Maartje Boer

THE COMPLEX ASSOCIATION BETWEEN SMU AND WELLBEING 273 8 Table A8.3 (continued) Random Effects Models, Life Satisfaction m5b m6b Level 1 ( n = 5,676) B SE p B SE p Wave -0.152 0.014 <0.001 -0.149 0.015 <0.001 SMU problems -0.104 0.010 <0.001 SNS viewing SNS posting -0.012 0.008 0.164 -0.008 0.009 0.396 SNS liking SNS responding IM viewing IM sending Level 2 ( n = 1,419) B SE p B SE p Female -0.106 0.035 0.002 -0.077 0.032 0.017 Pre-vocational education -0.001 0.038 0.978 0.049 0.040 0.223 Immigrant background 0.021 0.058 0.717 0.019 0.057 0.738 SMU problems -0.193 0.018 <0.001 SNS viewing SNS posting -0.040 0.026 0.124 0.017 0.029 0.558 SNS liking SNS responding IM viewing IM sending Random parameters B SE p B SE p Residual variance within 0.394 0.012 <0.001 0.379 0.011 <0.001 Residual variance between 0.275 0.016 <0.001 0.237 0.015 <0.001 Variance slope 0.006 0.003 0.102 0.005 0.004 0.175 Covariance slope- intercept <0.001 0.007 0.955 <0.001 0.008 0.979 95% prediction intervals B SE p B SE p LL- B -0.157 0.047 0.001 -0.145 0.054 0.007 UL- B 0.133 0.047 0.005 0.130 0.058 0.025 LL- β -0.271 0.081 0.001 -0.253 0.094 0.007 UL- β 0.231 0.081 0.005 0.226 0.101 0.025 Fit statistics Est. Est. Est. Est. Est. Est. Free parameters 11 13 Deviance/AIC/BIC 12806.3 12828.3 12901.4 12472.8 12498.8 12585.1 Model comparison 1 Est. Est. Est. Est. Est. Est. Δ Free parameters 2 2 Δ Deviance/AIC/BIC -8.6 -4.6 8.7 -8.9 -4.9 8.4 p -value deviance 0.014 0.012 Corrected p -value 2 0.009 0.007 Notes. SNS = social network sites; IM = instant messengers; SMU = social media use; Level 1 = yearly measurements; Level 2 = adolescents; B = unstandardized coefficient; SE = standard error, p = p- value; LL = 95% prediction interval lower limit; UL = 95% prediction interval upper limit; β = STDYX-standardized; Est. = estimate; Deviance = -2*loglikelihood; AIC = Akaike Information Criterion; BIC = Bayesian Information Criterion. 1 Model 5b was compared to Model 1b; Model 6b was compared to Model 2b. 2 The p- value for the deviance was corrected to take into account the boundary of the slope variance parameter (Hox, 2010b; Stoel et al., 2006).