Internet Use,Online Communication,and Ties in Americans’ Networks

Abstract

Drawing on nationally representative survey data, this article examines the implications of Internet use and online communication for strong and weak ties in Americans’ social networks. In line with the existing literature, this research shows that frequent Internet use and online communication are associated with a larger core discussion network and a more extensive position-generated network. More importantly, this research provides a finer tuned analysis by disaggregating the overall network into strong and weak ties. First, Internet use—but not online communication—is positively related to weak-tie based network extensity in the position-generated networks. Second, Internet use and online communication are positively related to the number and the proportion of strong ties in Americans' core discussion networks. These results help reconcile some of the conflicting findings and interpretations based on different network measures in the exiting literature.

Keywords

social networks strong ties weak ties Internet use online communication the name generator the position generator

There has been a growing concern about the decline of social capital and the increase of social isolation in America (Hampton, Sessions, & Her, 2011; McPherson, Smith-Lovin, & Brashears, 2006; Putnam, 1995, 2000). As Internet diffusion has been one of the most salient technological advancements since the mid-1990s, there has been a heated debate on whether the Internet increases, decreases, or has no impact on social capital. As social media have dramatically enhanced people’s capacity of maintaining and developing network contacts, scholars have further argued whether the Internet has triggered changes in network structure, leading to more superficial, transitory weak-tie relations at the expense of strong ties (McPherson et al, 2006; Turkle, 2011). However, only a small number of studies have examined strong and weak ties by established measures of tie strength and the literature is especially inconclusive on the Internet implications for strong ties. Drawing on nationally representative survey data in the United States, this research offers a finer tuned understanding by examining the relationships between Internet use and online communication and the number and the proportion of strong and weak ties in Americans’ networks.

Internet Use, Online Communication, and Social Networks

Early studies suggested that Internet use decreased sociability (Kraut et al., 1998; Nie & Erbring, 2000). However, a growing body of literature shows that the relationship between Internet use and personal networks is positive or neutral (Hampton & Wellman, 2001; Howard, 2004; Katz & Rice, 2002; Robinson & Martin, 2010; Shklovski, Kiesler, & Kraut, 2006; Vergeer & Pelzer, 2009; Wang & Wellman, 2010; Zhao, 2006). For instance, frequent e-mail use is associated with having a larger number of friends, colleagues, and kin (Boase, 2006). Certain types of Internet use (e.g., sharing photos and Instant Messaging [IM]) is positively associated with the size and the diversity of Americans’ core discussion networks (Hampton, Sessions, & Her, 2011). Using the Internet, using social networking services, and the frequency of Internet use at work are positively associated with the diversity of Americans’ position-generated networks (Hampton, Lee, & Her, 2011).

However, most studies have focused on aggregate network size or broad relational types (such as kin, non-kin, friends, or workmates, see reviews in Shklovski et al., 2006; Wellman, Boase, & Chen, 2002). Scholars have also examined the Internet implications for bonding and bridging social capital, building on Putnam’s concept of bonding social capital as the value generated through social networks among homogeneous people and bridging social capital the value assigned to connections among heterogeneous people (Brooks, Welser, Hogan, & Titsworth, 2011; Ellison, Steinfield, & Lampe, 2007; Steinfield, Ellison, & Lampe, 2008). Yet, defining social capital by its potential values tends to confuse the causes and consequences of social capital (Lin, 2001). As Putnam’s notion of bonding and bridging social capital indeed draws on Granovetter’s (1973) distinction of social connections in strong and weak ties, it is important to analyze strong and weak ties separately as they are of different nature and have different functions. In addition, most existing studies on Internet use and personal networks have used either the core discussion networks measured by the name generator or the position-generated networks based on the position generator. Only a few studies have utilized network data generated by both types of generators.

A typical personal network has a small number of strong ties at the core and a larger number of weak ties on the periphery. Strong ties involve emotional bonds, trust, and reciprocity. People feel close to their strong ties but less close to their weak ties. Accordingly, strong ties require more time, energy, and commitment to nurture and weak ties do not need intensive maintenance (Granovetter, 1973). Due to their different nature, strong and weak ties serve different social functions. On one hand, the trust and obligation embedded in strong ties facilitate the transfer of fine-grained information, tacit knowledge, and valuable resources, encourage collaboration, and enhance social control (Uzzi, 1996). Strong ties influence people’s opinions (Katz & Lazarsfeld, 1955). They offer emotional, instrumental, and financial support (Hurlbert, Beggs, & Haines, 2001; Wellman & Wortley, 1990). On the other hand, Granovetter (1973) sees a great potential of weak ties in linking individuals with diverse groups in the larger society as weak ties are more heterogeneous than strong ties and bridge otherwise separated groups. The exposure to diverse contacts brings people fresh information and perspectives, which helps them to develop cognitive flexibility and cultural capital (Erickson, 1996). In his seminal work, Granovetter defines the strength of ties as “a (probably linear) combination of the amount of time, emotional intensity, intimacy (mutual confiding), and reciprocal services” (1973, p. 1361). Marsden and Campbell analyzed three data sets from the United States and Germany and identified closeness or emotional intensity as the best indicator of tie strength, better than frequency or duration of contact or relational type (Marsden & Campbell, 1984, p. 482; see also Marsden, 1990; McPherson, Smith-Lovin, & Cook, 2001). In what follows, I draw on the existing literature to develop hypotheses and research questions on the relationship between Internet use, online communication, strong and weak ties.

Weak Ties

Most existing studies have argued for a positive relationship between Internet use, online communication, and weak ties due to the compatibility between technological affordances and the low-maintenance nature of weak ties (Donath & Boyd, 2004; Gil de Zúñiga & Valenzuela, 2011; Haythornthwaite, 2002). Using social networking site increases access to new contacts (Steinfield, DiMicco, Ellison, & Lampe, 2009). Social media facilitate pervasive awareness—the depth and breadth of information about one’s network contacts shared on social networking sites (Hampton, Sessions, & Her, 2011). Pervasive awareness may be especially significant to weak ties as people may already know more about their strong ties than weak ties. Both cross-sectional and longitudinal data have revealed a positive relationship between the intensity of Facebook use and weak-tie based bridging social capital among college students (Ellison et al., 2007; Steinfield et al., 2008). In addition, greater intensity of using an organizational social networking site is found to be positively associated with workers’ bridging social capital (Steinfield et al., 2009). Only a few scholars have raised concerns that new communication technologies may harm weak ties when people are too encapsulated by the intensified social interactions with their existing strong ties (Gergen, 2008). Overall, the literature is almost unanimous on a positive relationship between Internet use, online communication, and weak ties. Thus,

Hypothesis 1: The frequency of Internet use and online communication is positively related to the number of weak ties.

Strong Ties

Although a growing number of studies have examined the Internet implications for strong and weak ties (Baym, Zhang, & Lin, 2004; Boase, 2006, 2008; Chen, 2013; Jones et al., 2013; Pollet, Roberts, & Dunbar, 2011; Veenhof, Wellman, Quell, & Hogan, 2008), research on or related to strong ties has been inconclusive and can be roughly categorized into three groups: the reinforcement hypothesis, the displacement hypothesis, and the no impact hypothesis.

