (9%, 6%, 11%), and married (60%, 70%, 49%). For regression analyses, the categories were coded 0–4 to indicate lower versus higher levels of formal relationship commitment (single through married).

Residential population size was measured using the respondent’s five-digit zip code. Zip codes were linked to Federal Information Processing Standards codes and associated with Beale ciphers and countywide population data from the 2000 U.S. Census. The resulting measure is based on the logged population of a respondent’s county. The log transformation was performed to avoid heteroskedasticity. Taking the logarithm of the variable also facilitates intuitive interpretation because a single unit difference is expected to be less consequential in larger populations. The overall mean population was 889,018 (SD = 1,534,722; Median = 426,526). The mean for men was 835,432 (SD = 1,442,926; Median = 381,751) and the mean for women was 942,765 (SD = 1,618,971; Median = 445,342).

Results

Suggested interpretation of effect sizes

Our large sample size provided the power to detect even miniscule associations in analyses conducted with the overall sample (e.g., differences as small as a few percentage points, weak correlations, and small effect size ds). Therefore, we emphasize the effect size for notable comparisons. We highlight Cohen’s d when comparing means, which assesses the size of the difference between two means in standard deviation units. Cohen (1988) suggested that, in general, d values of .20, .50, and .80 be considered small, moderate, or large (respectively) in social science research. These values correspond roughly to Pearson’s r values of .10, .30, and .50, respectively. To indicate the direction of the gender difference, positive d values indicate that men had more friends than women did, and negative d values indicated that women had more friends than men (e.g., d = −.30 would indicate that, on average, women were slightly more likely to have more friends than men).

Similarly, in regression analyses, small β values are statistically significant. Although there are no general conventions regarding what constitutes a small, medium, or large β value, we suggest that values below .10 are not meaningful even if statistically significant. That is, if being 1 SD above the mean on the predictor variable results in only an average increase of .10 SDs on the outcome variable, controlling for the other variables, the finding is not likely to be of much interest.

Although our overall sample is large, some subsets of the sample are small. For example, there are very few married women who are over 65 with no children. Data are not reported for cells in which there are fewer than 30 respondents due to the likely unreliability of results based on such small samples.

Overview of data analysis strategy and data presentation

We first examined whether there was an effect of respondent gender on the six outcome variables (i.e., number of same-gender and cross-gender friends that one can discuss sex life with, count on to attend one’s birthday, and can call/text when in trouble). We conducted a multivariate analysis of variance because these six dependent measures are intercorrelated, followed by an examination of the univariate results. However, the effect size (partial η²) is only .001—indicating that this statistically significant relationship was not likely to be meaningful. Therefore, rather than focusing on the omnibus tests, we focused our attention on comparisons of specific cells of interest and their effect sizes.

Gender differences in number of same-gender, cross-gender, and total friends are reported in Table 1, along with effect size d for key comparisons. These differences are reported for the overall sample and across different age-groups as well as across different operational definitions of friendship. In Table 2, these comparisons are further refined by examining the number of friends people have according to gender, age, marital status, and parental status.

Table 1.

Mean number of friends according to gender, age, and type of friendship support.

	Friends you could talk to about your sex life
	Same-gender friendships			Cross-gender friendships			Total friendships
Age-group	Men M (SD)	Women M (SD)	d	Men M (SD)	Women M (SD)	d	Men M (SD)	Women M (SD)	d
18–24	4.5 (4.8)	3.5 (3.0)	.27***	2.4 (3.2)	1.7 (2.4)	.26***	6.9 (6.9)	5.2 (4.6)	.31***
25–35	3.5 (4.0)	3.1 (2.6)	.11***	1.9 (2.8)	1.2 (1.9)	.30***	5.3 (5.9)	4.3 (3.8)	.22***
36–49	2.4 (3.3)	2.6 (2.4)	−.06**	1.5 (2.5)	1.1 (1.7)	.22***	3.9 (5.1)	3.6 (3.5)	.07**
50–64	1.9 (2.9)	2.1 (2.3)	−.08**	1.3 (2.3)	.8 (1.6)	.26***	3.2 (4.6)	2.9 (3.3)	.08**
65+	1.8 (2.9)	2.1 (3.1)	−.10	1.3 (2.6)	1.0 (3.0)	.11	3.1 (4.9)	3.1 (5.7)	.00
All	2.5 (3.4)	2.8 (2.6)	−.10***	1.5 (2.6)	1.1 (1.9)	.18***	4.0 (5.3)	3.9 (3.9)	.02*
	Friends you expect to do something with you to celebrate your birthday
	Same-gender friendships			Cross-gender friendships			Total friendships
	Men M (SD)	Women M (SD)	d	Men M (SD)	Women M (SD)	d	Men M (SD)	Women M (SD)	d
18–24	5.8 (5.3)	4.9 (4.2)	.19***	3.7 (4.1)	3.7 (3.7)	.00	9.5 (8.5)	8.7 (7.2)	.12*
25–35	4.3 (4.4)	4.2 (3.8)	.01	2.8 (3.4)	2.8 (3.2)	.00	7.0 (7.1)	7.0 (6.4)	.01
36–49	2.8 (3.7)	3.3 (3.3)	−.14***	2.0 (3.1)	1.9 (2.6)	.05*	4.8 (6.3)	5.1 (5.3)	−.06**
50–64	2.3 (3.3)	3.1 (3.4)	−.24***	2.0 (2.9)	1.6 (2.5)	.13***	4.3 (5.9)	4.8 (5.3)	−.08**
65+	2.5 (3.7)	3.6 (3.9)	−.28***	2.4 (3.4)	1.8 (3.4)	.18*	4.9 (6.8)	5.4 (6.6)	−.06
All	3.0 (3.9)	3.8 (3.7)	−.20***	2.2 (3.2)	2.4 (3.0)	−.04**	5.3 (6.6)	6.1 (6.1)	−.14***
	Friends you ould call/text if you were in trouble late at night
	Same-gender friendships			Cross-gender friendships			Total friendships
	Men M (SD)	Women M (SD)	d	Men M (SD)	Women M (SD)	d	Men M (SD)	Women M (SD)	d
18–24	4.9 (4.4)	3.6 (3.1)	.36***	2.8 (3.3)	2.8 (2.7)	.00	7.7 (6.9)	6.4 (5.0)	.22***
25–35	4.1 (3.8)	3.4 (2.8)	.23***	2.1 (2.6)	2.2 (2.6)	−.05	6.2 (5.7)	5.6 (4.8)	.12***
36–49	3.5 (3.5)	3.1 (2.7)	.10***	1.9 (2.6)	2.0 (2.3)	−.05*	5.3 (5.5)	5.1 (4.5)	.04
50–64	2.9 (3.1)	2.9 (2.6)	.00	1.8 (2.4)	1.8 (2.0)	−.01	4.7 (5.0)	4.7 (4.1)	−.01
65+	2.8 (3.0)	2.9 (3.0)	−.02	2.0 (2.6)	2.0 (3.2)	.00	4.8 (5.2)	4.9 (5.8)	−.01
All	3.4 (3.5)	3.2 (2.8)	.05***	1.9 (2.6)	2.1 (2.4)	−.08***	5.3 (5.5)	5.4 (4.7)	−.01

