نقش انگیزش کاربران در ایجاد سرمایه اجتماعی و رفاه ذهنی: مورد بازی های شبکه های اجتماعی
|کد مقاله||سال انتشار||مقاله انگلیسی||ترجمه فارسی||تعداد کلمات|
|30100||2014||10 صفحه PDF||سفارش دهید||9482 کلمه|
Publisher : Elsevier - Science Direct (الزویر - ساینس دایرکت)
Journal : Computers in Human Behavior, Volume 39, October 2014, Pages 29–38
Social network games (SNGs)—which operate on a small scale and allow players to enjoy gaming with close friends—have exploded in popularity on the online social media scene in a very short time. This study explores the motivations that drive players to SNGs. The study investigates whether social capital serves as a moderating factor between these motivations and subjective well-being. Based on survey data (n = 560), the results show that SNG players seek entertainment, fantasy, the challenge of competition, and escapism when playing SNGs. The study finds that although social capital does not moderate the relationships between three motivations to play SNGs—entertainment, the challenge of competition, and escapism—and subjective well-being, it does moderate the relationship between the fantasy motivation and subjective well-being. Theoretical and practical implications and limitations are discussed.
The technology research firm Gartner reports that the global social gaming population includes at least 750 million players and is expected to double to 1.5 billion players by 2015 — a compound annual growth rate of 29 percent (GamblingData, 2012). Social network games (hereafter SNGs) have exploded in popularity on the online social media scene in a very short time (Casual Games Association, 2012). The explosive growth in popularity of social network sites (SNSs) such as MySpace and Facebook serves to highlight the value of rich social context in mediated environments (Kirman, 2010). SNGs represent a new media channel that amplifies social interaction between online users (Wohn, Lee, Sung, & Bjornrud, 2010). SNGs combine the entertainment function of existing mobile games and SNSs by connecting and stimulating interaction among users. This new media platform has opened the door to an era of SNG use that encompasses all socioeconomic classes (Yook & Ko, 2012). Marketers interested in leveraging these rapidly growing industries to attract customers seek to understand what motivates users in order to craft effective messages. Although there is a rich and rapidly growing body of literature on game players’ characteristics and motivations (Caplan et al., 2009, Dunne et al., 2010, Huffaker et al., 2009, Wohn et al., 2010 and Zhong, 2011), there has been very little research that focuses on the role of social capital as a moderating variable for subjective well-being in this environment. Building on a wide range of studies in the literature, this study builds a theoretical model to explain how factors that motivate individual SNG players reinforce their subjective well-being and also how social capital moderates the effects of these motivation factors on subjective well-being.
نتیجه گیری انگلیسی
5. Results 5.1. Measurement assessment and validation Although the measurement tools used in the study and based on the literature review are related to SNGs, it is important to test them for reliability and validity. For this purpose a multi-method approach to confirming the reliability and validity of the constructs was applied (McMillan and Hwang, 2002 and Sohn and Choi, 2013). The study first performed principal components factor analysis with varimax rotation on the initial items, employing a factor weight of 0.50 as the minimum cutoff value. It then examined the underlying factor structure to determine whether any new dimension within each dimension was conceptually meaningful and also to examine the psychometric properties of the scale. Throughout these stages, items can be deleted and refined to improve the reliability and validity of the scores on the scale (Hinkin, 1995). The factor analysis was executed by the maximum likelihood extraction method with varimax rotation. Varimax rotation was specifically used to identify variables that might indicate potential constructs, standardized factor loadings were examined at 0.5 and above on each potential construct, and one item with a factor loading below .05 was removed through this process (i.e., “I play SNGs because it stirs me up”). The result of a Bartlett’s test of sphericity was found to be significant (χ2 = 8856.9, df = 351, <0.01) while the Kaiser–Meyer–Olkin measure of sampling adequacy was 0.92, justifying application of the factor-analytic procedure. The first five factors extracted, in order of proportion of variance explained, were grouped as “social escapism,” “entertainment,” “challenge,” “fantasy,” and “social capital” to reflect the meanings and contexts of the