The second guideline assumes that a large number of indicators will yield many non-significant weights. Due to the fact that our measurement model consists of formative indicators, this test may be irrelevant. Guideline three assumes the co-occurrence of negative and positive indicator weights, which could lead to a misinteny the results In our case, only positive weights were observed (Table 3), and the suppressor effect was thus not tested The fourth guideline discusses absolute versus relative indicator contribution. Indicators with a non-significant or low weight can still have an important absolute ontribution. All related indicators must be independently assessed from other indicators to prevent misinterpretation of formative indicator results. As Table 3 shows ght of RESI is not significant. Other formative construct indicators are significant at the 0.01 level. It suggests that they contribute significantly to the form Attention Switching and Parental Monitoring composites. Overall while the statistical evidence regarding the validity of the formative conceptualization is not always conclusive, when synthesized with the theoretical ba nd opinions of subject-matter experts, it points to potential plausibility. We hence proceed with treating these concepts as formative composites As with all self-reported data, there is a potential for common method biases(Podsakoff et al., 2003). We performed statistical analyses to assess the potential sev problem in our data. First, Harmons one-factor test was performed. In a Principal Component Analysis with no rotation 12 components emerged; and the first compo explained only 10.5%of the variance. Second, following Podsakoff et al. (2003), we included in the PLS model a latent common method factor whose indicators incl items. We then calculated the variance explained in our endogenous construct, online game addiction, with- and without- the latent common methods variance factor odel. The first was 0.418, and the second was 0.408. The difference is very small. Both tests point to the conclusion that the method is unlikely to be a major source varation Hypothesis Testing Table 4 presents the estimates obtained from PLS analysis( Bootsrapping with 200 re-samples)3]. An R- value of 408 indicates that the model explains a substa amount of variance in addiction. The results provide some support for the hypothesized partial - mediation role of game playing and the direct effects of functional nee otivating factors, and prevention and harm reduction factors on the formation of online game addiction. Among the motivation factors, need to master the mechanic was positively associated with game playing but not with game addiction. As expected, needs for relationship and escapism were positively associated with game addiction. Need for advancement had no significant influence on addiction Table 4. Test of Hypotheses t-value Need for Rela eed for Escapism → Game Playing ime Playing Rationalization Education- Game Playing 3 cost→ 9 Perceived Co→Am **P<.001;种P<01;P<0 The data suggest that attention switching has a significant negative impact on game playing and addiction. It implies that alternative activities could distract adol attention from online games and thus reduce their risk of high levels of addiction. Furthermore, rationalization/ education and cost had significant influences on game no direct effect on online game addiction. Thus, they can alleviate the level of addiction through the reduction of ones online play time Dissuasion was expected to reduce game playing and addiction. Nevertheless, contrary to the expectation, it was positively associated with game playing and on diction, It seems that dissuasion does not serve as a prevention factor but turns out to be more of a remedy which is plausibly exercised only after high levels of gan d online game addiction are observed. We also found that Resource Restrictions was positively associated with addiction and had no significant impact on game pl expected that when fewer resources are available, people will be less likely to play online games and develop online game addiction. But as the data demonstrates, arThe second guideline assumes that a large number of indicators will yield many non-significant weights. Due to the fact that our measurement model consists of formative indicators, this test may be irrelevant. Guideline three assumes the co-occurrence of negative and positive indicator weights, which could lead to a misinterp the results. In our case, only positive weights were observed (Table 3), and the suppressor effect was thus not tested. The fourth guideline discusses absolute versus relative indicator contribution. Indicators with a non-significant or low weight can still have an important absolute contribution. All related indicators must be independently assessed from other indicators to prevent misinterpretation of formative indicator results. As Table 3 shows factor weight of RES1 is not significant. Other formative construct indicators are significant at the 0.01 level. It suggests that they contribute significantly to the form Attention Switching and Parental Monitoring composites. Overall while the statistical evidence regarding the validity of the formative conceptualization is not always conclusive, when synthesized with the theoretical ba and opinions of subject-matter experts, it points to potential plausibility. We hence proceed with treating these concepts as formative composites. Common Method Bias As with all self-reported data, there is a potential for common method biases (Podsakoff et al., 2003). We performed statistical analyses to assess the potential sev problem in our data. First, Harmon’s one-factor test was performed. In a Principal Component Analysis with no rotation 12 components emerged; and the first compo explained only 10.5% of the variance. Second, following Podsakoff et al. (2003), we included in the PLS model a latent common method factor whose indicators incl items. We then calculated the variance explained in our endogenous construct, online game addiction, with- and without- the latent common methods variance factor model. The first was 0.418, and the second was 0.408. The difference is very small. Both tests point to the conclusion that the method is unlikely to be a major source variation. Hypothesis Testing Table 4 presents the estimates obtained from PLS analysis (Bootsrapping with 200 re-samples)[3]. An R2 value of .408 indicates that the model explains a substa amount of variance in addiction. The results provide some support for the hypothesized partial -mediation role of game playing and the direct effects of functional nee motivating factors, and prevention and harm reduction factors on the formation of online game addiction. Among the motivation factors, need to master the mechanic was positively associated with game playing but not with game addiction. As expected, needs for relationship and escapism were positively associated with game play addiction. Need for advancement had no significant influence on addiction. Table 4. Test of Hypotheses Order Number Hypothesis Path coefficients t-value H1a Need for Relationship → Game Playing 0.11 2.76** H1b Need for Escapism → Game Playing 0.09 2.03* H1c Need for Mastering the Mechanics → Game Playing 0.21 3.53*** H1d Need for Advancement → Game Playing 0.01 0.25 H2a Need for Relationship → Addiction 0.09 2.28** H2b Need for Escapism → Addiction 0.14 3.55*** H2c Need for Mastering the Mechanics → Addiction 0.06 0.94 H2d Need for Advancement → Addiction 0.07 1.10 H3 Game Playing → Addiction 0.24 6.13*** H4a Attention Switching → Game Playing -0.11 2.41** H4b Attention Switching → Addiction -0.19 4.80 *** H5a Dissuasion→ Game Playing 0.08 2.31 * H5b Dissuasion → Addiction 0.22 6.34*** H6a Rationalization/ Education → Game Playing -0.17 4.34*** H6b Rationalization/ Education → Addiction -0.02 0.50 H7a Parental Monitoring → Game Playing -0.02 0.67 H7b Parental Monitoring → Addiction -0.10 2.66** H8a Resource Restriction → Game Playing 0.05 0.80 H8b Resource Restriction → Addiction 0.13 3.11** H9a Perceived Cost → Game Playing -0.13 3.53 *** H9b Perceived Cost → Addiction -0.02 0.24 *** p < .001; ** p < .01; * p < .05 The data suggest that attention switching has a significant negative impact on game playing and addiction. It implies that alternative activities could distract adol attention from online games and thus reduce their risk of high levels of addiction. Furthermore, rationalization/ education and cost had significant influences on game no direct effect on online game addiction. Thus, they can alleviate the level of addiction through the reduction of one’s online play time. Dissuasion was expected to reduce game playing and addiction. Nevertheless, contrary to the expectation, it was positively associated with game playing and on addiction. It seems that dissuasion does not serve as a prevention factor but turns out to be more of a remedy which is plausibly exercised only after high levels of gam and online game addiction are observed. We also found that Resource Restrictions was positively associated with addiction and had no significant impact on game pl expected that when fewer resources are available, people will be less likely to play online games and develop online game addiction. But as the data demonstrates, an