Optimal Social-Networking Strategy Is a Function of Socioeconomic Conditions

Abstract

In the two studies reported here, we examined the relation among residential mobility, economic conditions, and optimal social-networking strategy. In Study 1, a computer simulation showed that regardless of economic conditions, having a broad social network with weak friendship ties is advantageous when friends are likely to move away. By contrast, having a small social network with deep friendship ties is advantageous when the economy is unstable but friends are not likely to move away. In Study 2, we examined the validity of the computer simulation using a sample of American adults. Results were consistent with the simulation: American adults living in a zip code where people are residentially stable but economically challenged were happier if they had a narrow but deep social network, whereas in other socioeconomic conditions, people were generally happier if they had a broad but shallow networking strategy. Together, our studies demonstrate that the optimal social-networking strategy varies as a function of socioeconomic conditions.

Keywords

residential mobility social ecology social network cross-cultural differences well-being computer simulation sociocultural factors friendship

Like the reproductive strategies of plants and animals (Mayr, 1997; Tutin, 1979), human social-networking strategies primarily fall into two categories: (a) a broad, shallow strategy, in which individuals spread their time and resources widely across social connections and (b) a narrow, deep strategy, in which individuals concentrate their time and resources on a few very close friends or relatives. In the age of Facebook, many Americans seem to opt for a broad, shallow networking strategy. Yet, cross-cultural research has shown that having many friends is not always viewed positively outside of the United States. In Ghana, for instance, an individual who claimed to have more than 50 friends was considered “foolish” or “naive” (Adams & Plaut, 2003). In the present studies, we investigated the question of optimal social-networking strategy from a socioecological perspective (Oishi & Graham, 2010; Schug, Yuki, & Maddux, 2010). We asked in which socioeconomic contexts it is more advantageous to have narrow, deep ties versus broad, shallow ties.

In his classic book “Getting a Job,” sociologist Mark Granovetter (1974) documented the advantage of having multiple weak ties in job searches. In his study, the majority of individuals who recently found a job got the lead about the job opening from someone they met approximately only once a year. Granovetter found that having many acquaintances increased the chances of getting unique information on job opportunities. Most amazingly, individuals who got their job through acquaintances were making more money and were more satisfied with their job than those who got the job through close friends or relatives. Granovetter (1973) showed the functional value of having broad, shallow ties, and he called this phenomenon “the strength of weak ties.”

In addition to the functional value of broad, shallow ties, the American preference for a large social network may be explained by Americans’ high residential mobility (Oishi, 2010). Roughly half of Americans move their residence in any 5-year period (Schmitt, 2001). In a highly mobile society, it is difficult for individuals to maintain the same close friends over time because they or their friends are likely to move within a short period. To minimize the damage of a friend’s departure to a person’s social network, then, it is important to spread one’s time and resources among many friends instead of putting all one’s eggs in one basket. In other words, Americans might typically prefer a broad, shallow networking strategy in part because they intuitively know that they will lose some of their friends when they or their friends move away.

Another important ecological factor pertaining to friendship is economic conditions. In economically favorable conditions, under which most individuals lead a comfortable, self-sufficient life, friendships do not entail serious, high-cost helping such as covering a hospital bill and babysitting. In unfavorable economic conditions, under which individuals often need serious practical and material help from others, having a large number of friends might drain all of your time and resources and indeed be as “foolish” as Ghanaians perceive such a friendship to be (Adams & Plaut, 2003). In unfavorable economic conditions, then, having a small but deep social network should be advantageous because costly help can be expected from, and offered to, only a close friend or relative.

In short, we predicted that a broad, shallow networking strategy is optimal in a residentially mobile, economically favorable context, whereas a narrow, deep networking strategy is optimal in a residentially stable, economically unfavorable context. We conducted two studies to test this prediction, using two divergent methods—a computer simulation modeling the actions of hypothetical “agents” (Smith & Conrey, 2007) and an online survey of American adults living in diverse socioeconomic conditions (Buhrmester, Kwang, & Gosling, 2011).

Study 1: Computer-Simulation Study

We created an agent-based model because we were interested in a complex interaction among different levels of analysis (i.e., benefits individuals receive from their social network under various socioeconomic conditions; Latané & Bourgeois, 2001; Smith & Conrey, 2007). Multi-agent simulations have been used successfully to study complex social processes, such as social influence (Nowak, Szamrej, & Latané, 1990), formation of romantic relationships (Kalick & Hamilton, 1986), and residential segregation (Schelling, 1971). We manipulated the level of residential mobility and crisis probability (i.e., favorable or unfavorable economic conditions) at the level of community to identify the socioeconomic conditions in which it is more advantageous to have narrow, deep ties than broad, shallow ties.