Note. Positive d values indicate that men had more friends than women.

*p < .05; **p < .01; ***p < .001 for corresponding t-tests comparing men with women.

Table 2.

Number of same-gender, cross-gender, and total friends by key demographic factors.

	Total number of friends
	Unmarried						Married
	Children			No children			Children			No children
	Men M (SD)	Women M (SD)	d	Men M (SD)	Women M (SD)	d	Men M (SD)	Women M (SD)	d	Men M (SD)	Women M (SD)	d
18–24
Same	–	4.9 (3.8)	–	7.5 (5.8)	6.2 (4.5)	.26***	–	3.9 (3.0)	–	6.35 (4.4)	4.2 (2.8)	.69***
Cross	–	4.2 (3.5)	–	4.7 (4.5)	4.7 (4.0)	.00	–	2.8 (3.0)	–	3.5 (4.2)	3.4 (2.8)	.05
Total	–	9.1 (6.0)	–	12.2 (9.2)	10.9 (7.6)	.16**	–	6.6 (5.6)	–	9.9 (8.1)	7.6 (5.1)	.40
N	–	83	–	533	1,234		–	61	–	37	175
25–35
Same	5.9 (5.6)	4.3 (3.1)	.40***	6.7 (5.1)	5.8 (4.1)	.20***	5.0 (4.3)	4.5 (3.3)	.14**	5.9 (4.8)	5.2 (3.8)	.17**
Cross	4.4 (4.8)	3.0 (3.1)	.39***	4.2 (4.0)	4.1 (3.7)	.03	2.9 (3.2)	2.6 (2.8)	.11**	3.5 (3.4)	3.5 (3.4)	.00
Total	10.3 (9.3)	7.4 (5.3)	.45***	10.9 (8.2)	9.9 (7.1)	.13***	7.9 (6.8)	7.1 (5.4)	.14***	9.4 (7.1)	8.7 (6.6)	.11
N	140	376		926	1,857		828	1,168		394	910
36–49
Same	4.8 (4.8)	4.4 (3.4)	.09	5.1 (5.0)	4.9 (3.8)	.06	4.6 (4.2)	4.3 (3.3)	.08**	4.3 (3.8)	4.4 (3.7)	−.03
Cross	3.5 (3.8)	2.9 (3.8)	.20***	3.5 (4.1)	3.2 (3.3)	.08	2.9 (3.5)	2.5 (2.8)	.12***	3.0 (3.3)	2.9 (2.9)	.04
Total	8.3 (7.7)	7.3 (5.5)	.15**	8.6 (8.1)	8.1 (6.5)	.08	7.5 (7.0)	6.8 (5.5)	.11***	7.3 (6.7)	7.3 (6.0)	.00
N	506	952		476	713		3,083	1,961		339	417
50–64
Same	4.0 (3.9)	4.5 (3.8)	−.15**	3.9 (4.1)	4.1 (3.2)	−.06	4.0 (3.9)	4.0 (3.2)	.01	3.8 (4.0)	4.0 (3.6)	−.06
Cross	3.1 (3.3)	2.7 (2.8)	.14**	2.7 (3.1)	2.4 (2.6)	.11	2.9 (3.4)	2.3 (2.6)	.18***	2.7 (3.2)	2.5 (2.9)	.05
Total	7.1 (6.7)	7.2 (6.0)	−.02	6.6 (6.7)	6.5 (5.3)	.01	6.9 (6.8)	6.3 (5.3)	.09**	6.5 (6.7)	6.6 (6.2)	−.01
N	697	712		256	315		3,157	1,120		333	226
65+
Same	4.1 (3.8)	4.6 (4.2)	−.13	4.2 (3.9)	–	–	4.2 (4.3)	4.1 (3.6)	.02	–	–	–
Cross	3.7 (3.5)	2.4 (3.4)	.39**	3.4 (2.8)	–	–	3.5 (4.1)	2.8 (3.6)	.17	–	–	–
Total	7.8 (6.3)	7.0 (7.2)	.13	7.6 (6.5)	–	–	7.7 (3.1)	6.9 (6.5)	.10	–	–	–
N	170	102		33	–	–	683	115		–	–
All ages
Same	4.5 (4.5)	4.5 (3.5)	.00	6.2 (5.3)	5.6 (4.2)	.12***	4.4 (4.1)	4.2 (3.3)	.04	4.7 (4.3)	4.7 (3.7)	.00
Cross	3.4 (3.7)	2.9 (2.9)	.18***	4.0 (4.1)	4.0 (3.7)	.00	3.0 (3.5)	2.5 (2.8)	.14***	3.1 (3.3)	3.2 (3.2)	−.04
Total	7.9 (7.3)	7.3 (5.7)	.09***	10.2 (8.5)	9.6 (7.2)	.07***	7.4 (7.1)	6.8 (5.5)	.09***	7.8 (7.0)	7.9 (6.3)	−.02
N	1,534	2,225		2,224	4,142		7,773	4,425		1,123	1,739

Note. Data not reported for cells with <30 respondents. Positive d values indicate that men had more friends than women. The mean number of total friends for unmarried respondents was 9.0 (SD = 7.3) compared with 7.3 (SD = 6.5) for married respondents (d = .25***). OLS = ordinary least square.

