How Are BMI, Nutrition, and Physical Exercise Related? An Application of Ordinal Logistic Regression

• Published in: Life (Basel, Switzerland) Vol. 12; no. 12; p. 2098
• Main Authors: Wang, Hongwei, Quintana, Fernando G, Lu, Yunlong, Mohebujjaman, Muhammad, Kamronnaher, Kanon
• Format: Journal Article
• Language: English
• Published: Switzerland MDPI AG 01.12.2022; MDPI
• Subjects: Age; BMI; Body mass index; College students; Comparative analysis; Data analysis; Diet; Dietary fiber; Evaluation; Exercise; Food; Grain; Health aspects; Health care; health survey; Health surveys; Hispanic Americans; Medical research; Medicine, Experimental; Methods; Nutrition; Nutrition research; Obesity; ordinal logistic regression; Overweight; Physical exercise; Physical fitness; Proteins; Regression analysis; Regression models; South Texas; Sports training; Statistical models; Strength training; Students; Surveys; Type 2 diabetes; Variables; Texas; United States; United States > US; ordinal logistic regression; South Texas; health survey; BMI
• ISSN: 2075-1729; 2075-1729
• DOI: 10.3390/life12122098

This paper performs a detailed ordinal logistic regression study in an evaluation of a survey at a university in South Texas, USA. We show that, for categorical data in our case, ordinal logistic regression works well. The survey was designed according to the guidelines in diet and lifestyle from the American Heart Association and the United States Department of Agriculture and was sent out to all registered students at Texas A&M International University in Laredo, Texas. Data analysis included 601 students' results from the survey. Data analysis was conducted in Rstudio. The results showed that, compared with students who do not have enough whole grain food and exercise, those who have enough in both tend to have normal BMIs. As age increases, BMI tends to be out of the normal range. Because BMI in this research has three categories, applying an ordinal logistic regression model to describe the relationship between an ordered categorical response variable and more explanatory variables has several advantages compared with other models, such as the linear regression model.