The reinforcement hypothesis argues that Internet use and online communication help intensify the existing strong ties (Bargh & McKenna, 2004; Kraut et al., 2002; Quan-Haase & Wellman, 2005). The more people e-mail, the more they communicate face-to-face or via phone calls (Boase, 2008; Chen, Boase, & Wellman, 2002). This mutually reinforcing relationship is especially pronounced among strong ties (Baym et al., 2004). The frequency of Internet use is positively related to strong-tie based bonding social capital on Facebook among a sample of college students (Brooks et al., 2011). There is a positive relationship between Facebook use intensity and bonding social capital among college students (Ellison et al., 2007). In a similar vein, greater intensity of using an organizational social networking site is positively associated with workers’ bonding social capital (Steinfield et al., 2009). A new study of Facebook users shows that the greater the frequency of Facebook interaction between two users, the more likely they have strong ties with each other (Jones et al., 2013). However, there has been a lack of research on whether and the extent to which the intensification of communication with existing strong ties would add new strong ties into people’s networks. One study shows that frequent Internet use is positively associated with a greater stability in the number of strong ties but a greater fluctuation in the number of weak ties in Americans’ position-generated networks (Chen, 2013).

The displacement argument builds on the time-displacement hypothesis (Nie & Erbring, 2000), which assumes a hydraulic nature of time use and a trade-off between time spending online and face-to-face time with family and friends. In a similar vein, as the Internet enables more access to contacts outside of one’s immediate social world, there are concerns that the Internet may decrease people’s investment in strong, local, and communal relationships (Olds & Schwartz, 2009). In a study of American college students, “approximately one-third (35%) of male and about a quarter (23%) of female respondents indicated that the time they spend online takes up time they would prefer to spend engaging in other social interactions” (Jones, Johnson-Yale, Millermaier, & Perez, 2009, p. 259). Members of Internet-connected households are more likely to feel ignored by family members (Pierce, 2009). New communication technologies, especially mobile Internet, may contaminate the quality time with family and close friends when users interact “side-by-side” rather than “face-to-face” (Ito, Okabe, & Matsuda, 2005). Turkle (2011) argues that people use technologies to shield themselves from rather than engage in authentic interpersonal relationships, which can increase the sense of alienation. However, the existing studies are largely silent on whether and the ways in which the potential time displacement or alienation would be related to the number of strong ties in people’s networks.

In-between the reinforcement and the displacement camp, there are studies suggesting that new communication technologies may have limited impacts on strong ties (Bargh &McKenna, 2004; Baym et al., 2004). Research in Hong Kong shows no significant difference in the amount of time spending on social activities by Internet users and nonusers (Peng & Zhu, 2010). Close to 90% of American Internet users report that the Internet has not affected the amount of time they spend with family and friends (Kennedy, Smith, Wells, & Wellman, 2008). Research also fails to identify a significant difference in the number of close ties between Internet users and nonusers in America or in Canada (Boase, 2006; Veenhof et al., 2008). The use of IM and social networking sites is not associated with feeling emotionally closer to one’s offline network members (Pollet et al., 2011).

A few factors may contribute to the lack of significant Internet implications for the number of strong ties. First, the trust, emotional bond, and obligation embedded in strong ties make them less sensitive to communication barriers as people would use whatever means of communication available to maintain contact with their strong ties. Second, online interaction is structured by preexisting social networks and group boundaries (Cho & Lee, 2008) and people tend to use instant or text message to maintain contact with a small number of very close ties (Ling, 2008; Valkenburg & Peter, 2009). In a similar vein, while people may have hundreds of Facebook contacts, a typical user only has two-way Facebook communication with a handful contacts: four for male and six for female users (Marlow, 2009). Third, although the Internet makes it easier to build and maintain weak ties, it may not add new strong ties into people’s networks. People make friends online only occasionally and such friends do not easily become strong ties (Kraut et al., 1998; Mesch & Talmud, 2006; Wang & Wellman, 2010). However, there has been a lack of research on whether and the extent to which the intensification of communication with existing strong ties or the potential time displacement would be related to the number of strong ties in people’s networks. Thus, instead of a hypothesis, a research question is formulated:

Research Question 1: How is the frequency of Internet use and online communication related to the number of strong ties?

A Shift From Strong Ties to Weak Ties?

Scholars have been speculating whether the Internet has triggered changes in network structure, leading to more superficial, transitory relations at the expense of strong ties (McPherson et al., 2006). If Internet implications for network size vary by tie strength, it may also affect the composition of strong and weak ties in Americans’ networks. There can be a trade-off between strong and weak ties due to the limited capacity of the human brain to maintain a large number of direct network contacts (Dunbar, 2003). Mayhew and Levinger (1976) hypothesized that the more contacts crowded into people’s life, the less attention they would give to each contact. In such a scenario, strong ties may suffer more than weak ties as the former require more time and energy to cultivate than the latter. Recent research indicates that the mean emotional closeness in people’s networks—an indicator of the presence and the proportion of strong ties—is negatively associated with their network size (Roberts, Dunbar, Pollet, & Kuppens, 2009). Even though the absolute number of strong ties remains constant or even increases, the proportion of weak ties may increase if Internet use and online communication enable people to add more weak ties than strong ties. Indeed, using an organizational social networking site for adding new contacts is negatively related to workers’ strong-tie based bonding social capital (Steinfield et al., 2009). Thus,

Hypothesis 2: The frequency of Internet use and online communication is positively related to the proportion of weak ties.

Data and Method

Data and Sample

The data come from the Social Capital in the U.S. Survey conducted in 2008 (hereafter SCUS 2008 Survey), a national random digit dial telephone survey of currently or previously employed Americans aged 21 to 64. The project was sponsored by Academia Sinica, Taiwan, and Duke University. A total of 1,407 interviews were completed in 2008 and the response rate was 39%. The data were weighted, using the rake procedure in Stata to match the distribution of gender, race, age, marital status, and education in the survey to the Current Population Survey in 2008. To increase generalizability, sample weights were applied to analyses presented here.

Network Data Collected by Different Methods

Personal networks can be captured by either the core discussion networks measured by the name generator or the more extended networks measured by the position generator (Lin & Dumin, 1986; Marsden, 1987; van der Gaag, 2005). The name generator usually places an artificial limit on the number of names to be generated. For instance, the name generator in the General Social Survey (GSS) asks the respondent to nominate up to five persons with whom he or she discussed important matters in the last 6 months (Marsden, 1987). It tends to elicit more strong ties than weak ties and captures the small, kin-centered, dense, and homogeneous personal environments (Marsden & Campbell, 1984). However, it is important to point out that not all ties in the core discussion networks are strong as people do discuss important matters with weak ties too (Bearman & Parigi, 2004).