*p < .05; **p < .01; ***p < .001 for corresponding t-tests comparing men with women.

In Table 3, we report the link between demographic factors, including residential population size and total number of friends for men and women. In Table 4, we highlight how individuals report satisfaction with their friends according to gender, age, marital status, and parental status. Table 5 presents the results of a regression analysis examining the links among life satisfaction, same-gender and cross-gender friends, satisfaction with friends, and interactions between friendship quantity and quality, controlling for other variables.⁶ From the findings presented in these tables, we highlight the key patterns related to our primary research questions.

Table 3.

OLS regression examining predictors of total number of friends.

	Total number of friends
	Men β	Women β	Full sample β
Demographic predictors
Female			−.05***
Age	−.13***	−.18***	−.16***
Education level	−.00	.03**	.01
Personal income	.04***	.04***	.04***
Level of relationship commitment	−.04***	−.05***	−.04***
Child age categories (omitted:no child)
Child <6	−.04***	−.08***	−.06***
Child >6	−.06***	−.04***	−.06***
Under and over age 6	−.03**	−.06***	−.05***
Population size (logged)	.01	.05***	.02***
Constant	11.8***	9.1***	10.8***
Adjusted R ²	.03	.05	.04
N	11,409	11,375	22,784

Note. OLS = ordinary least square.

*p < .05; **p < .01; ***p < .001.

Table 4.

OLS regression examining predictors of satisfaction with friends.

	Satisfaction with friends
	Men β	Women β	Full sample β
Demographic predictors
Female			.07***
Age	.10***	.12***	.12***
Education level	−.01	.01	.00
Personal income	−.04***	−.03**	−.03***
Level of relationship commitment	−.02*	−.06***	−.04***
Child age categories (omitted:no child)
Child <6	−.02**	−.03***	−.03***
Child >6	−.03*	−.01	−.02*
Under and over age 6	−.04***	−.03**	−.03***
General satisfaction predictors
Job satisfaction	.21***	.17***	.19***
Health satisfaction	.28***	.21***	.24***
Friendship predictors
Total same-gender friends	.20***	.23***	.21***
Total cross-gender friends	.04***	.03**	.03***
Constant	2.42***	2.92***	2.6***
Adjusted R ²	.22	.18	.20
N	11,868	11,757	23,625

Note. OLS = ordinary least square.

*p < .05; **p < .01; ***p < .001.

Table 5.

OLS regression examining predictors of life satisfaction.

	Life satisfaction
	Men β	Women β	Full sample β
Demographic predictors
Female			.04***
Age	−.01	−.03**	−.02**
Education level	−.01	.03***	.01*
Personal income	.01	−.02*	−.00
Level of relationship commitment	.13***	.14***	.14***
Child age categories (omitted: no child)
Child <6	.00	.02*	.01
Child >6	−.01	.01	−.00
Under and over age 6	−.02	−.01	−.01*
General satisfaction predictors
Job satisfaction	.32***	.29***	.30***
Health satisfaction	.25***	.26***	.25***
Friendship predictors
Satisfaction with friends	.29***	.27***	.28***
Total same-gender friends	.02*	.01	.02*
Total cross-gender friends	.02*	.04***	.03***
Constant	.91***	1.5***	1.2***
Adjusted R ²	.44	.36	.40
N	11,852	11,740	23,592

Note. OLS = ordinary least square.

*p < .05; **p < .01; ***p < .001.

RQ1: Are there gender differences in number of friends, expressive versus instrumental forms of support, and satisfaction with friend relationships?

Number of friends

As shown in Table 1, men and women were remarkably similar in their number of same-gender, cross-gender, and total number of friends, regardless of the operational definition used. Overall, although t-tests sometimes indicated statistically significant differences (i.e., ***p < .001, **p < .01, *p < .05), the effect sizes were extremely small for gender differences in the number of friends people could talk with about their sex life (same-gender d = −.10***, cross-gender d = .18***, and total friends d = .02*), number of friends with whom they could expect to celebrate their birthday (same-gender d = −.20***, cross-gender d = −.04***, and total friends d = −.14*), and number of friends they could call or text if they were in trouble late at night (same-gender d = −.05***, cross-gender d = −.08***, and total friends d = −.01).

On average, men and women both reported approximately four friends they could talk to about their sex lives, five to six friends they could call when in trouble, and also five to six friends they could rely on for celebrating their birthday. Regardless of the operational definition of friendship, one can see from the rows for “all respondents” that both men and women reported more same-gender than cross-gender friends.

Gender differences were absent or very small across most age-groups reported in Table 1. Only one gender difference exceeded d = .30: men aged 18–24 had more same-gender friends they could call if they were in trouble late at night compared with women in the same age-group (d = .36***). The most consistent gender difference emerged for the question measuring the number of friends people can talk to about their sex lives. In almost all age categories, men reported more cross-gender friends with whom they can talk about their sex life than did women. The second most consistent gender difference was that women aged 36 and above reported having more same-gender friends who could be counted upon to celebrate their birthdays, compared with men in each older age-group. Even so, the big picture is that notable gender differences in friendships were mostly nonexistent, regardless of expressive or instrumental operational definition.

As shown in Table 2, regardless of age, marital status, and parental status, gender differences in friends were generally small to nonexistent. The notable exception to this pattern occurred among 25- to 35-year-old unmarried people with children. Unmarried men reported roughly three more friends than did unmarried women (d = .45***) and roughly 1.5 more same-gender friends (d = .40***) and cross-gender friends (d = .39***). That is, as one might expect, young single mothers appeared to have less time to invest in friendships. The second notable exception was that unmarried men older than 65 with children reported more cross-gender friends than did counterpart women (d = 39**).

Unlike some previous studies (see Gouldner & Strong, 1987), where answer options revealed researchers’ assumptions that all people have friends, we offered an opportunity for respondents to report zero friends in the various categories. The data reveal that virtually all men and women do think of themselves as having friends; only 4% of men and 2% of women reported having no friends for all three of the operational definitions of close friends.

Friendship satisfaction

Men and women were remarkably similar in their satisfaction with their friends. Overall, 72% of men and 75% of women were satisfied with their relationships with their friends, scoring above neutral on the 7-point Likert-type scale. As shown in Table 4, once demographic variables and other sources of satisfaction were controlled for, there was no noteworthy difference between men and women and their reported friendship satisfaction (β = .07***).