Method

In Excel 2007, we created 1,000 “agents” and modeled these agents’ social networks at two times. At Time 1, each agent had a randomly generated number of friends. To simulate individual differences in the number of friends in the real world, we varied the number of friends across agents, with a normal distribution around 6 for very close friends, 10 for close friends, and 30 for distant friends (these numbers were derived from behavioral data; Oishi, Kesebir, et al., 2012). To model the level of time and resources required for each type of friend, we assumed that the three types of friends required different levels of investment from the agent, which we quantified as 5 points for very close friends, 3 points for close friends, and 1 point for distant friends.

By Time 2, two aspects of the agent’s world were subject to change. The size of each of these changes was controlled by a parameter in the simulation that we could manipulate. The first aspect was that between Time 1 and Time 2, some of the agents’ friends could have moved away, as determined by the parameter mobility rate. Mobility rate was a number between 0 and 1 that represented the percentage of an agent’s friends who would have left by Time 2.

Our simulation was set up such that some loss of benefits from departed friends would be compensated for by benefits coming from new friends by Time 2. The benefits from friends made between Time 1 and Time 2 was set to be proportional to the number of an agent’s friends who moved away because (a) mobility rate should also index the number of newcomers to the agents’ world with whom friendships could be formed and (b) agents who lost friends would likely be motivated to replace them with new ones. However, the amount of benefits from new friends was set to be smaller (80%) than what old friends would be providing had they not left, because it takes time for new relationships to form and grow.

The second important factor determining the utility of various networking strategies is the frequency and magnitude of help that an agent needs. In our simulation, this was reflected in the crisis-probability parameter, which we could manipulate. The crisis-probability parameter was a number between 0 and 1, and it corresponded to the probability of an economic crisis occurring between Time 1 and Time 2, which would hurt the agent unless close friends or relatives could shield the agent from the negative effects of the crisis.

Each agent’s total investment in friends at Time 1 was computed by adding up the products of the number of friends of each type and the amount of fixed investment in that type of friend (i.e., 5, 3 or 1). For example, an agent with 2 very close, 10 close, and 25 distant friends would have a total friendship investment of 65 ([2 × 5] + [10 × 3] + [25 × 1] = 65).

To capture the degree of investment in deep ties, we computed the deep-tie index by dividing the number of points invested in very close friends by the sum of points invested in close friends and points invested in distant friends. The higher this index, the more an agent invested in deep ties relative to weak ties.

Payoff was a measure of the agent’s gain from the social network (i.e., the benefits drawn from friends) at Time 2. It was calculated using a combination of three components. The first component was the investment coming from friends who the agent had at Time 1 and who were still around by Time 2. According to equity theory (Hatfield, Walster, & Berscheid, 1978), agents’ friends on average invest at the same rate as agents themselves do (e.g., 3 points for close friends). The second component was the investment coming from friends who were made between Time 1 and Time 2. The payoff from new friends is given by the formula (1 + 2 × [random number between 0 and 1]) × ([number of friends who left before Time 2] × .8). The first product term in this formula determines the average payoff from new friends. It was set such that the number would range between 1 and 3, with an average of 2. This was roughly equivalent to the average payoff an agent had from all types of friends at Time 1. The multiplication by .8 represented the 20% loss of benefits due to time lags in making new friends and developing them to the level of older friendships.

The third component of payoff at Time 2 was loss from a crisis against which the agent could not be buffered through very close friends or relatives. When a crisis happened, the simulation compared the proportion of an agent’s very close friends who were not gone at Time 2 with the total number of remaining friends. If this proportion was larger than .1, the agent came out of the crisis unscathed. But if the proportion was smaller than .1, the agent lost 5 points.

Because the payoff index did not take into account the degree of investment in friends (i.e., the cost of friendship), the ultimate index for the agent’s well-being was the payoff: investment ratio. This was the ratio of an agent’s payoff at Time 2 to the agent’s investment at Time 1. It is the percentage of one’s initial investment into friends that was transformed into a gain at Time 2. The higher this number, the higher the benefits an agent got from friends relative to the initial investment. All the following results are based on 50 runs of the simulation for each parameter combination. We crossed five levels of residential mobility (0%, 10%, 20%, 30%, 40%) with five levels of crisis probability (0%, 20%, 40%, 60%, 80%), corresponding to 1,250 total runs of the simulation.