The position generator collects information about the respondent’s network contacts who fill a variety of positions in the occupational hierarchy (Lin, Fu, & Hsung, 2001). It is theory driven as the positions, ranging from high to low status, are sampled from a full list of occupations in the census and indicate the respondent’s access to a wide range of resources for instrumental and expressive returns. While the GSS name generator is centered on matters that people deem as important, the position generator is content free (Lin et al., 2001). The position generator is more effective than the GSS name generator in capturing weak ties (Chen & Tan, 2009; Erickson, 2004).

Most existing studies on Internet use and social networks have used either the core discussion networks measured by the name generator or the position-generated networks based on the position generator. Only a few studies have examined network data generated by both the name generator and the position generator. A Japanese study shows that while PC e-mail use and the size of people’s supportive networks are mutually reinforcing, it is not related to the size of position-generated network (Miyata & Kobayashi, 2008). A study of immigrant entrepreneurs demonstrates that Internet activities contribute to the global outreach of the position-generated network but not that of the core network (Chen & Wellman, 2009). Hampton (2011) examines the relationship between the core discussion networks (theorized as bonding social capital) and network extensity measured by the position generator (theorized as bridging social capital) and civic engagement. Yet, these studies do not differentiate strong ties from weak ties.

Measures: Dependent Variables

The SCUS 2008 Survey is one of the few studies that used both the name generator and the position generator to collect network data, allowing a comparison of the core discussion network and the position-generated network. The name generator replicated the one in the GSS. The question was formulated as follows: From time to time, most people discuss important matters with other people. Looking back over the last six months—who are the people with whom you discussed matters important to you?

The position generator in the survey gave the respondent a list of 22 occupations and asked whether he or she knew someone in each of the occupations. The question was formulated as follows: I am going to ask some general questions about jobs some people you know may now have. These people include your relatives, friends and acquaintances (acquaintances are people who know each other by face and name). If there are several people you know who have that kind of job, please tell me the one that occurs to you first. Is there anyone you know who is a ____ (occupation)? The 22 occupations sampled in the position generator included nurse, writer, farmer, lawyer, high school teacher, babysitter (housemaid), janitor, personnel manager, administrative assistant, hair dresser, accountant, guard, production manager, factory operator, computer programmer, receptionist, congressman/woman, taxi driver, professor, hotel bell boy, police officer, and CEO in a big company. These 22 occupations have been tested in nationally representative surveys in the United States, Canada, Europe, and Asia (Lin, 2001).

The dependent variables included the size of the core discussion network, the extensity of the position-generated network, the number of strong ties and weak ties, and the proportion of weak ties in the core discussion network and the position-generated network, respectively. The size of the core discussion network was measured by the number of people with whom the respondent had discussed important matters in the past 6 months (M = 2.50, SD = 2.06). The extensity of the position-generated networks was measured by the number of occupations in which the respondent had access to (M = 7.25, SD = 4.35). Table 1 reports the descriptive statistics.

Table 1.

Descriptive Statistics of Dependent and Independent Variables.

	N	M	SD	Min	Max
The core discussion network
Overall size	1,397	2.50	2.06	0	6
The number of strong ties	1,397	2.17	1.78	0	5
The number of weak ties	1,397	0.21	0.63	0	5
The proportion of weak ties	1,044	0.08	0.21	0	1
The presence of weak ties	1,397	0.13	0.34	0	1
The position-generated network
Network extensity	1,397	7.25	4.35	0	21
Network extensity via strong ties	1,397	3.12	2.74	0	19
Network extensity via weak ties	1,397	4.13	3.16	0	17
The proportion of weak ties	1,317	0.56	0.27	0	1
Internet use	1,397	10.99	15.96	0	153
Online communication	1,397	4.42	9.59	0	100
Age	1,397	43.41	10.91	21	64
Female	1,397	0.47	0.50	0	1
Education	1,397	2.93	1.18	1	5
Race
Whites	1,397	0.70	0.46	0	1
Blacks	1,397	0.10	0.31	0	1
Hispanics	1,397	0.14	0.35	0	1
Other	1,397	0.06	0.23	0	1
Partnered	1,397	0.65	0.48	0	1

Closeness or emotional intensity of a relationship has been considered the best indicator of tie strength, compared to the relational type, duration, or frequency of contact (Marsden & Campbell, 1984). Respondents were asked to evaluate the closeness of his or her relationship with each network contact, using a 5-point Likert-type scale where 1 is very close, 2 close, 3 so-so, 4 not close, and 5 not close at all. A tie was coded as strong if it was very close or close. Otherwise, it was coded as weak.

The number of strong ties in the core discussion network was the number of core discussion network contacts whom the respondent felt close or very close to (M = 2.17, SD = 1.78). The number of weak ties in the core discussion network was the number of core discussion network contacts whom the respondent did not feel close or very close to (M = .21, SD = .63). The proportion of weak ties in the core discussion network was measured by the percentage of weak ties in the core discussion network (M = .08, SD = .21). Overall, there were 13% of respondents who had at least one weak tie in their core discussion networks.

The strong-tie based extensity in the position-generated network was measured as the total number of occupations a respondent had access to via strong ties (M = 3.12, SD = 2.74). The weak-tie based extensity in the position-generated network was measured as the total number of occupations a respondent had access to via weak ties (M = 4.13, SD = 3.16). The proportion of weak ties in the position-generated network was measured by the percentage of weak ties in the position-generated network (M = .56, SD = .27).

Independent and Control Variables

The frequency of Internet use and the frequency of online communication were the key independent variables. The frequency of Internet use was measured by the respondent’s answer to the question on “how much time in a typical week do you spend on the Internet” (M = 10.99, SD = 15.96). The frequency of online communication was measured by the respondent’s answer to the question on “how much time in a typical week do you spend on the Internet communicating/contacting with” people the respondent had daily contact (M = 4.42, SD = 9.59). Both questions asked the respondents to give a specific number of hours.

Men, people with higher socioeconomic status, and married people tend to have larger, more diverse networks than women, people with lower socioeconomic status, and singles (Lin, 2001; McPherson et al., 2006). Thus, sociodemographic variables were controlled, including age and age square, gender, education, race/ethnicity, and marital status. The average age of the respondents was 43. The square term of age was included because network size tends to grow with age but declines around middle age. Gender was a dichotomous variable coded as 1 if the respondent was female and 0 otherwise (M = .47, SD = .50). Education was an ordinal variable, ranging from 1 less than high school, 2 high school, 3 associate college, 4 college, to 5 postgraduate (M = 2.93, SD = 1.18). Race had four categories. About 70% of the respondents were Whites, 10% Blacks, 14% Hispanics, and 6% other race. Partnered was a dichotomous variable coded as 1 if the respondent was married or lived with a partner and 0 otherwise (M = .65, SD = .48).

