These observations clearly indicate that mobile texting is used to bond with core ..... fact that teens' frequency of voice calling using mobile phones is declining .... in the same way for the adolescents (alpha = 0.83, Range: 3â12, Mean = 7.12,.
Journal of Computer-Mediated Communication
Tele-Cocooning: Mobile Texting and Social Scope∗ Tetsuro Kobayashi National Institute of Informatics, 2-1-2 Hitotsubashi Chiyoda-ku, Tokyo, 101-8430, Japan
Jeffrey Boase Ryerson University, School of Professional Communication, 350 Victoria Street, Toronto, Ontario, M5B 2K3
The present study examines the tele-cocooning hypothesis in the context of general trust using a nationally representative survey of Japanese youth. We find that although frequency of texting is positively correlated with general trust, this correlation is spuriously caused by how heavy mobile texters interpret the words ‘‘most people’’ in the general trust measurement. Heavy users assume that ‘‘most people’’ only refers to friends, family, and others going to the same school. When the effect of the ‘‘most people’’ assumption is controlled, the positive association between texting and general trust disappears. Further exploration of the data shows that heavy texting nevertheless has negative implications for social tolerance and social caution, both of which are theoretically proximate to general trust. Key words: Mobile communication, general trust, mobile texting, tele-cocooning, social scope. doi:10.1111/jcc4.12064
With the increasing functionality and widespread use of mobile phones, the need for studies on the social consequences of mobile phone use is now widely recognized. Using the tele-cocooning hypothesis, we examine the possible negative relationship between mobile texting (hereafter, texting) and general trust. The tele-cocooning hypothesis states that texting is associated with increasingly insular communication because it strengthens core ties at the expense of interactions with lesser-known weak ties. In this study, we focus on two manifestations of such insular communications: narrow social scope and low general trust. Although mobile phones can be used in a variety of ways, we examine texting in this study because it is how young people most frequently use mobile phones (Lenhart, 2012). We investigate these issues using a sample of Japanese youth, who have been among the earliest and heaviest adopters of texting in the world. Networked individualism, which is accelerated by the widespread use of mobile phones, requires that people develop new strategies and skills for handling problems (Rainie & Wellman, 2012). General trust serves as an important element of these strategies and skills because those distrustful of others cannot effectively make use of social capital embedded in human relationships (Yamagishi, 2011). Despite the great popularity of mobile communication, Japan has the lowest levels of general trust among East Asian countries. For example, results of the 2007 Pew Global Attitudes survey showed that 79% of respondents living in China agreed that ‘‘most people in ∗ Accepted
by previous editor Maria Bakardjieva
Journal of Computer-Mediated Communication 19 (2014) 681–694 © 2014 International Communication Association
681
society are trustworthy,’’ while only 43 percent of respondents living in Japan agreed with the same statement (Pew Research Global Attitudes Project, 2008). Why is mobile phone use in Japan so high when general trust is so low? In this study, we address this question by demonstrating that mobile texting among Japanese youth brings about tele-cocooning which narrows their social scope and lowers the level of general trust.
Tele-Cocooning: Insularity of mobile communication Although texting can potentially be used to maintain and build new relationships, research from a variety of sources indicates that texting during youth typically involves the intensive exchange of short messages among intimate and homogeneous peers (e.g. Campbell & Kwak, 2012a; Ishii, 2006; Ling, 2004, 2008; Wilken, 2011). Typical Japanese high school students exchange text messages with two to five close friends, and these exchanges tend to take place in more geographically limited areas than do e-mail exchanges using personal computers (Miyata, Wellman, Boase, & Ikeda, 2005). With a few notable exceptions (e.g. Boase & Kobayashi, 2008), the selectivity associated with mobile communication is believed to support and strengthen existing ties rather than build new ones (Ling, 2008). These observations clearly indicate that mobile texting is used to bond with core peer groups. The distinctive bonding nature of mobile texting has prompted concern about a dark side of mobile communication. If texting facilitates contact with close friends and family exclusively, and if it reduces heterogeneous encounters, it can have a cocooning effect by focusing one’s attention on existing close relationships at the expense of reaching out to new people (Campbell & Kwak, 2012a; Gergen, 2008; Habuchi, 2005; Ling, 2008). Put differently, if we assume that the resources people have to spend on personal communication are finite, the tele-cocooning hypothesis predicts that continual contact via texting strengthens existing strong ties at the expense of interactions with others who are less familiar. Although concerns about tele-cocooning appear repeatedly in the mobile communication literature, empirical tests of tele-cocooning have been limited to its effects on social networks. Campbell and Kwak (2012a) demonstrated that mobile-based political discussion with a small number of strong ties who mutually affirm political views may be associated with political demobilization. Similarly, Campbell and Kwak (2012b) showed that levels of open dialogue between people in small and like-minded strongtie networks decreased slightly when political discussion was mediated by mobile phone use. Note, however, that a series of studies by Campbell and Kwak (2012a, 2012b) indicate that mobile-mediated discussion in large, like-minded, strong-tie networks may foster both open and active citizenship. On the other hand, Ling (2008) demonstrated that mobile communication positively correlates with the social cohesion of strong ties, and he also pointed out the possibility that mobile phone use supports the cultivation of strong ties at the expense of important weak ties.
