اشتراک گذاری دانش و رسانه های اجتماعی: نوع دوستی، درک انگیزش دلبستگی آنلاین و درک تعهد به رابطه آنلاین

Abstract Social media, such as Facebook and Twitter, have become extremely popular. Facebook, for example, has more than a billion registered users and thousands of millions of units of information are shared every day, including short phrases, articles, photos, and audio and video clips. However, only a tiny proportion of these sharing units trigger any type of knowledge exchange that is ultimately beneficial to the users. This study draws on the theory of belonging and the intrinsic motivation of altruism to explore the factors contributing to knowledge sharing behavior. Using a survey of 299 high school students applying for university after the release of the public examination results, we find that perceived online attachment motivation (β = 0.31, p < 0.001) and perceived online relationship commitment (β = 0.49, p < 0.001) have positive, direct, and significant effects on online knowledge sharing (R2 0.568). Moreover, when introduced into the model, altruism has a direct and significant effect on online knowledge sharing (β = 0.46, p < 0.001) and the total variance explained by the extended model increases to 64.9%. The implications of the findings are discussed.

Online social media have become increasingly popular in the last few years. The rapidly increasing use of social media for sharing information has also triggered a great deal of academic interest (Osatuyi, 2013). For example, there were 757 million daily active users of Facebook on average in December 2013, with 2.7 billion ‘likes’ made daily on and off the Facebook site and 300 million photos uploaded (Tam, 2012 and Facebook, 2014). In addition to having a very large user base, Facebook encourages frequent interaction among users through such things as the exchange of ‘likes’, comments, photos, tags, polling, events, inbox messages, and online chatting. These figures pose an interesting question: What motivates individuals to share information and interact with other users to such a significant extent in the social media environment? In particular, does this social interaction go a step further and contribute to knowledge sharing and hence knowledge creation? While some previous empirical studies have measured knowledge sharing in terms of participation and interaction (Kapur and Kinzer, 2007 and Mazzolini and Maddison, 2007), others have suggested that knowledge sharing is complicated and cannot be attained through social media due to the extent of social interaction (Liao, 2006, Wang and Noe, 2010, Ma and Yuen, 2011 and Ghadirian et al., 2014). The motivation for the present study was prompted by the idea that it would be good if the interaction among users in the social media environment led to knowledge sharing behavior, as this would be an important step in the process of knowledge creation. Few studies have examined the motivations for online knowledge sharing behavior (Ghadirian et al., 2014). This study aims to fill this gap in the research by exploring the motivational factors that affect knowledge sharing among individuals, with a specific focus on how interpersonal relationships influence such sharing in the social media environment. An existing online knowledge sharing framework is extended to investigate the motivational factors relating to knowledge sharing and to further identify whether altruism is a key determinant of such behavior. An alternative explanation of knowledge sharing in the social media environment is discussed, with particular reference to the theory of the need for belonging among online users.

