Through public health studies, specifically on child obesity modeling, research scholars have been attempting to identify the factors affecting obesity using suitable statistical techniques. In recent years, regression, structural equation modeling (SEM) and partial least squares (PLS) regression have been the most widely employed statistical modeling techniques in public health studies. The main objective of this study to apply the Taguchi method to introduce a new pattern rather than a model for analyzing the body mass index (BMI) of children as a representative of childhood obesity levels mainly related to social media use. The data analysis includes two main parts. The first part entails selecting significant indicators for the proposed framework by applying SEM for primary and high school students separately. The second part introduces the Taguchi method as a realistic and reliable approach to exploring which combination of significant variables leads to high obesity levels in children. AMOS software (IBM, Armonk, NY, USA) was applied in the first part of data analysis and MINITAB software (Minitab Inc., State College, PA, USA) was utilized for the Taguchi experimental analysis (second data analysis part). This study will help research scholars view the data and a pattern rather than a model, as a combination of different factor levels for target factor optimization.
Keywords: childhood obesity modeling; complex analysis; overweight; public health study.