Results and discussion

Our central question in Study 1 concerned when a narrow, deep social network is advantageous and when it is not. Thus, the main outcome variable of interest was the degree of correlation between the deep-tie index and the payoff:investment ratio. A positive correlation means that narrow, deep ties are more advantageous than broad, shallow ties: The more investment in deep relative to weak ties, the higher the proportion of an agent’s payoff from friends relative to the initial investment. In contrast, a negative correlation indicates that investing relatively more in deeper than in weaker ties will generate a lower return on an agent’s friendship investment.

Figure 1 presents the key correlations between investment in deep ties in the social network and payoff:investment ratio as a function of crisis probability and residential mobility. As Figure 1 shows, when mobility was low (10%) and the probability of crisis was high (.8), as in Ghana, agents with narrow, deep ties were less affected by a crisis than agents with broad, shallow ties (r = .17, 99% confidence interval, CI = [.16, .18]). In contrast, agents were less affected by a crisis if they had broad, shallow ties when mobility was high (40%) and the probability of crisis was low (.2; r = −.12, 99% CI = [−.13, −.11]), when mobility was low (10%) and the probability of crisis was low (.2; r = −.02, 99% CI = [−.03, −.01]), and when both mobility and the probability of crisis were high (40% and .8, respectively; r = −.06, 99% CI = [−.07, −.04]).

Fig. 1.

Results from Study 1: correlation between the deep-tie index and the payoff:investment ratio as a function of the probability of an economic crisis occurring between two time points and residential mobility (the percentage of friends who moved away between those two time points).

When mobility was set to 0 and the probability of crisis was set to a number other than 0 in the simulation, there was a positive correlation between deep ties and the payoff:investment ratio. That is, if one lives in a completely stable society, it is better to have a social network with narrow, deep ties. Furthermore, when mobility was set to 0, the higher the crisis frequency, the better it was to have more narrow, deep ties. For a crisis probability of 20%, the correlation coefficient between deep-tie index and payoff:investment ratio equaled .27 (99% CI = [.26, .28]). The coefficient was .38 (99% CI = [.38, .39]) for a crisis probability of 40%, .48 (99% CI = [.48, .49]) for a crisis probability of 60%, and .57 (99% CI = [.57, .58]) for a crisis probability of 80%.

In reality, however, there are hardly any societies without residential mobility. When we set the mobility parameter to 10%, the correlations between the deep-tie index and the payoff:investment ratio became negative for low levels of crisis frequency (see Fig. 2). In contrast, the correlations were still positive for high levels of crisis probability. If the mobility rate was raised to a relatively high level of 20% (which is roughly equivalent to the annual American residential mobility), the relationship between the deep-tie index and the payoff:investment ratio became negative for all but high levels of crisis probability. When the mobility rate was set to 30% or 40%, the association between deep ties and the payoff:investment ratio was negative for any level of crisis frequency.

Fig. 2.

Results from Study 1: payoff:investment ratio as a function of the deep-tie index, residential mobility (the percentage of friends who moved away between two time points), and the probability of an economic crisis occurring between those two time points.

In sum, the simulation indicated that having broad, shallow ties is more advantageous in highly mobile environments regardless of crisis proneness, whereas having narrow, deep ties is more advantageous in residentially stable environments, particularly if crisis probability is high.

Study 2: Friendship Strategy and Subjective Well-Being in a Community Sample

The computer simulation that we used in Study 1 points to the specific contexts in which one type of networking strategy is superior to another. However, a simulation only reveals what should happen in a simplified world in which all our assumptions would hold true. It does not guarantee that it is what actually happens in the real world, where our assumptions may not hold (e.g., there are some long-term close friendships that no longer require much investment) and nonmodeled variables (e.g., age of the agents, local friendship norms) may complicate predicted relations between variables. This creates the need to match our results to behavioral data (Smith & Conrey, 2007). Therefore, we conducted Study 2 to test whether the results from the simulation would indeed be observed in real life. As an outcome measure, we chose subjective well-being because it captures how well or poorly people think their life is going and how much positive and negative emotion they feel (Diener, Suh, Lucas, & Smith, 1999). We predicted that a more adaptive networking strategy would lead to higher life satisfaction and the experience of positive emotions.