Texting and General Trust If texting is associated with the insularity of communication among strong ties with like-minded others, how is texting related to trust, especially general trust? General trust is a general belief in human benevolence. It suggests that trustworthiness is an aspect of human nature and that most people can be trusted despite some exceptions. High general trust is useful in situations with high opportunity costs and social uncertainty (Yamagishi & Yamagishi, 1994; Yamagishi, 2011). Yamagishi and Yamagishi (1994) demonstrated through large-scale surveys and a carefully designed series of experiments that general trust is higher in the U.S. than in Japan1 . Limited 682
Journal of Computer-Mediated Communication 19 (2014) 681–694 © 2014 International Communication Association
social mobility and a longer duration of committed relationships in Japan make both the opportunity cost of staying in stable relationships and the incentive for engaging in opportunistic behaviors low. If Japanese people behave opportunistically in their exclusive networks of committed relationships, they are likely to be ostracized. Furthermore, once ostracized, it is difficult to join other networks because those are exclusive as well, which makes the social consequences of being ostracized quite severe. The consequence of this extreme dependency on existing relationships implies that individuals can be ‘‘assured’’ that their close ties of friends, family, and school or workmates, will not behave opportunistically. Because people embedded in a tight web of committed relationships do not have to judge the trustworthiness of others, their level of general trust is low. To put it in Yamagishi’s (2011) words, ‘‘assurance destroys general trust.’’ Conversely, when social mobility is high, as it is in the U.S., opportunity cost increases because staying inside a network of stable, committed relationships makes it difficult to take advantage of outside opportunities. Having the social intelligence to judge others’ trustworthiness and to develop new, cooperative relationships is adaptive under such conditions. As a result, general trust is a form of social intelligence that allows individuals to cope with greater social mobility and uncertainty (Yamagishi, 2011). The committed relationships Yamagishi (2011) describes roughly correspond to the concept of strong ties, because both ideas are characterized by stability and mutually beneficial cooperation. Accordingly, if texting has a tele-cocooning effect, it should facilitate the formation of committed relationships. Hence, it is predicted that heavy texters will have lower levels of general trust.
Measurement of General Trust One possible threat to research on this topic is the measurement of general trust. In most cases, general trust is measured with a standard question: ‘‘Generally speaking, would you say that most people can be trusted or that you need to be very careful in dealing with people?’’ Similarly, Yamagishi and Yamagishi (1994) used multiple items to measure general trust, such as, ‘‘Most people are trustworthy,’’ ‘‘Most people trust others,’’ and ‘‘Most people are basically good and kind.’’ It is noteworthy that each of these items uses the phrase ‘‘most people,’’ and as Grootaert and Bastelaer (2002) have pointed out, the meaning of ‘‘most people’’ can vary across respondents, which makes the interpretation of such measures ambiguous (Delhey, Newton, & Welzel, 2011; Sturgis & Smith, 2010). Because the measured level of trust will naturally be higher when known others (rather than anonymous strangers) are the assumed referents, general trust will be overestimated when this occurs. Therefore, to assess the association between texting and general trust, the scope of ‘‘most people’’ should be fixed across respondents. This is not an easy task, however, because it may be that how one interprets the phrase ‘‘most people’’ is itself correlated with mobile phone use.