4.1. Descriptive statistics of the respondents The descriptive statistics of the respondents are presented in Table 1 below. Table 1. Descriptive statistics of the respondents (N = 299). Items/descriptive Gender: Male – 215 (71.9%); Female – 84 (28.1%) Average Age: 17.96 Most frequently used: Facebook (89%); Others (11%) In the last week,… Mean (1–10)(Std.Dev) How often did you visit there? 6.77 (2.413) How often did you use the message inbox there? 4.03 (2.571) How often did you share news there? 2.82 (2.022) How often did you post messages to all friends there? 3.19 (2.349) How often did you chat there? 4.05 (2.494) How often did you upload photo(s) there? 3.01 (2.247) How often did you upload video(s) there? 1.83 (1.557) How often did you make comment(s) there? 4.20 (2.469) How often did you edit your profile there? 2.60 (1.913) How often did you share music there? 2.60 (2.152) Table options 4.2. Instrument validation The descriptive statistics of the means and standard deviations of each item are presented in Table 2. Cronbach’s alpha was used to validate the internal consistency of the instrument. All of the constructs exhibited internal consistency with alpha values greater than 0.7, as suggested by previous studies (Nunnally & Bernstein, 1994). The convergent validity of the instrument was examined by confirmatory factor analysis via AMOS v20. As shown in the above table, all of the factor loadings were significant at either the p < 0.01 or p < 0.001 level, thereby demonstrating the convergent validity of the items in relation to the construct. Table 2. Descriptive statistics and confirmatory factor loadings of the constructs. Min Max Mean Std. dev. Cronbach’s alpha Factor loadings Perceived Online Attachment Motivation (POAM) POAM1 1 7 3.58 1.307 0.869 0.734a POAM2 1 7 3.33 1.296 0.718⁎⁎⁎ POAM3 1 7 3.48 1.270 0.775⁎⁎⁎ POAM4 1 7 3.74 1.300 0.710⁎⁎⁎ POAM5 1 7 3.66 1.269 0.818⁎⁎⁎ Perceived Online Relationship Commitment (PORC) PORC1 1 7 3.89 1.286 0.853 0.717a PORC2 1 6 3.81 1.243 0.766⁎⁎⁎ PORC3 1 7 3.72 1.278 0.767⁎⁎⁎ PORC4 1 7 3.43 1.380 0.674⁎⁎⁎ PORC5 1 7 3.65 1.221 0.762⁎⁎⁎ Online Knowledge Sharing Behavior (OKSB) OKSB1 1 7 3.99 1.197 0.864 0.735a OKSB2 1 7 4.11 1.173 0.710⁎⁎⁎ OKSB3 1 7 3.90 1.126 0.794⁎⁎⁎ OKSB4 1 7 3.94 1.123 0.795⁎⁎⁎ OKSB5 1 7 3.85 1.126 0.713⁎⁎⁎ Altruism (ALT) ALT1 1 7 4.07 1.251 0.918 0.636a ALT2 1 7 3.66 1.111 0.671⁎⁎⁎ ALT3 1 7 4.05 1.262 0.670⁎⁎⁎ ALT4 1 7 4.06 1.237 0.730⁎⁎⁎ ALT5 1 7 4.10 1.237 0.656⁎⁎⁎ ALT6 1 7 4.03 1.246 0.720⁎⁎⁎ ALT7 1 7 3.66 1.214 0.697⁎⁎⁎ ALT8 1 6 3.61 1.172 0.699⁎⁎⁎ ALT9 1 7 3.69 1.254 0.676⁎⁎⁎ ALT10 1 7 3.75 1.180 0.694⁎⁎⁎ ALT11 1 7 3.73 1.169 0.711⁎⁎⁎ ALT12 1 7 3.74 1.172 0.685⁎⁎⁎ a Regression weight set to 1. ⁎⁎⁎ p < 0.001. Table options 4.3. Model testing results The corresponding hypotheses were examined using structural equation modelling via AMOS v20. The analysis followed a series of steps designed to compare competing models. First, the original model from previous studies (Ma & Yuen, 2011), which includes the PORC and POAM toward OKSB, was used to estimate the explanatory and predictive power of the causal relationships (see Fig. 2). All of the goodness-of-fit indices exceed the suggested required values (Hair, Black, Babin, & Anderson, 2010), indicating that the model fits the data well. Consistent with previous studies (Ma & Yuen, 2011), POAM (β = 0.310, p < 0.001) and PORC (β = 0.493, p < 0.001) were found to have significant, strong, and direct effects on OKSB. POAM (β = 0.750, p < 0.001) also had a significant, direct, and positive effect on PORC. The explanatory power of the model was examined using R2 for OKSB (R2 = 0.568) and was found to be significant and strong, and comparable with previous studies ( Ma and Yuen, 2011 and Ma et al., 