Results

Analysis

The size of the core discussion network and the extensity of the position-generated networks aggregated or disaggregated by tie strength, can be count variables that do not follow a normal distribution, and may have an excessive number of zeros (McPherson, Smith-Lovin, & Brashears, 2009; Son, 2012). For example, 25% of respondents in the SCUS 2008 survey reported having zero contacts, 27% zero strong ties, and 87% zero weak ties in their core discussion networks. Even though there are fewer zeros in the position-generated network, 6% of respondents reported having zero contacts, 14% zero strong ties, and 12% zero weak ties in their position-generated networks. Thus, zero-inflated Poisson (ZIP) models were used to examine the relationship of Internet communication, online communication, and network variables. The ZIP model assumes two regimes in the population: Members in the first regime always have zero counts and members in the second regime can have either zero or positive counts. Accordingly, the ZIP model first generates a logit model to estimate the respondent’s membership in the first regime. Then, it uses a Poisson model to estimate the counts for respondents in the second regime (Long, 1997). The Vuong tests were used to compare the ZIP model with the standard Poisson model and indicated the use of the former was appropriate.

The Number of Weak Ties and Strong Ties

Appendix A showed that more frequent Internet use and online communication were significantly related to a larger core discussion network and a more extensive position-generated network.¹ Table 2 showed the results of ZIP models estimating the number of weak ties in the core discussion network (Models 1–2) and the weak-tie based extensity in the position-generated networks (Models 3–4). Models 1 and 2 showed that neither Internet use nor online communication was significantly related to the number of weak ties in the core discussion network. Models 3 and 4 showed that Internet use—but not online communication—was positively related to weak-tie based extensity in the position-generated network. For instance, if the time spending on Internet use per week was increased by 10 hr, the weak-tie based network extensity would increase by a factor of exp(0.03) = 1.03, holding all other variables in the model constant. Thus, Hypothesis 1 that the frequency of Internet use and online communication would be positively related to the number of weak ties was only partially supported.

Table 2.

The Number of Weak Ties (Zero-Inflated Poisson Models).

	Core Discussion Networks		Position-Generated Networks
	Model 1	Model 2	Model 3	Model 4
Internet use	−0.009		0.003*
Internet use	(0.012)		(0.001)
Online communication		−0.022		0.003
Online communication		(0.019)		(0.002)
Age	0.092	0.091	0.049**	0.049**
Age	(0.063)	(0.064)	(0.018)	(0.018)
Age squared	−0.001	−0.001	−0.001*	−0.001*
Age squared	(0.001)	(0.001)	(0.000)	(0.000)
Female	−0.061	−0.060	−0.102*	−0.104*
Female	(0.151)	(0.152)	(0.042)	(0.042)
Education	0.059	0.059	0.089***	0.093***
Education	(0.065)	(0.064)	(0.018)	(0.017)
Race (Whites as reference)
Blacks	−0.362	−0.370	−0.084	−0.086
Blacks	(0.273)	(0.266)	(0.061)	(0.062)
Hispanics	0.913***	0.894***	−0.172**	−0.173**
Hispanics	(0.193)	(0.194)	(0.062)	(0.062)
Other	−1.207	−1.208	−0.329*	−0.325*
Other	(0.677)	(0.676)	(0.152)	(0.152)
Partnered	−0.032	−0.036	0.091	0.091
Partnered	(0.162)	(0.162)	(0.049)	(0.050)
Constant	−2.556	−2.512	0.149	0.145
Constant	(1.382)	(1.404)	(0.409)	(0.410)
Inflate
Internet use	0.003		−0.002
Internet use	(0.015)		(0.006)
Online communication		0.008		0.008
Online communication		(0.025)		(0.009)
Constant	0.967***	0.960***	−2.151***	−2.213***
Constant	(0.192)	(0.162)	(0.136)	(0.126)
Log likelihood	−718.55	−716.97	−3,485.81	−3,491.15

Note. Robust standard errors in parentheses.

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

Table 3 showed the results of ZIP models estimating the number of strong ties in the core discussion networks (Models 1–2) and the strong-tie based extensity in the position-generated networks (Models 3–4). The results showed that frequent Internet use and online communication were positively related to more strong ties in the core discussion network (Models 1–2) but not in the position-generated network (Models 3–4). If the respondent were to increase the time spending on Internet use by 10 hr per week, the expected number of strong ties in the core discussion network would increase by a factor of exp(0.03) = 1.03; if the respondent were to increase the time spending on digital communication by 10 hr per week, the expected number of strong ties in the core discussion network would increase by a factor of exp(0.07) = 1.07, while holding all other variables in the model constant.

Table 3.

The Number of Strong Ties (Zero-Inflated Poisson Models).

	Core Discussion Networks		Position-Generated Networks
	Model 1	Model 2	Model 1	Model 2
Internet use	0.003*		0.002
Internet use	(0.002)		(0.001)
Online communication		0.007***		0.002
Online communication		(0.002)		(0.002)
Age	−0.015	−0.016	0.016	0.017
Age	(0.016)	(0.016)	(0.022)	(0.022)
Age squared	0.000	0.000	−0.000	−0.000
Age squared	(0.000)	(0.000)	(0.000)	(0.000)
Female	0.170***	0.173***	0.037	0.036
Female	(0.044)	(0.044)	(0.055)	(0.055)
Education	0.076***	0.079***	0.043*	0.045*
Education	(0.020)	(0.019)	(0.022)	(0.021)
Race (Whites as reference)
Blacks	−0.313***	−0.318***	0.384***	0.386***
Blacks	(0.082)	(0.081)	(0.068)	(0.068)
Hispanics	−0.180**	−0.173**	−0.058	−0.053
Hispanics	(0.062)	(0.061)	(0.080)	(0.079)
Other	−0.124	−0.113	−0.127	−0.122
Other	(0.170)	(0.176)	(0.164)	(0.163)
Partnered	0.037	0.046	0.165**	0.166**
Partnered	(0.047)	(0.046)	(0.061)	(0.061)
Constant	1.030**	1.036**	0.559	0.544
Constant	(0.340)	(0.337)	(0.465)	(0.465)
Inflate
Internet use	−0.005		−0.000
Internet use	(0.010)		(0.007)
Online communication		−0.006		−0.007
Online communication		(0.009)		(0.012)
Constant	−1.297***	−1.330***	−2.103***	−2.076***
Constant	(0.148)	(0.110)	(0.144)	(0.132)
Log likelihood	−2,528.53	−2,526.30	−3,240.66	−3,241.61

Note. Robust standard errors in parentheses.

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

Thus, the answer to Research Question 1 on the relationship between Internet use and online communication and the number of strong ties seems to depend on whether it is about the core discussion network or the position-generated network. The more time spending on Internet use and online communication, the more strong ties the respondent would have in the core discussion networks. However, neither Internet use nor online communication was significantly related to the strong-tie based extensity in the position-generated networks. That is, both the reinforcement hypothesis and the no impact hypothesis were partially supported, while the displacement hypothesis was not supported.

The Proportion of Weak Ties

Table 4 reported the results of the generalized linear models estimating the proportion of weak ties in the core discussion networks (Models 1-2) and in the position-generated networks (Models 3-4). Both Internet use and online communication were negatively related to the proportion of weak ties in the core discussion network (b = −0.015, b = −0.027, p ≤ .05, respectively). By contrast, none of them had a significant relationship with the proportion of weak ties in the position-generated network. Thus, Hypothesis 2 that the frequency of Internet use and online communication would be positively related with a higher proportion of weak ties was rejected. Furthermore, Appendix B showed that more frequent online communication was significantly related to a smaller likelihood of having at least one weak tie in the core discussion networks.

