Maartje Boer

THE COURSE OF PROBLEMATIC SMU 209 7 Figure 7.2 Average Co-Trajectory of Problematic SMU and SMU Frequency, n = 1,419 Notes. Data labels show the model estimated means based on the one-class solution. Vertical bars denote the 95% confidence intervals. was not significant ( p = 0.172). Furthermore, the Entropy of the 5- and 6-class models was below 0.700, which suggests inaccurate classification accuracy. Hence, the 1- to 4-class solutions seemed more eligible. From these models, we selected and further interpreted the 4-class solution, because this model showed the best model fit according to all fit indices, and the Entropy was appropriate (0.719). Table 7.2 Model Fit Indices and Classification Accuracy LCGA models (n = 1,419) Par. C BIC Entropy LMR-LRT value LMR-LRT p -value BLRT p -value Min. class size Max. class size Min. probability Max. probability 10 1 25439.225 1,419 1,419 1 1 17 2 23588.660 0.723 1864.667 <0.001 <0.001 691 728 0.916 0.918 24 3 23081.855 0.708 546.846 <0.001 <0.001 387 578 0.847 0.891 31 4 22868.471 0.719 259.088 <0.001 <0.001 224 527 0.728 0.896 38 5 22806.901 0.673 110.204 0.035 <0.001 238 350 0.748 0.874 45 6 22780.090 0.665 76.115 0.172 <0.001 139 348 0.650 0.802 Notes. LCGA = Latent Class Growth Analysis; Par. = number of free parameters; C = number of classes; BIC = Bayesian Information Criterion, LMR-LRT = Lo-Mendell-Rubin Likelihood Ratio Test; BLRT = Bootstrap Likelihood Ratio Test; Min. = minimum; Max. = maximum.