2012) (see Fig. 1 and Table 4). Full-size image (16 K) Fig. 1. Testing of the online knowledge sharing model (competing model 1). Figure options Table 3. Summary of the testing of the extended online knowledge sharing model. Hypothesis Causal paths Coefficients Results Hypothesis H1a POAM → OKSB 0.189⁎ Supported Hypothesis H1b POAM → PORC 0.749⁎⁎⁎ Supported Hypothesis H2 PORC → OKSB 0.242⁎⁎ Supported Hypothesis H3 ALT → OKSB 0.456⁎⁎⁎ Supported R2: OKSB = 0.649 R2: PORC = 0.562 ⁎ p < 0.05. ⁎⁎ p < 0.01. ⁎⁎⁎ p < 0.001. Table options Table 4. A summary of the testing of the competing models. Causal paths Model 1 Model 2 Model 3 POAM → OKSB 0.310⁎⁎⁎ – 0.189⁎ POAM → PORC 0.750⁎⁎⁎ – 0.749⁎⁎⁎ PORC → OKSB 0.493⁎⁎⁎ – 0.242⁎⁎ ALT → OKSB – 0.743⁎⁎⁎ 0.456⁎⁎⁎ R2: OKSB 0.568 0.552 0.649 ΔR2 0.081 ⁎ p < 0.05. ⁎⁎ p < 0.01. ⁎⁎⁎ p < 0.001. Table options The second model was constrained to only the additional construct, Altruism. All of the goodness-of-fit indices exceed the suggested required values (Hair et al., 2010), indicating that this model also fits the data well. The explanatory power of this model is comparable with the first, with an R2 equal to 0.552. The causal path testing shows that Altruism had a positive, direct, and strong effect on OKSB (β = 0.743, p < 0.001) (see Fig. 2 and Table 4). Full-size image (11 K) Fig. 2. Testing of altruism on online knowledge sharing (competing model 2). Figure options For the final model, the first two models were combined to produce the extended online knowledge sharing model, which includes the effects of POAM, PORC, and Altruism on OKSB. The explanatory power of the model for individual constructs was again examined using R2 for OKBC. The results for the model with sample size N = 299 produced a list of goodness-of-fit indices, including TFI (0.923), CFI (0.933), and RMSEA (0.058). The values of these and all other indices exceed the suggested values, indicating that the model fits the data well ( Hair et al., 2010). The testing results are summarized in the figure and tables below (see Fig. 3, Table 3 and Table 4). Full-size image (19 K) Fig. 3. Testing of the extended online knowledge sharing model (competing model 3). Figure options Together, POAM, PORC, and Altruism explain 64.9% of the variance observed in OKSB. Altruism appears to contribute more to the observed explanatory power than the other constructs. At the same time, POAM accounts for 56.2% of the variance observed in PORC. The predictive power of the model was examined and the postulated hypotheses tested based on the path coefficients between the constructs. The data support most of the causal paths in the postulated model: POAM (β = 0.189, p < 0.05) has a positive, direct, and significant effect on OKSB. Thus, Hypothesis H1a and H1b is supported. The coefficient suggests that every standard unit increase in POAM will strengthen an individual’s OKSB by 0.189 units. However, as the above figures show, this predictive strength is significantly reduced when Altruism is included in the model. POAM also had a positive, direct, and significant effect on PORC (β = 0.749, p < 0.001). Thus, Hypothesis H2 is supported. PORC has a positive, direct, and significant effect on OKSB (β = 0.242, p < 0.01). Thus, Hypothesis H3 is supported. Again, the presence of Altruism seems to have had a strong mediating effect such that the strength of the effect of PORC on OKSB is reduced by half. Altruism has a positive, direct, and significant effect on OKSB (β = 0.456, p < 0.001). Thus, Hypothesis H3 is supported. In comparison, the effect of Altruism on OKSB is nearly double that of PORC and nearly triple that of POAM. To conclude, this final model fits the data well and provides better explanatory power than the previous competing models.