RQ2: What are the other demographic predictors of number of friends and friendship satisfaction?

Number of friends

In correlational results (not shown), there were only weak associations among number of friends and demographic variables such as age, education level, personal income, parental status (children under 6, over 6, or both under and over 6), and level of relationship commitment (all rs < .20). Similarly, the associations between these variables and satisfaction with friends were weak (all rs < .10).

As shown in Table 1, it appears that the most notable association is between age and number of friends. Younger respondents reported more same-gender friends than older respondents, and this difference was particularly pronounced among men (e.g., for number of friends men could talk about their sex lives with, the results were 4.5 friends for 18- to 24-year-olds and 3.5 for 25- to 35-year-olds, compared with 1.9 friends for 50- to 64-year-olds and 1.8 for 65+-year-olds). The results in Table 2 suggest that young men and women who were unmarried and childless tended to report the largest number of friends across the various measures of friendship. Young married men and women with children tended to report fewer friends than their unmarried counterparts and married counterparts with no children. As shown in Table 3, however, the overall associations of age, parental status, relationship status, and population size were only weakly related to number of friends, with age being the strongest of the predictors. These variables accounted for only 4% of the variance in the number of friends.

Friendship satisfaction

As shown in Table 4, regression analysis revealed that education level, income, and relationship involvement had no significant association with satisfaction with friends or that the associations were very weak. When controlling for these variables, however, age was the best demographic predictor of satisfaction with friends. Older individuals tend to be more satisfied with their friendships (β = .12***). Furthermore, when controlling for other important sources of satisfaction (i.e., health and job) both men and women with more cross-gender friends were not necessarily more satisfied with their friends, whereas both women and men with more same-gender friends reported being significantly and notably more satisfied with their friends (β = .21***). That is, there was no gender difference in this pattern regarding same- versus cross-gender friendships.

RQ3: What is the link between number of friends, satisfaction with friendships, and overall life satisfaction?

The average self-report of life satisfaction was 5.4 (SD = 1.4) for men and 5.4 (SD = 1.3) for women. Average satisfaction with friends was 5.3 (SD = 1.4) for men and 5.4 (SD = 1.5) for women. The correlation between satisfaction with friends and life satisfaction for men was .49***, and the relationship between satisfaction with friends and life satisfaction for women is .40***. As shown in Table 5, satisfaction with friends was a notable predictor of overall life satisfaction (β = .29*** for men and .28*** for women), even when controlling for demographic factors and other dimensions of satisfaction.

For men, the correlation between life satisfaction and number of same-gender friends was .14*** and cross-gender friends was .11***. For women, the correlation between life satisfaction and number of same-gender friends was .14*** and cross-gender friends was .11***. When controlling for other variables, the associations between number of same-gender and cross-gender friends and life satisfaction were statistically significant for men and women. The β values, however, were extremely small (β = .03 to .23; see Table 5).

Discussion

Strengths and limitations

Most previous studies of friendship were small in scale and narrow in scope, confined to a single geographic location or age cohort. We examined friendship using a large national sample that was sufficiently diverse on many demographic variables of interest and had a substantial number of respondents within each category to ensure robust between-group comparisons. By including less frequently studied demographic variables—such as age, marital status, and parental status—we were able to determine that gender differences in close friendship patterns are small across many different subgroups. Additionally, we distinguished between same-gender and cross-gender friends in our analyses and determined that, for the most part, people count on friends of both genders.

Perhaps most important, our study succeeded in avoiding common biases toward “female-typical” definitions of friendship that assess only personal disclosure and other intimate talk but neglect instrumental support as a critical aspect of friendship. We were able to assess both aspects of friendship because we had the opportunity to design multiple items related to our questions about friendship patterns, in contrast to many previous large-scale studies that relied on a single probe taken from a general survey. In so doing, we found that men and women were similar to each other in number of friends they have across different types of friendship support. These data challenge previous studies that concluded that men were more instrumentally inclined in their friendships and women were more emotive.

Other strengths of our research design should be noted. The results of the present study show that more fine-grained cohorts are needed when analyzing friendship across the life course. More than 20% of our very large sample would have been excluded from Fischer and Oliker’s (1983) original study because their research excluded young single parents and older unmarried individuals. Our findings may diverge from theirs, not only because 30 years have passed but also due to our inclusion of these additional categories of respondents and our use of a diverse, national sample that included nonurban residents (in contrast to Fischer and Oliker’s local urban sample).

Another difference between our design and Fischer and Oliker’s (1983) points to a limitation in our approach. Past research on adult friendships in the U.S. demonstrates that women are more likely to have their kin among their close friends, whereas men are more likely to have coworkers among theirs (Campbell, 1988; Fischer & Oliker, 1983; Marsden, 1987). Fischer and Oliker’s (1983) analysis differentiated between kin, nonkin, and coworkers, but we were unable to do likewise because our data set lacked these distinctions. Nonetheless, it is highly likely that our respondents ruled out both relatives and romantic partners for inclusion in their total count of friends, insofar as the wording of prior items asked them to think about friends separately from family members and partners.

Given women’s propensity to regard kin as close friends, this may have reduced the number of friends women reported in our study. Similarly, many previous studies of friendship do not distinguish between friends who are kin and those who are nonkin (Aries & Johnson, 1983; Caldwell & Peplau, 1982; Lenton & Webber, 2006); this can lead to undercounting of people who do double duty as both relatives and friends. Future researchers would be well advised to make clear distinctions in their definitions of types of friendships.

Many previous studies were designed such that they presumed that respondents have at least one friend and did not provide “zero” as an answer option. A notable strength of our study is that we made no such presumptions and provided respondents with the opportunity to indicate that they had zero friends in a particular category. We found strong evidence that virtually all respondents believed that they had at least one close friend; in fact, most of our respondents had at least two to six close friends. This is reasonably consistent with the recent estimates of number of close friends (Demir, Orthel, & Andelin, 2013; Demir & Özdemir, 2010). Thus, we were able to provide data in strong contrast to McPherson, Smith-Lovin, and Brashears’s (2006) controversial finding that about a quarter of respondents lacked even a single confidant, supporting the conclusion of Paik and Sanchagrin (2013) that McPherson’s finding was the result of methodological issues.