corresponding items. A screen plot also supported the retention of the first five factors extracted. After collecting and cleaning the data, tests for internal consistency were run and item-to-total correlation was calculated and examined. Item-to-total correlation should be greater than .5, as suggested by Nunnally (1967). Correlation analysis was used to check for multicollinearity among the variables. The means of the squares of the correlation coefficients fell between 0.10 and 0.42, and the values of the variance inflation factor (VIF) for all of the variables satisfied the VIF level of 10, which suggests the presence of multicollinearity. The squared correlation coefficients between the dependent and independent variables were calculated to be 0.49 (the inter-item correlations do not exceed 0.70). All items used in the study satisfied the abovementioned criteria. The internal consistency of the measures was evaluated by Cronbach’s alpha, which was also used to test the reliability of this study’s instruments, and, as seen in Table 3, the scale reliabilities. For composite reliability to be adequate, a value of .70 or higher has been recommended (Nunnally & Bernstein, 1994). In all five constructs, Cronbach’s alpha exceeded the standard acceptance norm of .70. Because maximum likelihood estimation procedures were used in this study, it is important that the normality assumption not be severely violated (Curran, West, & Finch, 1996). The recommended range for the mean of skewness and kurtosis is ±1.96 (see Hair, Anderson, Tatham, & Black, 1998, for skewness and kurtosis standards; see Kline, 1998, and Joreskog, 1973, for guidelines for avoiding severe non-normality). To observe whether the standard was met, the observed variables’ skewness and kurtosis were examined: univariate skewness of the observed variables ranged from −.022 to .140, and univariate kurtosis ranged from −.086 to −.414. Therefore, the data used in this study can be regarded as fairly normal for the purpose of additional analyses. Fornell and Larker (1981) listed three procedures for assessing convergent validity. These are the item reliability of each measure, the composite reliability of each construct, and the average variance extracted (AVE; see Bagozzi and Yi, 1988 and Hair et al., 1998). Hair et al. (1998) suggested that an item is significant if its factor loading is greater than 0.50. As shown in Table 2, the factor loadings of all the items in the measure range from 0.58 to 0.91, thus meeting that threshold and demonstrating convergent validity at the item level. Discriminant validity was assessed by comparing the correlations of components with the AVE. For composite reliability to be adequate, a value of .70 or higher has been recommended (Nunnally & Bernstein, 1994). The final indicator of convergent validity is the AVE, which measures the amount of variance captured by the construct in relation to the amount of variance that is attributable to measurement error (Fornell & Larker, 1981). The AVE also satisfies the standard of 0.5, implying that the measurement indexes exhibit convergent validity. Table 2. Analysis of correlations associated with SNG users. Item number Escapism Entertainment Challenge Fantasy Social capital SWB Escapism 3 0.85a Entertainment 4 0.16⁎⁎ 0.79 Challenge 3 0.17⁎⁎ 0.25⁎⁎ 0.86 Fantasy 3 0.20⁎⁎ 0.27⁎⁎ 0.31⁎⁎ 0.88 Social capital 9 0.13⁎⁎ 0.11⁎⁎ 0.10⁎⁎ 0.10⁎⁎ 0.91 SWB 5 0.33⁎⁎ 0.20⁎⁎ 0.21⁎⁎ 0.21⁎⁎ 0.42⁎⁎ 0.86 Note: The correlation coefficients (n = 560) are squared. ⁎⁎ p < .05. SWB: subjective well-being. a Diagonal elements show Cronbach’s alpha. Table options Table 3. Construct item statistics for SNG-playing motivations. Factor Items Standardized Factor Loadings Skewness Kurtosis Escapism Es1 .84 −.096 −.089 (Mean: 3.57, S.D: .78) Es2 .77 −.047 −.675 Es3 .83 .023 −.368 Entertainment En1 .81 −.572 .416 (Mean: 3.31, S.D: .92) En2 .86 −.419 −.035 En3 .93 −.264 −.358 En4 .78 −.089 −.728 Challenge Ch1 .84 −.509 −.354 (Mean: 3.14, S.D: .91) Ch2 .84 −.179 −.390 Ch3 .79 −.726 .265 Fantasy Fa1 .79 −.162 −.329 (Mean: 2.46, S.D: .88) Fa2 .86 −.119 −.431 Fa3 .78 −.077 −.360 Social capital SC1 .76 −.389 −.565 (Mean: 3.77, S.D: .89) SC2 .79 −.328 −.498 SC3 .77 −.185 −.534 SC4 .76 −.446 −.331 SC5 .84 −.246 −.192 SC6 .85 −.364 −.196 SC7 .88 −.366 −.481 SC8 .89 −.453 −.266 SC9 .87 −.423 −.264 Cronbach’s alpha .85 .79 .86 .88 .91 C.R .91 .84 .92 .94 .95 AVE .78 .74 .79 .80 .82 C.R: composite reliability, AVE: average variance extracted. Table options In order to test for discriminant validity, the AVE of each of the two potential factors was compared with the square of the correlation between the two potential factors. As seen in Table 4, the means of the squares of the correlation coefficients (r2) are less than the AVE. The AVE should be greater than