Table 4.

The Proportion of Weak Ties (Generalized Linear Model Models).

	Core Discussion Networks		Position-Generated Networks
	Model 1	Model 2	Model 3	Model 4
Internet use	−0.015*		0.000
Internet use	(0.006)		(0.002)
Online communication		−0.027*		−0.003
Online communication		(0.014)		(0.004)
Age	0.143	0.148	0.055	0.056*
Age	(0.076)	(0.076)	(0.028)	(0.028)
Age squared	−0.001	−0.001	−0.001	−0.001
Age squared	(0.001)	(0.001)	(0.000)	(0.000)
Female	−0.188	−0.181	−0.201**	−0.200**
Female	(0.183)	(0.182)	(0.070)	(0.070)
Education	−0.055		0.064*
Education	(0.075)	−0.067	(0.029)	0.069*
Race (Whites as reference)		(0.076)		(0.028)
Blacks	0.049	0.048	−0.326**	−0.325**
Blacks	(0.301)	(0.300)	(0.105)	(0.105)
Hispanics	1.255***	1.250***	0.069	0.068
Hispanics	(0.209)	(0.208)	(0.105)	(0.105)
Other	−1.242	−1.273	−0.245	−0.241
Other	(0.669)	(0.669)	(0.192)	(0.192)
Partnered	−0.218	−0.229	−0.049	−0.054
Partnered	(0.202)	(0.201)	(0.079)	(0.079)
Constant	−5.493***	−5.582***	−1.092	−1.106
Constant	(1.521)	(1.535)	(0.604)	(0.604)
Log likelihood	−244.32	−244.28	−673.04	−672.94

Note. Robust standard errors in parentheses.

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

Comparing With Results on Kin and Non-Kin Ties

I further compared the results on strong and weak ties with the results on kin and non-kin ties. The more time spending on Internet use and online communication, the more extensive the position-generated networks via non-kin ties. Yet, only online communication was positively related to the number of non-kin ties in the core discussion network (Appendix C).² Neither Internet use nor online communication was significantly related to kin ties in the core discussion network or in the position-generated network (Appendix D). Appendix E showed that neither Internet use nor online communication was significantly related to the proportion of non-kin ties in the core discussion networks or the position-generated networks. Overall, the results showed that it would be problematic to use kin and non-ties as a proxy of strong and weak ties.

Discussion and Conclusion

Much attention has been focused on searching for the smoking guns responsible for the decline of social capital (Putnam, 1996, 2000). Pessimists have speculated that the Internet may have played a negative role (Gergen, 2008; Kraut et al., 1998; Nie & Erbring, 2000; Turkle, 2011). However, a growing body of literature has shown that Internet use and online communication, in general, do not harm sociability and often increase network size and diversity (Hampton, Lee, & Her, 2011; Hampton, Sessions, & Her, 2011; Hampton & Wellman, 2001; Howard, 2004; Katz & Rice, 2002; Shklovski et al., 2006; Vergeer & Pelzer, 2009; Wang & Wellman, 2010; Zhao, 2006). Yet, few studies have investigated whether Internet use and online communication have different implications for strong and weak ties. Even fewer studies have investigated whether and how the relationship may vary by network data generated by different methods.

Drawing on nationally representative survey data, this article addresses these knowledge gaps and examines the Internet implications for strong and weak ties in Americans’ core discussion networks and position-generated networks. In line with the existing literature, this research shows that frequent Internet use and online communication are associated with a larger core discussion network and a more extensive position-generated network. More importantly, this research provides a finer tuned analysis by disaggregating the overall network into strong and weak ties. These results help reconcile some of the conflicting findings and interpretations based on different network measures in the exiting literature. It draws the following conclusions.

First, the bottom line is that Internet use and online communication are not related with a smaller number of strong ties or weak ties in Americans’ networks. Moreover, frequent Internet use and online communication are related to a larger number of strong ties in the core discussion network. In addition, frequent Internet use—but not online communication—is related to greater weak-tie based extensity in the position-generated networks.

Second, this research demonstrated that it is important to examine the Internet implications by tie strength and network data generated by different methods as the results may vary significantly. For instance, whether the reinforcement hypothesis or the no impact hypothesis on the Internet implications for strong ties is supported varies by networks measured by different methods. While the Internet intensifies and enriches communication among strong ties, it is only positively associated with a greater number of strong ties in the core discussion network but not with a greater extensity via strong ties in the position-generated network. Internet use is only significant to greater extensity via weak ties in the position-generated network but not to a greater number of weak ties in the core discussion network.

Third, the Internet has not yet facilitated a shift in network structure from strong to weak ties, either in the core discussion network or in the position-generated network. McPherson and colleagues (2006) found that non-kin ties—as a proxy of weak ties—experienced the largest decline in Americans’ core discussion networks over the last two decades. However, this research does not find a significant relationship between Internet use, online communication, and the proportion of non-kin ties.

These findings seem to paint a rosy picture of frequent Internet use and online communicators having the best of both worlds: maintaining a strong network core and a more extensive position-generated network. There is, however, one concern. This research shows that, the more frequent Internet use and online communication, the greater the proportion of strong ties and the smaller the proportion of weak ties in the core discussion network. That is, the intensified digital communication among strong ties may crowd out weak ties in the core discussion networks. This research lends support to the notion that new communication technologies may harm weak ties when people are too encapsulated by the intensified social interactions with their existing strong ties (Gergen, 2008).

The research has several limitations that call for future research. First, the survey data are cross-sectional and insufficient to determine causality. The relationship between Internet use, online communication, and social networks has become increasingly reciprocal. A better understanding requires panel data. Second, the measurement of Internet use and online communication is limited to the number of hours spending online in general and on communication with daily contacts. The focus on daily contacts may overestimate the relationship between online communication and strong ties. The survey has no detailed information about the mode of online communication (via e-mail, IM, text message, social networking sites, or online chat rooms and forums) and the purpose of online communication. Research shows that specific mode of digital communication tends to have differential relationship with network size and diversity (Hampton, Lee, & Her, 2011; Hampton, Sessions, & Her, 2011). Thus, future research needs to explore the variations of online communication and tie strength by specific mode and purposes of digital communication.

Individuals need a mix of strong ties and weak ties to gain reliable access to a wider range of information, perspectives, and resources. Strong ties are needed for local solidarity and weak ties for integration. The research makes important contributions by offering a more nuanced examination of the Internet implications for personal networks by tie strength and network data generated by different methods.

The Internet is a bundle of versatile technologies that can help people develop both strong and weak ties. However, the extent to which such potential can be realized is more about how people use the Internet than what the Internet can technically afford. As technologies and how we use them still evolve, we need to be mindful about how the relationship between online communication and social networks vary by strong and weak ties.

Footnotes

Appendix A

The Size of the Core Discussion Networks and the Extended Networks (Zero-Inflated Poisson Models).