Hypotheses If texting leads to network insularity, as the tele-cocooning hypothesis maintains, then heavier texting should be associated with a narrower definition of ‘‘most people.’’ This process could take place in two ways. First, because intensive texting with close others inside cocoons takes time, it may limit or reduce interactions with more distant others. In this case, ties with people outside of cocoons are more likely to decay, leading to decreased network size among heavy texters. This could lead to an overreferencing of close others when heavy texters answer general trust questions that use the words ‘‘most people.’’ Journal of Computer-Mediated Communication 19 (2014) 681–694 © 2014 International Communication Association
683
Second, texting may not shrink users’ personal networks into small cocoons by decaying weaker ties, but it may cognitively limit users’ social horizons. Intensive texting to those one has strong ties with may chronically increase their cognitive accessibility, keeping them at the foremost of one’s mind. In this case, when heavy texters are asked about ‘‘most people,’’ they may tend to think of those they text frequently as exemplars because these ties are readily accessible in memory. Given these two possible scenarios, the first hypothesis is: H1: Frequency of texting is negatively associated with the perceived scope of ‘‘most people.’’ If H1 is supported, a positive but spurious association between texting and general trust would be detected. Even if texting and general trust have no meaningful connection, the reported general trust of heavy texters would be higher merely because they are more likely to imagine known others (rather than anonymous strangers) as the referents of ‘‘most people.’’ For this reason, our second hypothesis states: H2: The frequency of texting is positively associated with levels of general trust when the perceived scope of ‘‘most people’’ is not controlled. As previously noted, the relationship in H2 is spurious if H1 is supported. If the scope of ‘‘most people’’ is held constant, then the tele-cocooning hypothesis predicts that the association between texting and general trust will be negative because the insular communication in committed relationships that is strengthened by texting will decrease general trust. Thus, we hypothesize that: H2a: The positive association between the frequency of texting and levels of general trust becomes negative when the perceived scope of ‘‘most people’’ is controlled. These three hypotheses test the implications of the tele-cocooning hypothesis for general trust and directly address issue of validity in the general trust measure. As noted, however, the standard measurement of general trust is not soundly appropriate to test the tele-cocooning hypothesis. For this reason we utilize two other dependent variables that do not use the phrase ‘‘most people’’ to measure, but are nonetheless theoretically proximate to general trust. We use the Caution Scale (Yamagishi & Yamagishi, 1994) and the Social Tolerance Scale (Ikeda & Kobayashi, 2008; Kobayashi, 2010), both of which are relevant to general trust measurement while circumventing the ‘‘most people’’ wording. People who generally trust others do not have to be overly cautious because they can effectively cooperate with heterogeneous others, and because they are socially intelligent enough to distinguish trustworthy from untrustworthy people (Yamagishi, 2011). As expected, Yamagishi and Yamagishi (1994) found a negative correlation between general trust and caution in dealing with others. Therefore, if texting is negatively associated with general trust, it will be positively associated with caution in dealing with others. H3a: Texting is positively associated with caution in dealing with others. On the other hand, because those with higher general trust can develop cooperative relationships with a diverse range of people, they do not have to exclude potential partners in social exchanges who have different values or attitudes. People from outside of one’s committed relationships tend to have different views and opinions, and they are a source of non-redundant information and new opportunities (Granovetter, 1973). If those with high general trust are good at cooperating with heterogeneous others from outside their intimate circle, they must also be socially tolerant of others in order to reduce opportunity costs. Accordingly, if texting is negatively associated with general trust, it will also be negatively associated with social tolerance. H3b: Texting is negatively associated with social tolerance. 684
Journal of Computer-Mediated Communication 19 (2014) 681–694 © 2014 International Communication Association
