Application of Logistic Regression in Modelling Depression among Students at Murang’a University of Technology
Loise Muthoni Wahome *
Murang’a University of Technology, Murang'a, Kenya.
Jane Wangui Runo
Mathematics and Actuarial Science Department, Murang’a University of Technology, Murang'a, Kenya.
*Author to whom correspondence should be addressed.
Abstract
Depression is a common mental health disorder worldwide that can adversely affect an individual's thoughts, emotions, communication, and ability to perform everyday activities. Despite its significant impact, stigma surrounding mental illness may discourage early recognition and treatment, resulting in some cases remaining undiagnosed during the mild or moderate stages. When left unaddressed, severe depression may result in substantial functional impairment, disability or suicidal behaviour. Previous studies have examined the prevalence of depression across different populations, with findings varying considerably depending on the characteristics of the study population and the factors considered. This study applied the logistic regression to model the prevalence of depression among students at Murang'a University of Technology (MUT). The specific objectives were to determine the prevalence of depression among MUT students; Identify the factors associated with depression among MUT students and evaluate the predictive performance of logistic regression in modelling depression among MUT students. A sample of 1448 students drawn from the different schools within the university participated in the study. Data were collected using questionnaires that captured respondents' sociodemographic characteristics and other factors associated with depression. The questionnaires were distributed to participants through social media platforms. Proportionate stratified random sampling was first used to ensure representation of students from all schools in accordance with their population sizes, after which simple random sampling was applied to select individual respondents within each stratum. The collected data was analysed using both descriptive and inferential statistical techniques. Depressive symptoms were assessed using the Patient Health Questionnaire-9 (PHQ-9), with a score of 10 or above used as the threshold for identifying depression. The findings indicated that 25.97% of the students exhibited depressive symptoms, comprising 19.61% with moderate symptoms and 6.35% with severe symptoms. Depressive symptoms were significantly more common among male students , students in their third and fourth years of study. The variables that were significantly related to depression were, gender , social support network , financial situation and past abuse, trauma and neglect . The predictive performance of the logistic regression model was evaluated using confusion matrix-based measures. The model demonstrated strong performance, achieving an accuracy of 0.7624, sensitivity of 0.5625, specificity of 0.8341, positive predictive value of 0.5488 and negative predictive value of 0.8416. The findings highlight the importance of developing interventions that address the specific risk and protective factors associated with depression among university students. Further research should also examine the long-term effectiveness and outcomes of such interventions, thereby contributing to the growing body of knowledge on mental health within higher education settings.
Keywords: Depression, university students, logistic regression, patient health questionnaire-9 (PHQ-9), odds ratio, confusion matrix, social support, financial situation, predictive performance, Murang’a University of Technology