Nonlinear structural equation model approaches and an application in R programme

Structural Equation Modeling (SEM) is a statistical approach used to examine causal and mutual relationships between observed and latent variables. In cases where the data set is not linear or is not suitable for normal distribution, classical methods do not always give good results. In application of this thesis, three approaches were considered within the scope of nonlinear structural equation modeling used in cases where SEM assumptions were not met. The first is the latent moderated structural equation modeling approach used to analyze nonlinear relationship types using latent variables, such as interaction effect or quadratic effects, and the second is the mixture structural equation modeling approach used to approximate nonlinear latent variables with mixtures of linear relationships. Nonlinear was developed by Kelava, Nagengast and Brandt (2014) as an additional semi-parametric method. In application of the thesis, the effects of loneliness and life satisfaction on internet addiction were investigated among students of Düzce University Medical School UCLA loneliness scale, life satisfaction scale and internet addiction scale were used in designing the research model. When the findings of the study were examined, it was seen that nonlinear structural equation mixture modeling approach is more useful than the other two approaches. Loneliness latent variable had a significant positive effect on internet addiction, while life satisfaction latent variable had a significant negative effect. In addition, it was found that both the interaction effect of loneliness and life satisfaction and the quadratic effect of life satisfaction have negatively effects on internet addiction.