Method

Participants in Study 2 were 247 Americans (108 men, 139 women; 184 European Americans, 16 African Americans, 15 Hispanic Americans, 28 Asian Americans, and 4 of other ethnicities; mean age = 31.11 years, SD = 11.81 years, range = 15–81 years) who were recruited through Amazon’s Mechanical Turk. The data were collected in June 2011. Of the 247 participants, 222 provided zip codes that we were able to match with Census 2000 data (U.S. Census Bureau, 2000). Participants completed a brief survey that was made to parallel the computer simulation described previously. Namely, participants listed three types of friends: very close (defined to participants as “someone you feel very close to, so close that it would be hard to imagine life without”), close (defined as “not quite as close as those in the inner circle, but still close”), and distant (defined as people the participant feels “less close to, but who are still important”). To make these types of friends vivid, we asked participants to list the initials of one friend of each type. Then, we told them to imagine that their time, energy, and money were limited to 60 points, and we asked them to describe how they would distribute this finite resource (60 points) to their three types of friends. As in the computer-simulation study, we created the index of deep ties using the following formula: points given to very close friends divided by the sum of points given to close friends and points given to distant friends.

We assessed participants’ cognitive aspects of subjective well-being using the Satisfaction With Life Scale (Diener, Emmons, Larsen, & Griffin, 1985; α = .92), and we assessed emotional aspects of subjective well-being by asking participants how often they had felt three positive emotions (happy, excited, and content; α = .82) and four negative emotions (sad, angry, bored, and worried; α = .83) in the month previous to the study. Participants rated these emotions on 7-point scales from 1 (not at all) to 7 (a lot). To gauge participants’ socioeconomic environments, we asked them to list the zip code of their current residence. We then obtained information regarding each zip code via Census 2000. As in previous research (Oishi, Miao, Koo, Kisling, & Ratliff, 2012; Oishi et al., 2007), we calculated the residential mobility of the population in each zip code using the following formula: the number of individuals who were not living in the current house in that zip code in 1995 divided by the number of total residents in that zip code in 2000. We also obtained the median family income for each zip code from Census 2000. The median family income is the conceptual equivalent of the crisis-probability factor simulated in Study 1, with lower family income corresponding to a higher crisis probability.

Results and discussion

We first formed the latent subjective well-being factor using three indicators: life satisfaction, positive affect, and the lack of negative affect (calculated by subtracting the mean negative affect from 7; higher numbers indicated the relative absence of negative affect). Using Mplus (Version 4.2; Muthén & Muthén, 2007), we then predicted latent subjective well-being from participants’ deep-tie index (z scored), the residential mobility of the population in their current zip code (z scored), the median family income in their current zip code (z scored), and the two- and three-way interactions of these variables, as well as control variables (gender, age, race). The fit of the model was acceptable, χ²(20) = 33.76, p = .01, comparative-fit index = .92, root-mean-square error of approximation = .06, standardized root-mean-square residual = .025.

We found the expected significant three-way interaction among residential mobility, median income, and networking strategy, b = 0.40 (95% CI = [0.03, 0.77]), β = 0.22, z = 2.11, p < .05. A simple-slopes analysis (Aiken & West, 1991) showed that, consistent with the simulation in Study 1, participants who had a narrow, deep friendship strategy had higher levels of subjective well-being than did participants who had a broad, shallow friendship strategy in a residentially stable and poor zip code (i.e., 1 SD below the mean in residential mobility and 1 SD below the mean in median family income, respectively), b = 0.94, SE = 0.43, t(211) = 2.20, p < .05 (see Fig. 3). In contrast, in the three other socioeconomic conditions (i.e., high mobility and rich zip code, high mobility and poor zip code, and low mobility and rich zip code), the broad, shallow strategy was positively associated with subjective well-being. The slope for the residentially stable and poor zip code was significantly different from all three other slopes, ts(211) > 2.59, ps < .02.

Fig. 3.

Results from Study 2: latent subjective well-being as a function of the deep-tie index, residential mobility of friends, and median family income in the participant’s zip code (low = 1 SD below the mean; high = 1 SD above the mean).