Relatedly, a possible limitation of our study is that our pull-down charts with numbers of friends of each gender in each category may have enabled some respondents to enter a rough estimate of their number of friends rather than accurately counting exactly how many of their friends fit that description. One might assume that the more common “name generation” method would yield a more precise and meaningful answer; however, commonly used name-generator items have recently been criticized for their methodological faults (Paik & Sanchagrin, 2013)—including respondent and/or interviewer fatigue—leading to erroneous results.

Our study suffered the same limitation that survey research often encounters; results were derived only from people who elected to participate in a research study. Nevertheless, despite being nonrepresentative, our sample was unusually large and diverse in terms of geographical region of origin, age, gender, relationship status, educational attainment, and income. Such an unusually large convenience sample was well suited to explore the links between demographic factors, number of friends, and friendship quality.

It is unclear how the sampling procedure used in this study would produce different associations among these variables compared to other recruitment methods. As is true of Internet samples in general, our collection of data from an Internet website yielded a sample of respondents with relatively higher incomes and educational levels when compared with the national population. We note that selection biases introduced by differential access to the Internet have been minimized as Internet use has grown more commonplace (Rainie & Horrigan, 2005; U.S. Census Bureau, 2012a).

We also note a very important difference between our sample and the ones often solicited on websites. Many Internet samples are taken from specialty websites that draw in visitors who share particular demographic factors or interests. We had access to the home page of a major and multifaceted website that draws visitors for diverse reasons including hard news, trending popular culture news, weather, sports, and financial updates. Enabling more robust comparisons between groups, our broad access yielded a sample that was more diverse with respect to gender, age, socioeconomic status, and geographic region than most Internet surveys; in fact, our sample was much more diverse than nonprobability samples generated by many of the more traditional data gathering methods (Gosling, Vazire, Srivastava, & John, 2004).

Finally, like most large-scale surveys, our survey elicited self-reports of various aspects of respondents’ friendships, yet provided no means by which researchers could cross-check the resulting data with the other half of any particular friendship dyad. Rubin’s (1985) study indicates that respondents may list individuals as close or even best friends, while those very same “best” friends may fail to share an identical perception of the depth or nature of the relationship. Possibly, then, if contacted, some of the friends enumerated by our respondents might dispute our respondents’ understanding of the relationships in question. However, van der Poel’s (1993) research gives us reason to suspect that friends would confirm mutual support.

Notable findings and concluding remarks

Several findings are worth highlighting. First, within-gender variation on our other demographic variables was more striking than across-gender variation. Overall, the average number of friends diminishes with age for both men and women. For every type of support (i.e., celebrating one’s birthday, talking about one’s sex life, and being able to call or text if in trouble late at night), the numbers were higher among emerging and young adults compared with middle-aged or older adults. Age-related decreases in number of friends can be attributed to a number of factors.

Among the elderly, decreases may be related to the death of close friends or health limitations that limit social participation and engagement in leisure activities (Rook, Charles, & Heckhausen, 2006). Among individuals in their 30s and 40s, this may be related to increased professional and familial obligations leaving less time and/or energy to maintain close friendships. Additionally, Carstensen (2006) found a “pruning effect” whereby individuals reduce their number of friends into smaller, closer networks starting in their 30s and 40s.

Although these differences are interesting, we remind the reader that the demographic variables represented in Table 3 accounted for only 4% of the variance in number of friends. We look to future researchers to examine other predictors—possibly including personality factors like self-esteem, extroversion, and identity formation—which might have more explanatory value when predicting number of friends (Demir, 2008; Diener, Suh, Lucas, & Smith, 1999).

People with children tended to have fewer friends, particularly for women and for people who have children both younger and older than 6 years of age. Thus, although many studies of friendship are content to confine their analyses to simple gender comparisons, our data, just like Fischer’s and Oliker’s (1983), indicate that demographic factors other than gender interacted with gender to account for far more variation in number of friends than gender alone. These findings underscore a point that feminist social scientists have driven home for years; researchers’ propensity for ferreting out gender differences obscures the extent to which differences are often accounted for by other factors (Epstein, 1988).

Second, our data support previous findings (Booth & Hess, 1974; Rose, 1985) that both married men and married women have more same-gender than cross-gender friends. We found this to be true by every measure of friendship support as well as in the total number of friends. In fact, this was also true for unmarried men and women who were in committed relationships relative to their “single” same-age counterparts. This lends some support to the notion that cross-gender friendships may pose a challenge to heterosexual relationships in the form of potential sexual attraction between heterosexual male and female friends (Halatsis & Christakis, 2009; O’Meara, 1989).

Third, our analyses were enriched by the inclusion of several life cycle categories missing from Fischer and Oliker’s original study (1983). Fischer and Oliker’s study presumed a traditionally linear progression by which a single (heterosexual) person first marries and then has children; unmarried parents of all ages were excluded from the analysis, as were unmarried adults aged 36–64. But a contemporary Census Bureau report on women’s fertility indicates that of women aged 15–44 who gave birth between June 2007 and June 2008, 37.5% were unmarried (Lewin, 2010). Our data pertaining to young unmarried parents suggest that single mothers have more friends than young married mothers. The Census Bureau also reported that 28% of those single mothers cohabited with their partners (Lewin, 2010); their friendship patterns may more closely resemble those of married women than of single women not cohabiting with a partner. Future friendship research should examine the effects of cohabiting with one’s partner among single mothers—our unmarried mother category did not distinguish between those cohabiting and not cohabiting.

Fourth, although we expected residential population size to matter, it did not. Despite the fact that people in smaller towns may know their friends and neighbors their whole lives, we nonetheless expected big city residents to have more opportunities to connect with a greater number of people who share their interests. Yet consistent with one of the few articles that incorporated city size as a variable (Marsden, 1987), our results did not show a meaningful relationship between residential population size and number of friends for either men or women.

Just as there were surprisingly only small differences by education and income, the lack of difference by population size may simply reflect the basic fact that humans are social beings. Other studies, using broader definitions of friends and social networks, may have found differences by core demographic variables. Insofar, as our operational definition got at the essence of very close friends, it may be that regardless of community size, education, and socioeconomic status most people will find a select few close friends who are socially and emotionally available to them and on whom they can depend.