Method Survey Targeting Japanese adolescents aged from 8 to 18 years, a nationally representative drop-and-collect survey was fielded in the summer of 2009.2 First, all cities in Japan were organized by region and population size into 62 strata. Next, 100 primary sampling units (PSU) were randomly selected in each stratum in proportion to its population size. Then, 17 target respondents were randomly sampled in each PSU using basic registries of residents. This is a standard method that is widely used in Japan to produce representative samples. Of the 1,700 target respondents, valid responses were received from 1,002 (59%) individuals. To include school-related variables in the analyses, about 2.5% of respondents who were not in school (i.e., who were 18 years old and had graduated from high school at the time of the survey) were excluded from these analyses. Using 2005 census information, the sample was weighted using poststratification weights calculated from distributions of sex, age, and the population of each stratum. Measurement General Trust General trust was measured by summing the scores for three items from the General Trust Scale (Yamagishi & Yamagishi, 1994). The items ‘‘Most people are trustworthy,’’ ‘‘Most people trust others,’’ and ‘‘Most people are basically good and kind’’ were measured using 4-point scales (α = 0.85, Range: 3–12, Mean = 7.98, SD = 0.05). Immediately after responding to the general trust items, a follow-up question asked respondents to indicate what ‘‘most people’’ meant, using the prompt: ‘‘What kind of people did you imagine as ‘most people’ when answering the three items above [general trust items]? Please circle all that apply below.’’ The multiple-choice format for this question had eight possible responses: ‘‘All of humanity, not only Japanese people,’’ ‘‘All Japanese people,’’ ‘‘People living in the same prefecture,’’ ‘‘People living in the same city, ward, town, or village,’’ ‘‘People living in the neighborhood (within an approximately 10-minute walk from my home),’’ ‘‘People going to the same school or working at the same place of work,’’ ‘‘Friends or acquaintances,’’ and ‘‘Parents, brothers, sisters, or other relatives.’’ Caution in Dealing With Others Caution in dealing with others was measured by summing the scores for the following three items from the Caution Scale (Yamagishi & Yamagishi, 1994): ‘‘People are only interested in their own welfare,’’ ‘‘In this society, one has to be alert or someone will take advantage of you,’’ and ‘‘In this society, one must be constantly afraid of being cheated.’’ The items were measured using 4-point scales (α = 0.72, Range: 3–12, Mean = 8.15, SD = 0.06). A larger value indicates greater caution in dealing with others. The correlation with general trust was –0.17 (p < .01). Social Tolerance. Social tolerance was measured using the following item from the Social Tolerance Scale (Ikeda & Kobayashi, 2008; Kobayashi, 2010): ‘‘What do you think when you and the following people have different views or opinions?’’ The wording was slightly modified to suit adolescent respondents, who answered this question for their ‘‘Father,’’ ‘‘Mother,’’ ‘‘Close friends of the same generation,’’ and ‘‘Acquaintances of the same generation’’ using a 4-point scale: 4 = No problem having different opinions; 3 = Not a big problem having different opinions; 2 = I would prefer that we have the same Journal of Computer-Mediated Communication 19 (2014) 681–694 © 2014 International Communication Association
685
opinions; and 1 = It would be better if we had the same opinions. Social tolerance scores were figured by summing across responses for the four referents (α = 0.84, Range: 4–16, Mean = 11.91, SD = 0.10). The correlation with general trust was 0.00 (n.s.)3 .
Frequency of Mobile Texting The frequency of outgoing and incoming mobile texting on an average day was measured using the following 6-point scale (one for outgoing, one for incoming): 1 = less than one message; 2 = 1–5 messages; 3 = 6–10 messages; 4 = 11–25 messages; 5 = 26–50 messages; 6 = 51 messages or more. The incoming and outgoing scales were summed to create a single measurement of the frequency of mobile texting (α = 0.96, Range: 2–12, Mean = 6.02, SD = 0.12). Control Variables The effects of generalized reciprocity (Putnam, 2000), participation in voluntary associations (Brehm & Rahn, 1997), and the general trust of guardians (Sturgis et al., 2010) were controlled because they are major sources of the general trust of adolescents.4 The general trust of guardians was available because this survey collected data from adolescents and their guardians. Sex, age, the type of school (elementary school, junior high school, high school, or college), and whether the school was highly ranked were also included as control variables. School rank was included because a previous study in Japan indicated a negative correlation between school rank and general trust among high school students (Kobayashi & Ikeda, 2007).