	Core Discussion Networks		Position-Generated Networks
	Model 1	Model 2	Model 3	Model 4
Internet use	0.003*		0.003**
Internet use	(0.001)		(0.001)
Online communication		0.007***		0.003*
Online communication		(0.002)		(0.002)
Age	−0.014	−0.015	0.027	0.028
Age	(0.017)	(0.017)	(0.015)	(0.015)
Age squared	0.000	0.000	−0.000	−0.000
Age squared	(0.000)	(0.000)	(0.000)	(0.000)
Female	0.137**	0.140**	−0.029	−0.031
Female	(0.043)	(0.043)	(0.034)	(0.034)
Education	0.074***		0.077***
Education	(0.020)	0.076***	(0.014)	0.080***
Race (Whites as reference)		(0.019)		(0.014)
Blacks	−0.323***	−0.327***	0.100*	0.098*
Blacks	(0.080)	(0.080)	(0.045)	(0.045)
Hispanics	−0.022	−0.014	−0.149**	−0.148**
Hispanics	(0.060)	(0.059)	(0.048)	(0.048)
Other	−0.167	−0.155	−0.236	−0.232
Other	(0.174)	(0.180)	(0.136)	(0.136)
Partnered	0.005	0.014	0.139***	0.140***
Partnered	(0.045)	(0.044)	(0.037)	(0.038)
Constant	1.161**	1.169***	1.071***	1.068***
Constant	(0.358)	(0.355)	(0.317)	(0.316)
Inflate
Internet use	−0.003		−0.003
Internet use	(0.007)		(0.007)
Online communication		−0.005		0.010
Online communication		(0.008)		(0.009)
Constant	−1.293***	−1.309***	−2.798***	−2.878***
Constant	(0.126)	(0.101)	(0.158)	(0.146)
Log likelihood	−2,719.80	−2,716.52	−4,102.50	−4,107.47

Note. Robust standard errors in parentheses.

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

Appendix B

The Presence of Weak Ties in the Core Discussion Networks (Logistic).

	Model 1	Model 2
Internet use	0.991
Internet use	(0.006)
Online communication		0.975*
Online communication		(0.012)
Age	1.045	1.049
Age	(0.078)	(0.078)
Age squared	1.000	1.000
Age squared	(0.001)	(0.001)
Female	0.974	0.982
Female	(0.172)	(0.173)
Education	1.114	1.118
Education	(0.080)	(0.080)
Race (Whites as reference)
Blacks	0.720	0.723
Blacks	(0.189)	(0.190)
Hispanics	2.747***	2.739***
Hispanics	(0.576)	(0.573)
Other	0.417	0.420
Other	(0.313)	(0.314)
Partnered	0.865	0.847
Partnered	(0.160)	(0.158)
Constant	0.036*	0.034*
Constant	(0.056)	(0.054)
Log likelihood	−531.23	−529.82

Note. Robust standard errors in parentheses.

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

Appendix C

The Number of Non-Kin Ties (Zero-Inflated Poisson Models).

	Core Discussion Networks		Position-Generated Networks
	Model 1	Model 2	Model 3	Model 4
Internet use	0.003		0.004**
Internet use	(0.003)		(0.001)
Online communication		0.007*		0.005**
Online communication		(0.003)		(0.002)
Age	0.041	0.043	0.038**	0.038**
Age	(0.037)	(0.028)	(0.014)	(0.014)
Age squared	−0.000	−0.000	−0.000*	−0.000*
Age squared	(0.000)	(0.000)	(0.000)	(0.000)
Female	−0.003	0.044	−0.101**	−0.103**
Female	(0.089)	(0.070)	(0.037)	(0.037)
Education	0.099*	0.058	0.095***	0.098***
Education	(0.041)	(0.033)	(0.016)	(0.015)
Race (Whites as reference)
Blacks	−0.321*	−0.334**	0.061	0.057
Blacks	(0.144)	(0.115)	(0.052)	(0.052)
Hispanics	0.235*	0.201*	−0.138*	−0.138*
Hispanics	(0.114)	(0.094)	(0.055)	(0.055)
Other	−0.288	−0.224	−0.352**	−0.348**
Other	(0.546)	(0.292)	(0.108)	(0.109)
Partnered	−0.136	−0.145*	0.073	0.076
Partnered	(0.092)	(0.073)	(0.041)	(0.041)
Constant	−0.699	−0.636	0.586	0.591
Constant	(0.856)	(0.594)	(0.313)	(0.314)
Inflate
Internet use	−0.000		−0.006
Internet use	(0.005)		(0.007)
Online communication		−0.004		0.011
Online communication		(0.007)		(0.009)
Constant	−0.051	−0.097	−2.222***	−2.336***
Constant	(0.123)	(0.088)	(0.139)	(0.124)
Log likelihood	−1,651.04	−1,658.37	−3,853.23	−3,856.83

Note. Robust standard errors in parentheses.

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

Appendix D

The Number of Kin Ties (Zero-Inflated Poisson Models).

	Core Discussion Networks		Position-Generated Networks
	Model 1	Model 2	Model 3	Model 4
Internet use	−0.001		0.000
Internet use	(0.002)		(0.002)
Online communication		0.003		−0.001
Online communication		(0.005)		(0.004)
Age	−0.036	−0.034	−0.002	−0.002
Age	(0.024)	(0.025)	(0.027)	(0.027)
Age squared	0.000	0.000	−0.000	−0.000
Age squared	(0.000)	(0.000)	(0.000)	(0.000)
Female	0.311***	0.309***	0.161**	0.162**
Female	(0.065)	(0.065)	(0.061)	(0.062)
Education	0.054*	0.053	−0.001	−0.000
Education	(0.027)	(0.027)	(0.026)	(0.025)
Race (Whites as reference)
Blacks	−0.315**	−0.327**	0.246***	0.246***
Blacks	(0.111)	(0.111)	(0.068)	(0.068)
Hispanics	−0.189*	−0.196*	−0.078	−0.078
Hispanics	(0.091)	(0.092)	(0.101)	(0.103)
Other	0.133	0.105	0.191	0.190
Other	(0.249)	(0.250)	(0.271)	(0.271)
Partnered	0.293***	0.304***	0.278***	0.277***
Partnered	(0.084)	(0.084)	(0.072)	(0.072)
Constant	0.911	0.855	0.516	0.515
Constant	(0.521)	(0.532)	(0.553)	(0.555)
Inflate
Internet use	−0.026		0.002
Internet use	(0.019)		(0.007)
Online communication		−0.007		0.002
Online communication		(0.020)		(0.012)
Constant	−1.054***	−1.262***	−1.846***	−1.830***
Constant	(0.193)	(0.172)	(0.176)	(0.170)
Log likelihood	−2,139.04	−2,140.26	−2,463.54	−2,463.53

Note. Robust standard errors in parentheses.

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

Appendix E

The Proportion of Non-Kin Ties (Generalized Linear Model Models).

