Operational definition of key concepts

國

立 政 治 大 學

‧

N a tio na

l C h engchi U ni ve rs it y

27

Given the above analysis, we are led to ask of the weak ties theory is indeed applicable to Chinese online social networks? The question leads to our first hypothesis,

H1: The importance of weak ties as source of information is significantly greater than that of strong ties.

Our analysis above also shows that tie strength is conceptually different from guanxi, yet in Chinese societies guanxi plays a major role in social networks and resources are often shared among in-group members in the same guanxi network (Shin, Ishman, & Sanders, 2007).

Trust and communication were also found to be two important elements in information diffusion (Ramasamy, Goh, & Yeung, 2006).

This leads us to question if Chinese Facebook users‟ guanxi with strong ties and that with weak ties is different, and if guanxi plays a role in selecting counterpart for information exchange:

H2 There is significant difference between Facebook users‟ guanxi with strong ties and their guanxi with weak ties, and

H3: Strong guanxi positively predicts respondents‟ selection of strong ties in exchanging information.

H4: Weak guanxi negatively predicts respondents‟ selection of weak ties in exchanging information.

3.2 Operational definition of key concepts

Tie strength

A “tie” in this study refers to a “contact person” or a “Friend” on respondents‟ Facebook Friend list. Since the publication of Granovetter's weak ties theory (1973), the volume of research using tie strength concept to examine information resources exchange in social networks are overwhelmingly larger than empirical studies attempted to measure or test the concept (Mathews, White, Soper, & von Bergen, 1998; Petróczi, Nepusz, & Bazsó, 2007).

Likewise, different dimensions of tie strength have been discussed widely for social network

‧ 國

立 政 治 大 學

‧

N a tio na

l C h engchi U ni ve rs it y

28

research in general, rather than used specifically for examining relations of online social network users.

Most of the primary indicators of tie strength in the existing literature of social ties or network research were developed in offline context. (Petróczi, et al., 2007) summarized eleven indicators including primary indicators such as frequency, intimacy/closeness, as well as indicators which were infrequently used were specific to offline context, for example, desire for companionship, and multiple social context (breadth of topics), they were therefore inapplicable in this study. “Virtual communities are created/maintained and held together by computer-mediated communication (CMC), therefore components

such as help provided and received, time spent together or even communication may have different meanings” (Petróczi, et al., 2007, p.41). Despite the abundant literature on tie strength research, therefore, there is no standardized scale available.

Emotional intensity/closeness is seen as most distinctive indicator of tie strength for some studies (e.g. (Marsden & Campbell, 1984). However, other researchers reached contradicting conclusions. “Those studies that measure ties and tie strength using the single word

„knowing‟ may misinterpret the composition of tie,” (McCarty, 1996, p. 14) pointed out. He questioned the predictive power of closeness, as closeness /knowing may be understood differently by respondents (McCarty, 1996).

A considerable number of tie strength studies were targeted on finding predictors and indicators of tie strength (e.g. (Mathews, et al., 1998), rather than to quantify tie strength.

It is however important to quantitatively distinguish strong ties from weak ties, because only then the claims and theories which rely on the concept of tie strength can be tested

(Granovetter, 1973, Petróczi, et al., 2007).

Empirical studies focusing on tie strength in virtual communities are rather limited (Petróczi, et al., 2007). Among the few, there are two studies relevant to the current study. Muncer et al.

(Muncer, Loader, Burrows, Pleace, & Nettleton, 2000) defined tie as “having at least one posting between two participants and used the number of postings on each strand and frequency to indicate strength” (Petróczi, et al., 2007, p.41).

‧ 國

立 政 治 大 學

‧

N a tio na

l C h engchi U ni ve rs it y

29

Actual interaction between users had been used to distinguish strong ties and weak ties on Facebook in the research by Viswanath, Mislove, Cha, & Gummadi (2009). They found that the median number of posts per user pair is two in the first year the link was established and 81% of the pairs exchange less than five wall posts in a year. They divided the users into two groups: high and low rate interaction. The average number of posts exchanged between high-rate interaction groups is 16.2, and the median number is 10 posts. The user pairs may also interact via messages, photo sharing, application, and chat (Viswanath, et al., 2009).

There are two major levels of analysis for network study, individual actors, and actor and network. Organization represents a higher level of analysis than individuals in terms of scope and complexity of the entities being examined. This macro perspective may not apply to the study of online social network information exchange, as it is individual-based and considered micro-level (Borgatti & Foster, 2003).

Tie strength in this study is measured from the dyadic perspective, meaning the intensity is viewed for each pair of individuals (e.g. where A and B are friends or not; the frequency of contact between them), rather than for each node (e.g. age or gender of each actor from the perspective of network structure) (Borgatti & Foster, 2003).

Taking the above into consideration, “frequency of contact” is used here as it is clear and hence easier for respondents to understand and respond to questions. “Strong ties” in this study are defined as user pairs that interact three times or more per week via Facebook, by way of wall posts, comments, video and application sharing, chat, or messages. User pairs that exchanged less than five wall posts in a year were considered low interaction in research (Viswanath, et al., 2009). Thus, “weak ties” in this study is defined as user pairs that interact once in three months, via wall posts, comments, video and application sharing, chat, or messages.

Information exchange

Information exchange is dyadic communicating behavior that diffuses information. In this study questions relating to such exchanges were targeted at primarily information on job opportunities, the same as that Granovetter used in his 1973 study. The focus on job-related

‧ 國

立 政 治 大 學

‧

N a tio na

l C h engchi U ni ve rs it y

30

information helps respondents to relate to the situation described in the question and properly respond to it. But in this study, leisure activities information was also added to see if different types of information would entail different patterns of behavior.

Guanxi

For the present study, guanxi is viewed strong societal relationship involving favor, mutual benefit, reciprocity and personal connection (Chiu, et al., 2007). According to the past research, trust is also an important component in guanxi. Thus “trust” is added into the guanxi scale (Osland, 1990; Ramasamy, et al., 2006). The guanxi scale in this study was developed on the basis of the above components, and the items from guanxi scale by Zhuang, Xi, & Tsang (Zhuang, Xi, & Tsang, 2010). But as the Zhuang, et al. (2010) study was designed to examine organizational guanxi behavior, some of the items had been removed or rephrased to adapt to the purpose of this study, for example, “If not for my company, I would rather not have a connection with him (them).”