In addition to the three-way interaction, we found a significant interaction between networking strategy and median family income, b = −0.52 (95% CI = [−0.96, −0.07]), β = −0.28, z = −2.29, p < .05. A simple-slopes analysis revealed that participants living in a poor zip code were happier if they had a narrow, deep networking strategy than if they had a broad, shallow strategy, b = 0.60, t(211) = 2.98, p < .01. Conversely, people living in a rich zip code were happier if they had a broad, shallow friendship strategy than if they had a narrow, deep friendship strategy, b = −0.44, t(211) = −2.09, p < .05.

We also found a marginally significant two-way interaction between networking strategy and residential mobility, b = −0.28 (95% CI = [−0.62, 0.05]), β = −0.28, z = −1.66, p = .10. People living in a residentially stable zip code tended to be happier if they had a narrow, deep friendship strategy than a broad, shallow friendship strategy, b = 0.35, t(211) = 1.75, p = .08, whereas people living in a residentially mobile zip code did not show such a pattern, b = −0.21, t(211) = −0.95, n.s. Gender, age, and race were all nonsignificant factors, zs < |0.75|, ps > .44. Similarly, there were no main effects of residential mobility, median family income, or networking strategy, zs < |0.48|, ps > .64.

General Discussion

In the studies reported here, we used two divergent methods to investigate the optimal networking strategy in various socioeconomic conditions. In Study 1, we utilized an agent-based computer simulation (Latané & Bourgeois, 2001; Smith & Conrey, 2007). Consistent with previous cross-cultural research (Adams & Plaut, 2003), our results showed that a narrow, deep networking strategy, in which individuals invest their time and resources in a small number of very close friends, yielded a better payoff on that investment (i.e., allowed people to recover more quickly from an economic crisis) in a residentially stable, crisis-prone context (like the situation in Ghana) than did a broad, shallow networking strategy. In other socioeconomic conditions (i.e., those in which the population is residentially mobile and crisis probability is high, the population is residentially mobile and crisis probability is low, and the population is residentially stable and crisis probability is low), however, a broad, shallow friendship strategy had a better payoff.

In Study 2, we collected data from American adults living in diverse zip codes. Consistent with the results of the computer simulation, findings in Study 2 showed that Americans living in a residentially stable, poor zip code were happier when they had a narrow, deep networking strategy than when they had a broad, shallow networking strategy. Again consistent with the results of the computer simulation, findings indicated that Americans living in other socioeconomic conditions were happier if they had a broad, shallow networking strategy than if they had a narrow, deep networking strategy. Study 2 also demonstrated that a narrow, deep networking strategy is generally better in a residentially stable or in an economically impoverished community.

There are two important implications of the current findings for the literature on culture, close relationships, and well-being. First, our findings suggest that the previously reported cross-cultural differences in the nature of friendship (e.g., Adams & Plaut, 2003) could be partly explained by socioeconomic conditions such as residential mobility and economic factors. Although Ghana and the United States are very different in terms of culture, climate, economy, and politics, we were able to identify zip codes within the United States where a Ghanaian-style narrow, deep networking strategy was indeed optimal. Thus, cross-cultural variation in interpersonal relationships observed in the past (e.g., Ghana vs. the United States, India vs. the United States) might not be categorical. Instead, a shift in socioeconomic conditions might change Ghanaian and Indian networking strategies over time. In general, in economically challenging conditions, one’s chance of obtaining major help is higher if one has deep ties with a small number of friends than if one has shallow ties with a large number of friends. Likewise, if the community is residentially stable, friends are there for you forever. In such a condition, investing in a small number of friends seems sensible and adaptive. It is, however, critical in the future to test whether our findings will replicate in diverse developmental and cultural contexts.

Second, it is interesting to note that evolutionary psychologist David Buss (2000) idealized narrow, deep ties and viewed them as the basis of human happiness. Indeed, until recently, people lived under economically challenging and residentially stable conditions. Having narrow, deep ties was optimal in such an environment, and a narrow, deep networking strategy thus might have been the default social-networking strategy for a long time. The optimal social-networking strategy, however, might have changed with changing socioeconomic conditions.

Despite limitations (e.g., assumptions in the computer simulation, behavioral data only from the United States, a sample of mostly younger adults), our studies paint a clear picture that socioeconomic factors such as residential mobility and economic security play an important role in determining the most adaptive networking strategy. As residential mobility decreases and economic recession deepens in the United States, the optimal social-networking strategy might shift from the broad but shallow to the narrow but deep, even in a nation known best for the strength of weak ties.