Fifth, friendship quality is important for both men’s and women’s life satisfaction. This is consistent with research that has found that the quality of close friendships is equally important for women’s and men’s emotional well-being (Demir & Davidson, 2013).

Finally, and most surprising to us, was the virtual lack of difference between women and men. Women’s and men’s patterns were strikingly similar not only in the number and gender of their friends but also in the type of support they received from close friends. In fact, our data suggest that same-gender and cross-gender friendships defy gendered expectations of friend relationships. For example, women did not rely exclusively upon other women for intimacy or primarily upon men for instrumental help late at night; men did not rely only on women friends for intimate disclosure. Rather, for both men and women, same-gender friends were understood as being available to provide a variety of types of expressive and instrumental support.

We found it heartening that men had as many friends as women did with whom they could share intimacies and upon whom they could count to help them celebrate their birthdays, and we believe it reflects women’s great advances in social and economic independence that women reported having more same-gender than cross-gender friends they could rely on late at night to come to their rescue. This again suggests that previous research has overstated the degree to which men and women have distinct styles of friendship.

Of course, it is possible that our results reflect genuine patterns of cultural change. Three and a half decades ago, men’s close friendship was deemed “largely noticeable by its absence” (Levinson, 1978, p. 335), but today men benefit from a variety of types of same-gender friendships, including close friendships characterized by self-disclosure (Greif, 2009) as well as instrumentality. Perhaps, decades after the advent of second-wave feminism, contemporary masculinities better accommodate intimacy and the sharing of support in men’s same-gender friendships.

We end with a reminder that satisfaction with friends was a significant predictor of overall life satisfaction after controlling for demographic variables, number of friends, and satisfaction with other aspects of respondents’ lives. This is in line with existing research that found that friendship quality was more strongly associated with emotional well-being (i.e., life satisfaction) than friendship quantity (Demir, Orthel, & Andelin, 2013). In fact, satisfaction with friends was a better predictor of life satisfaction than was number of friends. In other words, where life satisfaction is concerned, our findings indicate that it is not how many friends a person has that matters most, but rather how happy one feels with the friendships he or she has that matters. Quality, or at least perceived quality, can trump quantity. Overall, the importance of friendship in our respondents’ lives underscores the need for further scholarship on friends’ unique contribution to people’s well-being. As Muraco (2012, p. 15) concludes, “friendship is one of the most significant, yet socially ignored, relationships.”

Footnotes

Acknowledgments

We thank ELLE magazine and NBCNews.com for access to data from the “ Sex, Stress and Success Survey,” Julia Sommerfeld and Ben Dickinson for their early editorial contributions, and Kimberly Steadman for information on website visitors. We also thank Judith Treas, Kim Elsesser, and the two anonymous reviewers for their helpful advice and comments on early versions of this manuscript. Janet Lever acknowledges the late Helen P. Gouldner for the inspiration to study friendship decades ago.

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

Notes

References

Aries

E. J.

Johnson

F. L.

(1983). Close friendship in adulthood: Conversational content between same-gender friends. Sex Roles, 9, 1183–1196.

Auckett

Ritchie

Mill

(1988). Gender differences in friendship patterns. Sex Roles, 19, 57–66.

Bleske-Rechek

A. L.

Buss

D. M.

(2001). Opposite-sex friendships: Sex differences and similarities in initiation, selection, and dissolution. Personality and Social Psychology Bulletin, 27, 1310–1323.

Bleske-Rechek

A. L.

Somers

Micke

Erickson

Matteson

Stocco

… Ritchie

(2012). Benefit or burden? Attraction in cross-gender friendship. Journal of Social and Personal Relationships, 29, 569–596.

Booth

Hess

(1974). Cross-gender friendship. Journal of Marriage and the Family, 36, 38–46.

Bowman

J. M.

(2008). Gender role orientation and relational closeness: Self-disclosure behavior in same-gender male friendships. The Journal of Men’s Studies, 16, 316–330.

Braithwaite

D. O.

Kellas

J. K.

(2006). Shopping with and for friends: Everyday communication at the shopping mall. In Wood

J. T.

Duck

(Eds.), Composing relationships: Communication in everyday life (pp. 86–95). Belmont, CA: Thomson/Wadsworth.

Brashears

M. E.

(2011). Small networks and high isolation? A reexamination of American discussion networks. Social Networks, 33, 331–341.

Burleson

B. R.

Holmstrom

A. J.

Gilstrap

C. M.

(2005). Guys can’t say that to guys: Four experiments assessing the normative motivation account for deficiencies in the emotional support provided by men. Communication Monographs, 72, 468–501.

10.

Cable

Bartley

Chandola

Sacker

(2013). Friends are equally important to men and women, but family matters more for men’s well-being. Journal of Epidemiology and Community Health, 67, 166–171.

11.

Caldwell

M. A.

Peplau

L. A.

(1982). Sex differences in same-gender friendship. Sex Roles, 8, 721–732.

12.

Campbell

K. E.

(1988). Gender differences in job-related networks. Work and Occupations, 15, 179–200.

13.

Cancian

F. M.

(1987). Love in America: Gender and self-development. Cambridge, UK: Cambridge University Press.

14.

Carstensen

L. L.

(2006). The influence of a sense of time on human development. Science, 312, 1913–1915.

15.

Cohen

(1988) Statistical power analysis for the behavioral sciences (2nd ed.). Hillsdale, NJ: Erlbaum.

16.

Cohen

T. F.

(1992). Men’s families, men’s friends: A structural analysis of constraints on men’s social ties. In Nardi

P. M.

(Ed.), Men’s friendships (pp. 115–131). Newbury Park, CA: Sage.

17.

Cornwell

Laumann

E. O.

Schumm

L. P.

(2008). The social connectedness of older adults: A national profile. American Sociological Review, 73, 185–203.

18.

Crosnoe

(2000). Friendships in childhood and adolescence: The life course and new directions. Social Psychology Quarterly, 63, 377–391.

19.

Demir

(2008). Sweetheart, you really make me happy: Romantic relationship quality and personality as predictors of happiness among emerging adults. Journal of Happiness Studies, 9, 257–277.