the means of the squares for all correlation coefficients ( Fornell & Larker, 1981), and for this data set the AVE falls between .74 and .82, and the means of the squares of the correlation coefficients fall between .10 and .42, indicating that AVE is greater than the means of the squares of the correlation coefficients (r2). The results of Convergent and Discriminant Validity indicted that AVE/r2 is greater than 1. This also satisfies the requirement of discriminant validity for research hypothesis model verification ( Bagozzi and Yi, 1988 and Fornell and Larker, 1981). Table 4. Results of convergent and discriminant validity. Construct AVE r r2 AVE/r2 C.V D.V Escapism .78 vs. Entertainment .754 .569 1.370 O O vs. Challenge .715 .511 1.415 O O vs. Fantasy .703 .494 1.579 O O vs. Social capital .756 .571 1.366 O O Entertainment .74 vs. Escapism .754 .569 1.300 O O vs. Challenge .698 .487 1.519 O O vs. Fantasy .656 .430 1.721 O O vs. Social Capital .714 .510 1.451 O O Challenge .79 O O vs. Escapism .715 .511 1.546 O O vs. Entertainment .704 .496 1.592 O O vs. Fantasy .687 .472 1.673 O O vs. Social Capital .709 .503 1.571 O O Fantasy .80 vs. Escapism .703 .494 1.619 O O vs. Entertainment .675 .456 1.754 O O vs. Challenge .676 .457 1.750 O O vs. Social Capital .711 .506 1.581 O O Social capital .82 vs. Escapism .756 .571 1.436 O O vs. Entertainment .645 .416 1.271 O O vs. Challenge .642 .412 1.990 O O vs. Fantasy .688 .473 1.734 O O Note: AVE: average variance extracted, r: correlation between factor of interest and remaining factors, r2: squared correlation factor interest and remaining factors, C.V: convergent validity (AVE > .50), D.V: discriminant validity (AVE/r2 > 1). Table options 5.2. Hypothesis tests Hierarchical moderated regression analyses were applied to test the hypotheses. The analysis was carried out in three steps. In the first step, the SNG players’ motivation-related factors were entered. In step 2 of each analysis, the centered NA was entered using social capital terms. In step 3, the study entered the interaction terms formed with SNG players’ motivation terms and the hypothesized moderator. Table 4 presents the results of the hierarchical moderated regression analysis for SNG players’ motivations. The result in step 1 shows that SNG players’ motivations are significantly related to subjective well-being (for escapism, β = 0.36, p < 0.05; for entertainment, β = 0.16, p < 0.05; for the challenge, β = 0.14, p < 0.05; and for fantasy-seeking, β = 0.10, p < 0.05, R2 = 0.41). The escapism, entertainment, challenge, and fantasy-seeking motivations are shown to have statistically significant effects on subjective well-being. Thus, H1-1, H1-2, and H1-3 were supported. Step 2 shows that social capital (β = 0.43, p < 0.05) is significantly related to subjective well-being in the SNG environment. In total, the two constructs entered in the second step explain 58% of the variance in subjective well-being beyond what could be explained by the SNG-related motivation factors. Further, step 3 discloses our interaction results. The interactions of three of the tested SNG players’ motivations with social capital were not significant predictors of subjective well-being (for escapism × social capital, β = 0.02, p > 0.05; for entertainment x social capital, β = 0.02, p > 0.05; for challenge × social capital, β = 0.04, p > 0.05). On the other hand, interactions of SNG players’ fantasy motivations with social capital were significant predictors of subjective well-being (for fantasy × social capital, β = 0.08, p < 0.05). Thus, of the four interaction hypotheses, only hypothesis H2-4 (regarding fantasy-seeking × social capital) was supported (see Table 5). Table 5. Hierarchical moderated regression results for subjective well-being as a dependent variable. Variables Step 1 Step 2 Step 3 β t B t B t Independent variables Escapism 0.36 10.54⁎⁎ 0.26 9.02⁎⁎ 0.18 1.40 Entertainment 0.16 3.76⁎⁎ 0.08 2.29⁎⁎ 0.14 0.93 Challenge 0.14 3.62⁎⁎ 0.11 3.25⁎⁎ 0.27 1.79 Fantasy 0.10 2.72⁎⁎ 0.07 2.18⁎⁎ 0.25 1.73 Moderator variable Social capital (SC) 0.43 15.3⁎⁎ 0.28 2.38⁎⁎ Interactions Escapism * SC 0.02 0.63 Entertainment * SC 0.02 0.50 Challenge * SC 0.04 1.13 Fantasy * SC 0.08 2.28⁎⁎ R2 0.41 0.58 0.59 F 98.2⁎⁎ 23.0⁎⁎ 2.15⁎⁎ ⁎⁎ p < .05. Table options 5.3. Additional analyses Additional tests were conducted on the relationship between the four motivation factors and social capital using regression analysis. The tests show that escapism and fantasy-seeking were not significantly related to social capital (for escapism, β = 0.06, p > 0.05; for fantasy-seeking, β = 0.03, p > 0.05). Tests revealed statistically significant relationships between entertainment and the challenge and social capital (for entertainment, β = 0.17, p < 0.05; for the challenge, β = −0.88, p < 0.05).