Footnotes

Acknowledgements

We thank Casey Eggleston, Minha Lee, Thomas Talhelm, Felicity Miao, Marguerite Beattie, Yuxin Wang, and Jueyu Wu for their invaluable comments.

Declaration of Conflicting Interests

The authors declared that they had no conflicts of interest with respect to their authorship or the publication of this article.

References

Adams

Plaut

V. C.

(2003). The cultural grounding of personal relationship: Friendship in North American and West African worlds. Personal Relationships, 10, 333–347.

Aiken

L. S.

West

S. G.

(1991). Multiple regression: Testing and interpreting interactions. Newbury Park, CA: SAGE.

Buhrmester

Kwang

Gosling

S. D.

(2011). Amazon’s Mechanical Turk: A new source of inexpensive, yet high-quality, data? Perspectives on Psychological Science, 6, 3–5.

Buss

D. M.

(2000). The evolution of happiness. American Psychologist, 55, 15–23.

Diener

Emmons

R. A.

Larsen

R. L.

Griffin

(1985). The Satisfaction With Life Scale. Journal of Personality Assessment, 49, 71–75.

Diener

Suh

E. M.

Lucas

R. E.

Smith

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

Granovetter

M. S.

(1973). The strength of weak ties. American Journal of Sociology, 78, 1360–1380.

Granovetter

(1974). Getting a job: A study of contacts and careers. Chicago, IL: University of Chicago Press.

Hatfield

Walster

G. W.

Berscheid

(1978). Equity: Theory and research. Boston, MA: Allyn & Bacon.

10.

Kalick

S. M.

Hamilton

T. E.

(1986). The matching hypothesis reexamined. Journal of Personality and Social Psychology, 51, 673–682.

11.

Latané

Bourgeois

M. J.

(2001). Successfully simulating dynamic social impact: Three levels of prediction. In Forgas

J. P.

Williams

K. D.

(Eds.), Social influence: Direct and indirect processes (pp. 61–76). New York, NY: Psychology Press.

12.

Mayr

(1997). This is biology: The science of the living world. Cambridge, MA: Harvard University Press.

13.

Muthén

L. K.

Muthén

B. O.

(2007). Mplus user’s guide (4th ed.). Los Angeles, CA: Author.

14.

Nowak

Szamrej

Latané

(1990). From private attitude to public opinion: A dynamic theory of social impact. Psychological Review, 97, 362–376.

15.

Oishi

(2010). The psychology of residential mobility: Implications for the self, social relationships, and well-being. Perspectives on Psychological Science, 5, 5–21.

16.

Oishi

Graham

(2010). Social ecology: Lost and found in psychological science. Perspectives on Psychological Science, 5, 356–377.

17.

Oishi

Kesebir

Miao

F. F.

Talhelm

Endo

Uchida

. . . Norasakkunkit

(2012). Residential mobility increases motivation to expand social network: But why? Manuscript submitted for publication.

18.

Oishi

Miao

F. F.

Koo

Kisling

Ratliff

(2012). Residential mobility breeds familiarity-seeking. Journal of Personality and Social Psychology, 102, 149–162.

19.

Oishi

Rothman

A. J.

Snyder

Zehm

Hertel

A. W.

. . . Sherman

G. D.

(2007). The socioecological model of procommunity action: The benefits of residential stability. Journal of Personality and Social Psychology, 93, 831– 844.

20.

Schelling

T. C.

(1971). Dynamic models of segregation. Journal of Mathematical Sociology, 1, 143–186.

21.

Schmitt

(2001, August 6). Census data show a sharp increase in living standard. The New York Times. Retrieved from www.nytimes.com/2001/08/06/national/06CENS.html

22.

Schug

Yuki

Maddux

(2010). Relational mobility explains between- and within-culture differences in self-disclosure to close friends. Psychological Science, 21, 1471–1478.

23.

Smith

E. R.

Conrey

F. R.

(2007). Agent-based modeling: A new approach for theory-building in social psychology. Personality and Social Psychology Review, 11, 87–104.

24.

Tutin

C. E. G.

(1979). Mating patterns and reproductive strategies in a community of wild chimpanzees (Pan troglodytes schweinfurthii). Behavioral Ecology and Sociobiology, 6, 29–38.

25.

U.S. Census Bureau. (2000). American FactFinder. Retrieved from http://factfinder2.census.gov/faces/nav/jsf/pages/index.xhtml