	Core Discussion Networks		Position-Generated Networks
	Model 1	Model 2	Model 3	Model 4
Internet use	0.001		0.005
Internet use	(0.004)		(0.003)
Online communication		0.003		0.003
Online communication		(0.006)		(0.005)
Age	0.088	0.087	0.066*	0.068*
Age	(0.050)	(0.050)	(0.031)	(0.032)
Age squared	−0.001	−0.001	−0.001	−0.001
Age squared	(0.001)	(0.001)	(0.000)	(0.000)
Female	−0.094	−0.094	−0.190*	−0.191*
Female	(0.122)	(0.122)	(0.081)	(0.081)
Education	−0.006	−0.008	0.143***	0.152***
Education	(0.053)	(0.053)	(0.034)	(0.034)
Race (Whites as reference)				−0.142
Blacks	−0.050	−0.053	−0.141	(0.111)
Blacks	(0.196)	(0.196)	(0.111)	0.096
Hispanics	0.513**	0.517**	0.097	(0.124)
Hispanics	(0.171)	(0.171)	(0.124)	0.142
Other	−0.459	−0.460	0.135	(0.225)
Other	(0.493)	(0.493)	(0.223)	−0.013
Partnered	−0.818***	−0.813***	−0.010	(0.091)
Partnered	(0.130)	(0.130)	(0.091)	−0.984
Constant	−2.075	−2.070	−0.972	(0.658)
Constant	(1.074)	(1.074)	(0.654)	0.003
Log likelihood	−569.56	−569.46	−591.03	−591.58

Note. Robust standard errors in parentheses.

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

Acknowledgment

I thank colleagues from the Social Capital Group at Duke University for advice and assistance and the two anonymous reviewers for their comments. Data are drawn from the thematic research project “Social Capital: Its Origins and Consequences.” The principal investigators of the project are Nan Lin, Yang-Chih Fu, and Chih-Jou Jay Chen. This project is funded by Academia Sinica, Taiwan, through its Research Center for Humanities and Social Sciences, and the Institute of Sociology.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Notes

References

Bargh

J. A.

McKenna

K. Y. A.

(2004). The Internet and social life. Annual Review of Psychology, 55, 573–590.

Baym

N. K.

Zhang

Y. B.

Lin

M. C.

(2004). Social interactions across media: Interpersonal communication on the Internet, telephone and face-to-face. New Media & Society, 6, 299–318.

Bearman

P. S.

Parigi

(2004). Cloning headless frogs and other important matters: Conversation topics and network structure. Social Forces, 83, 535–557.

Boase

(2006). America online and offline: The relationship of personal networks to email and other communication media (PhD Thesis). Department of Sociology, University of Toronto, Toronto, ON, Canada.

Boase

(2008). Personal networks and the personal communication system. Information, Communication and Society, 11, 490–508.

Brooks

Welser

H. T.

Hogan

Tittsworth

(2011). Socioeconomic status updates: Family SES and emergent social capital in college student Facebook networks. Information, Communication & Society, 14, 1–22.

Chen

(2013). Growth or decline: Changes in Americans’ Social Capital. In Lin

Y. C.

Chen

C. J.

(Eds.) Social Capital and Its Institutional Contingency: A Study of the United States, Taiwan and China (pp. 259–280). London, UK: Routledge.

Chen

Boase

Wellman

(2002). The global villagers: Comparing Internet users and uses around the world. In Wellman

Haythornthwaite

(Eds.), The Internet in everyday life (pp. 74–113). Oxford, UK: Blackwell.

Chen

Tan

(2009). Understanding transnational entrepreneurship through a network lens: Theoretical and methodological considerations. Entrepreneurship Theory and Practice, 33, 1079–1091.

10.

Chen

Wellman

(2009). Net and jet: The Internet use, travel and social networks of Chinese Canadian entrepreneurs. Information, Communication and Society, 12, 525–547.

11.

Cho

Lee

J. S.

(2008). Collaborative information seeking in intercultural computer-mediated communication groups: Testing the influence of social context using social network analysis. Communication Research, 35, 548–573.

12.

Donath

Boyd

(2004). Public displays of connection. BT Technology Journal, 22, 71–82.

13.

Dunbar

R. I. M.

(2003). The social brain: Mind, language, and society in evolutionary perspective. Annual Review of Anthropology, 32, 163–181. doi:10.1146/annurev.anthro.32.061002.093158

14.

Ellison

N. B.

Steinfield

Lampe

(2007). The benefits of Facebook “friends:” Social capital and college students’ use of online social network sites. Journal of Computer-Mediated Communication, 12, Article 1. http://jcmc.indiana.edu/vol12/issue4/ellison.html

15.

Erickson

B. H.

(1996). Culture, class, and connections. American Journal of Sociology, 102, 217–251.

16.

Erickson

B. H.

(2004). The distribution of gendered social capital in Canada. In Flap

Volker

(Eds.), Creation and returns of social capital: A new research program (pp. 27–50). London: Routledge.

17.

Gergen

K. J.

(2008).Mobile communication and the transformation of the democratic process. In Katz

J. E.

(Ed.), Handbook of mobile communication studies (pp. 297–310). Cambridge, MA: MIT Press.

18.

Gil de Zúñiga

Valenzuela

(2011). The mediating path to a stronger citizenship: Online and offline networks, weak ties and civic engagement. Communication Research, 38, 397–421.

19.

Granovetter

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

20.

Hampton

K. N.

(2011). Comparing bonding and bridging ties for democratic engagement. Information, Communication & Society, 14, 510–528.

21.

Hampton

K. N.

Lee

C. J.

Her

E. J.

(2011). How new media affords network diversity: Direct and mediated access to social capital through participation in local social settings. New Media & Society, 13, 1031–1049.

22.

Hampton

K. N.

Sessions

L. F.

Her

E. J.

(2011). Core networks, social isolation, and new media. Information, Communication & Society, 14, 130–155.

23.

Hampton

Wellman

(2001). Long distance community in the network society—Contact and support beyond Netville. American Behavioral Scientist, 45, 476–495.

24.

Haythornthwaite

(2002). Strong, weak, and latent ties and the impact of new media. The Information Society, 18, 385–401.

25.

Howard

P. N.

(2004). Embedded media: Who we know, what we know, and the context of life online. In Howard

P. N.

Jones

(Eds.), Society online: The Internet in context (pp. 1–27). Thousand Oaks, CA: Sage.

26.

Hurlbert

J. S.

Beggs

J. J.

Haines

(2001). Social networks and social capital in extreme environments. In Lin

Cook

K. S.

Burt

R. S.

(Eds.), Social capital theory and research (pp. 209–231). New York, NY: Aldine de Gruyter.

27.

Ito

Okabe

Matsuda

(2005). Personal, portable, pedestrian: Mobile phones in Japanese life. Cambridge, MA: MIT Press.

28.

Jones

J. J.

Settle

J. E.

Bond

R. M.

Fariss

C. J.

Marlow

Fowler

J. H.

(2013). Inferring tie strength from online directed behavior. PLoS One, 8, e52168.

29.

Jones

Johnson-Yale

Millermaier

Perez

F. S.

(2009). US college students’ Internet use: Race, gender and digital divides. Journal of Computer-mediated Communication, 14, 244–264.

30.