20.

Demir

Davidson

(2013). Toward a better understanding of the relationship between friendship and happiness: Perceived responses to capitalization attempts, feelings of mattering, and satisfaction of basic psychological needs in same-gender best friendships as predictors of happiness. Journal of Happiness Studies, 14, 525–550.

21.

Demir

Orthel

Andelin

A. K.

(2013). Friendship and happiness. In David

S. A.

Boniwell

Ayers

A. C.

(Eds.), The Oxford handbook of happiness (pp. 860–870). Oxford, UK: Oxford University Press.

22.

Demir

Özdemir

(2010). Friendship, need satisfaction and happiness. Journal of Happiness Studies, 11, 243–259.

23.

Demir

Weitekamp

L. A.

(2007). I am so happy ‘cause today I found my friend: Friendship and personality as predictors of happiness. Journal of Happiness Studies, 8, 191–211.

24.

Diener

Seligman

M. E. P.

(2002). Very happy people. Psychological Science, 13, 81–84.

25.

Diener

Suh

E. M.

Lucas

R. E.

Smith

H. L.

(1999). Subjective well-being: Three decades of progress. Psychological Bulletin, 125, 276–302.

26.

Elsesser

K. M.

Lever

(2011). Does gender bias against female leaders persist? Quantitative and qualitative data from a large-scale survey. Human Relations, 64, 1555–1578.

27.

Epstein

C. F.

(1988). Deceptive distinctions: Sex, gender and the social order. New Haven, CT: Yale University Press.

28.

Fehr

(1995). Friendship processes. Thousand Oaks, CA: Sage.

29.

Felmlee

Muraco

(2009). Gender and friendship norms among older adults. Research on Aging, 31, 318–344.

30.

Fiori

K. L.

Antonucci

T. C.

Cortina

K. S.

(2006). Social network typologies and mental health among older adults. Journal of Gerontology, Series B: Psychological Sciences and Social Sciences, 61, 25–32.

31.

Fischer

C. S.

(1982). To dwell among friends: Personal networks in town and city. Chicago, IL: University of Chicago Press.

32.

Fischer

C. S.

Oliker

S. J.

(1983). A research note on friendship, gender, and the life cycle. Social Forces, 62, 124–133.

33.

Fraley

R. C.

(2007). Using the Internet for personality research: What can be done, how to do it, and some concerns. In Robins

R. W.

Fraley

R. C.

Krueger

R. F.

(Eds.), Handbook of research methods in personality psychology (pp. 130–149). New York, NY: Guilford.

34.

Frederick

D. A.

Peplau

L. A.

Lever

(2008). The Barbie mystique: Satisfaction with breast size and shape across the lifespan. International Journal of Sexual Health, 20, 200–211.

35.

Fuller-Iglesias

H. R.

Webster

N. J.

Antonucci

(2013). Adult family relationships in the context of friendship. Research in Human Development, 10, 184–203.

36.

Gable

S. L.

Gonzaga

G. C.

Strachman

(2006). Will you be there for me when things go right? Supportive responses to positive event disclosures. Journal of Personality and Social Psychology, 91, 904–917.

37.

Galperin

Haselton

M. G.

Frederick

D. A.

Poore

von Hippel

Buss

D. M.

Gonzaga

G. C.

(2013). Sexual regret: Evidence for evolved sex differences. Archives of Sexual Behavior, 42, 1145–1161.

38.

Gosling

S. D.

Vazire

Srivastava

John

O. P.

(2004). Should we trust web-based studies? A comparative analysis of six preconceptions about Internet questionnaires. American Psychologist, 59, 93–104.

39.

Gouldner

Strong

M. S.

(1987). Speaking of friendship. New York, NY: Greenwood Press.

40.

Greif

G. L.

(2009). The buddy system: Understanding male friendship. Oxford, UK: Oxford University Press.

41.

Grov

Gillespie

B. J.

Royce

Lever

(2011). Perceived consequences of casual online sexual activities on heterosexual relationships: A U.S. Online survey. Archives of Sexual Behavior, 40, 429–439.

42.

Gundelach

Kreiner

(2004). Happiness and satisfaction in advanced European countries. Cross-Cultural Research, 38, 359–386.

43.

Halatsis

Christakis

(2009). The challenge of sexual attraction within heterosexuals’ cross-gender friendship. Journal of Social and Personal Relationships, 26, 919–937.

44.

Jones

D. C.

(1991). Friendship satisfaction and gender: An examination of sex differences in contributors to friendship satisfaction. Journal of Social and Personal Relationships, 8, 167–185.

45.

Kahn

J. R.

McGill

B. S.

Bianchi

S. M.

(2011). Help to family and friends: Are there gender differences at older ages? Journal of Marriage and Family, 73, 77–92.

46.

Kanter

R. M.

(1977). Men and women of the corporation. New York, NY: Basic Books.

47.

Kimmel

(2013). The gendered society (5th ed.). Oxford, UK: Oxford University Press.

48.

Lefkowitz

E. S.

Boone

T. L.

Shearer

C. L.

(2004). Communication with best friends about sex-related topics during emerging adulthood. Journal of Youth and Adolescence, 33, 339–351.

49.

Lenton

A. P.

Webber

(2006). Cross-gender friendships: Who has more? Sex Roles, 54, 809–820.

50.

Levinson

D. J.

(1978). The Seasons of a man’s life. New York, NY: Alfred A. Knopf.

51.

Lewin

(2010). Census finds single mothers and live-in partners. The New York Times. Retrieved from http://www.nytimes.com/2010/11/06/us/06moms.html?_r=0

52.

Liebler

C. A.

Sandefur

G. D.

(2002). Gender differences in the exchange of social supportwith friends, neighbors, and co-workers at midlife. Social Science Research, 31, 364–391.

53.

Löckenhoff

C. E.

Carstensen

L. L.

(2004). Socioemotional selectivity theory, aging, and health: The increasingly delicate balance between regulating emotions and making tough choices. Journal of Personality. Special Issue: Emotions, Personality, and Health, 72, 1395–1424.

54.

Lucas

R. E.

Dyrenforth

P. S.

Diener

(2008). Four myths about subjective well-being. Social and Personality Psychology Compass, 2, 2001–2015.