Results About 51% of the sample reported using mobile phones, with a great deal of age-related variability. Six percent of 8-year-olds, 17% of 11-year-olds, and 33% of 12-year-olds reported using mobile phones. The jump between ages 11 and 12 reflects the fact that many adolescents start to use mobile phones when they enter junior high school. Likewise, there is another jump between ages 14 (48%) and 15 (82%), which is when students enter high school. One hundred percent of the 18-year-olds in our sample reported using mobile phones. About 97% of mobile phone users in this sample were texters. The frequency distribution for outgoing mobile texting indicated that 11% of the sample averaged less than one text message a day, 35% sent 1–5 messages, 21% sent 6–10 messages, 16% sent 11–25 messages, 12% sent 26–50 messages, and 5% sent 51 messages or more. The frequency distribution for incoming mobile texting indicated that 11% of the sample averaged less than one message a day, 34% sent 1–5 messages, 22% sent 6–10 messages, 15% sent 11–25 messages, 14% sent 26–50 messages, and 5% sent 51 messages or more. H1: Frequency of texting is negatively associated with the perceived scope of ‘‘most people.’’ The ‘‘full sample’’ bars in Figure 1 show how respondents interpreted the term ‘‘most people’’ in General Trust Scale items. About 20% of the respondents endorsed the two most general interpretations (‘‘All of humanity’’ and ‘‘All Japanese people’’) that are the theoretical targets of the General Trust Scale. However, more than 50% of the respondents endorsed ‘‘Friends or acquaintances,’’ and about 40% interpreted ‘‘most people’’ to mean ‘‘People going to the same school’’ or ‘‘family members.’’ This result at least partially reflects the fact that the sample is limited to adolescents. Nevertheless, this variability in the interpretation of ‘‘most people’’ corroborates the findings from previous studies that cast doubt on the theoretical targets of the General Trust Scale (e.g. Grootaert & Bastelaer, 2002). 686
Journal of Computer-Mediated Communication 19 (2014) 681–694 © 2014 International Communication Association
0
20
40
60
80
100
All of humanity, not only Japanese people All Japanese people People living in the same prefecture People living in the same city, ward, town, or village
Full sample Narrow scope cluster Broad scope cluster
People living in the neighborhood People going to the same school Friends or acquaintances Parents, brothers, sisters, or other relatives
Figure 1 How respondents interpreted the phrase ‘‘most people’’ (MA, %)
Next, to test H1, a cluster analysis was performed to categorize the respondents according to their interpretation of ‘‘most people.’’ Defining similarity as Pearson’s phi coefficients, a hierarchical cluster analysis employing Ward’s method generated two clear clusters, which are labeled ‘‘Narrow scope cluster’’ and ‘‘Broad scope cluster’’ in Figure 1. The most prominent feature of the narrow scope cluster (n = 666) is the high percentage of respondents who endorsed ‘‘Friends or acquaintances,’’ ‘‘People going to the same school,’’ and ‘‘Parents, brothers and sisters, or relatives.’’ Those in the broad scope cluster (n = 336) interpreted the ‘‘most people’’ phrase to mean ‘‘All of humanity’’ or ‘‘All Japanese people.’’ This means that about two thirds of the respondents interpreted ‘‘most people’’ more narrowly than the intended interpretation used in the development of the General Trust Scale. Next, a logistic regression was performed using cluster (0 = Narrow scope cluster and 1 = Broad scope cluster) as the dichotomous dependent variable (Table 1). The N for the analysis was reduced to 437 because we analyzed the subsample of respondents who used mobile phones. Despite the extensive controls, the frequency of mobile texting had a significant negative association with the scope of ‘‘most people.’’ That is, the more frequently adolescents texted, the more likely they were to belong to the narrow scope cluster, which supports H1. H2: The frequency of texting is positively associated with levels of general trust when the perceived scope of ‘‘most people’’ is not controlled. H2a: The positive association between the frequency of texting and levels of general trust becomes negative when the perceived scope of ‘‘most people’’ is controlled. To test H2 and H2a, ordinary least squares (OLS) regressions were conducted using general trust as the dependent variable (Table 2). To test H2, the magnitude of unmediated association between mobile texting and general trust was estimated without controlling for the scope of ‘‘most people’’ (Model 1). To test H2a, the dichotomous clusters of assumed ‘‘most people’’ was included as an independent variable to see whether the coefficient of mobile texting would be negative (Model 2). Journal of Computer-Mediated Communication 19 (2014) 681–694 © 2014 International Communication Association
687
Table 1 Logistic regression model predicting the interpretations of ‘‘most people’’ Dependent variable: Scope of ‘‘most people’’ (0: Narrow scope cluster, 1: Broad scope cluster) coef. (B) Sex Age Junior high school High school Some college Academic-track school General trust of the guardian Participation in voluntary association Generalized reciprocity Frequency of mobile texting Constant N Pseudo R-squared
−0.54∗∗ (0.19) 0.01 (0.09) 1.92∗ (0.80) 2.33∗ (0.95) 2.32† (1.16) −0.02 (0.08) −0.16∗ (0.06) −0.14∗∗ (0.05) 0.03 (0.03) −0.09∗ (0.04) 0.36 (1.15) 437 0.08
∗∗
P