Katz

Lazarsfeld

(1955). Personal influence: The part played by people in the flow of mass communications. Glencoe, IL: Free Press.

31.

Katz

Rice

R. E.

(2002). Social consequences of Internet use: Access, involvement, and interaction. Cambridge, MA: MIT Press.

32.

Kennedy

Smith

Wells

A. T.

Wellman

(2008). Networked families. Washington, DC: Pew Internet & American Life Project.

33.

Kraut

Kiesler

Boneva

Cummings

J. N.

Helgeson

Crawford

A. M.

(2002). Internet paradox revisited. Journal of Social Issues, 58, 49–74.

34.

Kraut

Patterson

Lundmark

Kiesler

Mukopadhyay

Scherlis

(1998). Internet paradox: A social technology that reduces social involvement and psychological well-being? American Psychologist, 53, 1017–1031.

35.

Lin

(2001). Social capital: A theory of social structure and action. New York, NY: Cambridge University Press.

36.

Lin

Dumin

(1986). Access to occupation through social ties. Social Networks, 8, 365–385.

37.

Lin

Y.-C.

Hsung

R.-M.

(2001). The position generator: Measurement techniques for investigations of social capital. In Lin

Cook

Burt

R. S.

(Eds.), Social capital: Theory and research (pp. 57–84). New York, NY: Aldine De Gruyter.

38.

Ling

(2008). New tech, new ties: How mobile communication is reshaping social cohesion. Cambridge, MA: The MIT Press.

39.

Long

J. S.

(1997). Regression models for categorical and limited dependent variables. Thousand Oaks, CA: Sage.

40.

Marlow

(2009, March 9). Maintained relationships on Facebook [Web log message]. Retrieved from http://overstated.net/2009/03/09/maintained-relationships-on-facebook

41.

Marsden

P. V.

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

42.

Marsden

P. V.

(1990). Network data and measurement. Annual Review of Sociology, 16, 435–63.

43.

Marsden

P. V.

Campbell

K. E.

(1984). Measuring tie strength. Social Forces, 63, 482–501.

44.

Mayhew

B. H.

Levinger

R. L.

(1976). Size and the density of interaction in human aggregates. American Journal of Sociology, 82, 86–110.

45.

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.

46.

McPherson

Smith-Lovin

Brashears

M. E.

(2009). Models and marginals: Using survey evidence to study social networks. American Sociological Review, 74, 670–681.

47.

McPherson

Smith-Lovin

Cook

J. M.

(2001). Birds of a feather: Homophily in social networks. Annual Review of Sociology, 27, 415–444.

48.

Mesch

Talmud

(2006). The quality of online and offline relationships: The role of multiplexity and duration of social relationships. The Information Society, 22, 137–148.

49.

Miyata

Kobayashi

(2008). Causal relationship between Internet use and social capital in Japan. Asian Journal of Social Psychology, 11, 42–52.

50.

Nie

N. H.

Erbring

(2000). Internet and society: A preliminary report. Stanford, CA: Stanford Institute for the Quantitative Study of Society.

51.

Olds

Schwartz

R. S.

(2009). The lonely American: Drifting apart in the twenty-first century. Boston, MA: Beacon Press.

52.

Peng

Zhu

(2010). A game of win-win or win-lose? Revisiting the internet’s influence on sociability and use of traditional media. New Media & Society, 11, 568–586.

53.

Pierce

(2009). Family time decreasing with internet use. Retrieved from http://uscnews.usc.edu/digital_media/family_time_decreases_with_internet_use.html

54.

Pollet

T. V.

Roberts

S. G. B.

Dunbar

R. I. M.

(2011). Use of social network sites and instant messaging does not lead to increased offline social network size, or to emotionally closer relationships with offline network members. Cyberpsychology, Behavior, and Social Networking, 14, 253–258. doi:210.1089/cyber.2010.0161

55.

Putnam

(1995). Bowling alone: America’s declining social capital. Journal of Democracy, 6, 65–78.

56.

Putnam

(2000). Bowling alone: The collapse and revival of American community. New York, NY: Simon & Schuster.

57.

Quan-Haase

Wellman

(2005). Hyperconnected network: Computer mediated community in a high-tech organization. In Heckscher

Adler

(Eds.), Collaborative community in business and society. New York, NY: Oxford University Press.

58.

Roberts

S. G. B.

Dunbar

R. I. M.

Pollet

T. V.

Kuppens

(2009). Exploring variation in active network size: Constraints and ego characteristics. Social Networks, 31, 138–146. doi:10.1016/j.socnet.2008.12.002

59.

Robinson

J. P.

Martin

(2010). IT use and declining social capital? Journal of Social Science Computer Review, 28, 45–63.

60.

Shklovski

Kiesler

Kraut

R. E.

(2006). The Internet and social interaction: A meta-analysis and critique of studies, 1995-2003. In Kraut

Brynin

Kiesler

(Eds.), Computers, phones and the Internet: Domesticating information technology (pp. 251–264). New York, NY: Oxford University Press.

61.

Son

(2013). Social capital and institutional constraints: A comparative analysis of China, Taiwan and the US. London, England: Routledge.

62.

Steinfield

DiMicco

J. M.

Ellison

N. B.

Lampe

(2009). Bowling online: Social networking and social capital within the organization. In Proceedings of the fourth International Conference on Communities and Technologies. New York, NY: ACM Press.

63.

Steinfield

Ellison

N. B.

Lampe

(2008). Social capital, self-esteem, and use of online social network sites: A longitudinal analysis. Journal of Applied Developmental Psychology, 29, 434–445.

64.

Turkle

(2011). Alone together: Why we expect more from technology and less from each other. New York, NY: Basic Books.

65.

Uzzi

(1996). The sources and consequences of embeddedness for the economic performance of organizations: The network effect. American Sociological Review, 61, 674–698.

66.

Valkenburg

Peter

(2009). The effects of instant messaging on the quality of adolescents’ existing friendships: A longitudinal study. Journal of Communication, 59, 79–97.

67.

van der Gaag

(2005). The measurement of individual social capital (PhD dissertation). University of Groningen, Groningen, The Netherlands.

68.

Veenhof

Wellman

Quell

Hogan

(2008). How Canadians’ use of the Internet affects social life and civic participation. Connectedness Series No. 16, Statistics Canada, Ottawa, ON, Canada.

69.

Vergeer

Pelzer

(2009). Consequences of media and Internet use for offline and online network capital and well-being. A causal model approach. Journal of Computer-mediated Communication, 15, 189–210.

70.

Wang

Wellman

(2010). Social connectivity in America: Changes in adult friendship network size from 2002 to 2007. American Behavioral Scientist, 53, 1148–1169.

71.

Wellman

Boase

Chen

(2002). The networked nature of community: Online and offline. IT Society, 1, 151–165.

72.

Wellman

Wortley

(1990). Different strokes from different folks: Community ties and social support. American Journal of Sociology, 96, 558–588.

73.

Zhao

(2006). Do internet users have more social ties? A call for differentiated analyses of Internet use. Journal of Computer-mediated Communication, 11, 844–862.