55.

Marion

Laursen

Zettergren

Bergman

L. R.

(2013). Predicting life satisfaction during middle adulthood from peer relationships during mid-adolescence. Journal of Youth and Adolescence, 42, 1299–1307.

56.

Marks

S. R.

(1994). Intimacy in the public realm: The case of co-workers. Social Forces, 72, 843–858.

57.

Marsden

P. V.

(1987). Core discussion networks of Americans. American Sociological Review, 52, 122–131.

58.

McPherson

Smith-Lovin

Brashears

M. E.

(2006). Social isolation in America: Changes in core discussion networks over two decades. American Sociological Review, 71, 353–375.

59.

Monsour

Harris

Kurzweil

Beard

(1994). Challenges confronting cross-gender friendships: “Much ado about nothing?” Sex Roles, 31, 55–77.

60.

Muraco

(2012). Odd couples: Friendships at the intersection of gender and sexual orientation. Durham, NC: Duke University Press.

61.

NBC News. (2012). Media kit. Retrieved from http://www.nbcnews.com/id/31066137

62.

O’Meara

J. D.

(1989). Cross-gender friendship: Four basic challenges of an ignored relationship. Sex Roles, 21, 525–543.

63.

Paik

Sanchagrin

(2013). Social isolation in America: An artifact. American Sociological Review, 78, 339–360.

64.

Peplau

L. A.

Frederick

D. A.

Yee

Maisel

Lever

Ghavami

(2009). Body image satisfaction in heterosexual, gay, and lesbian adults. Archives of Sexual Behavior, 38, 713–725.

65.

Perissinotto

C. M.

Cenzer

I. S.

Covinsky

K. E.

(2012). Loneliness in older persons: A predictor of functional decline and death. Archives of Internal Medicine, 172, 1078–1084.

66.

Pinquart

Sörensen

(2000). Influences of socioeconomic status, social network, and competence on subjective well-being in later fife: A meta-analysis. Psychology and Aging, 15, 187–224.

67.

Quantcast (2013). Traffic and demographic statistics for MSNBC. Retrieved from http://www.quantcast.com/msnbc.com

68.

Rainie

Horrigan

(2005). A decade of adoption: How the Internet has woven itself into American life. Retrieved from http://www.pewinternet.org/PPF/r/148/report_display.asp

69.

Reisman

J. M.

(1990). Intimacy in same-gender friendships. Sex Roles, 23, 65–82.

70.

Requena

(1995). Friendship and subjective well-being in Spain: A cross-national comparison with the United States. Social Indicators Research, 35, 271–288.

71.

Rook

K. S.

Charles

S. T.

Heckhausen

(2006). Aging and health. In Friedman

H. S.

Silver

R. C.

(Eds.), Handbook of health psychology (pp. 234–262). New York, NY: Oxford University Press.

72.

Rook

K. S.

Ituarte

P. H. G.

(1999). Social control, social support, and companionship in older adults’ family relationships and friendships. Personal Relationships, 6, 199–211.

73.

Rose

S. M.

(1985). Same- and cross-gender friendships and the psychology of homosociality. Sex Roles, 12, 63–74.

74.

Roy

Benenson

J. F.

Lilly

(2000). Beyond intimacy: Conceptualizing sex differences in same-gender friendships. The Journal of Psychology, 1, 93–101.

75.

Rubin

L. B.

(1985). Just friends: The role of friendship in our lives. New York, NY: Harper & Row.

76.

Ryle

(2011). Questioning gender: A sociological exploration. Thousand Oaks, CA: Pine Forge Press.

77.

Sabin

E. P.

(1993). Social relationships and mortality among the elderly. Journal of Applied Gerontology, 12, 44–60.

78.

Sias

P. M.

Bartoo

(2007). Friendship, social support, and health. In L’Abate

(Ed.), Low cost approaches to promote physical and mental health: Theory, research, and practice (pp. 455–472). New York, NY: Springer.

79.

Siebert

D. C.

Mutran

E. J.

Reitzes

D. C.

(1999). Friendship and social support: The importance of role identity to aging adults. Social Work, 44, 522–533.

80.

Taylor

S. E.

(2010). Social support: A review. In Friedman

H. S.

(Ed.), Oxford handbook of health psychology (pp. 189–214). New York, NY: Oxford University Press.

81.

U.S. Census Bureau. (2012a). Digital nation: Expanding Internet usage. Retrieved from http://www.ntia.doc.gov/files/ntia/publications/ntia_internet_use_report_february_2011.pdf

82.

U.S. Census Bureau. (2012b). Educational attainment of the population 25 years and over, by selected characteristics. Retrieved from http://www.census.gov/hhes/socdemo/education/data/cps/2011/tables.html

83.

U.S. Census Bureau. (2012c). Households and families: 2010: 2010 census briefs. Retrieved from http://www.census.gov/prod/cen2010/briefs/c2010br-14.pdf

84.

U.S. Census Bureau. (2012d). Income, poverty, and health insurance coverage in the United States. Retrieved from http://www.census.gov/prod/2012pubs/p60-243.pdf

85.

van der Horst

Coffe

(2012). How friendship network characteristics influence subjective well-being. Social Indicators Research, 107, 509–529.

86.

van der Poel

M. G. M.

(1993). Delineating personal support networks. Social Networks, 15, 49–70.

87.

van Diepen

A. M. L.

Mulder

C. H.

(2009). Distance to family members and relocations of older adults. Journal of Housing and the Built Environment, 24, 31–46.

88.

Veniegas

R. C.

Peplau

L. A.

(1997). Power and the quality of same-gender friendships. Psychology of Women Quarterly, 21, 279–297.

89.

Walker

(1994). Men, women, and friendship: What they say, what they do. Gender and Society, 8, 246–265.

90.

Wright

P. H.

(1988). Interpreting research on gender differences in friendship: A Case for moderation and a plea for caution. Journal of Social and Personal Relationships, 5, 367–373.

91.

Wright

P. H.

(2006). Toward an expanded orientation to the comparative study of women’s and men’s same-gender friendships. In Dindia

Canary

D. J.

(Eds.), Sex differences and similarities in communication (2nd ed., pp. 37–57). Mahwah, NJ: Lawrence Erlbaum Associates.