Application of Multinomial Logistic Regression on Physical Violence in the Youth in Makurdi – Nigeria

Abstract

Youth physical violence remains a major public health and social problem, particularly in developing countries such as Nigeria. This study examined the demographic, family, psychological, and economic factors associated with youth physical violence in Makurdi Metropolis using a Multinomial Logistic Regression (MLR) approach. Data were obtained through a structured questionnaire administered to youths aged 15–35 years, with responses collected via Google Forms using a random sampling technique. Youth physical violence was measured as a categorical outcome based on the frequency of physical fights, classified into multiple non-ordinal categories. Descriptive statistics were employed to summarize respondents’ characteristics after which multinomial logistic regression analysis was applied using R statistical software. The findings revealed that age and sex were significant demographic predictors of youth violence, with youths aged 21–30 years and males having a higher likelihood of engaging in violent behavior. Family support was found to be a significant protective factor, as youths with strong family support were less likely to engage in physical violence. Psychological factors showed mixed effects, with levels of stress and anxiety demonstrating significant associations with youth violence. Economic factors, particularly employment status, were also significant, as unemployed youths were more likely to engage in violent behavior compared to their employed counterparts, while tertiary education showed no significant effect. Model diagnostics confirmed good model fit, absence of multicollinearity and satisfaction of key regression assumptions. The study concludes that youth physical violence in Makurdi is influenced by a complex interaction of demographic, family, psychological, and economic factors. It recommends targeted interventions focusing on employment creation, family-based support systems, and youth-centered social programs to reduce violent behavior. The use of multinomial logistic regression provides a robust analytical framework for understanding the multifaceted nature of youth physical violence and offers valuable insights for policymakers and stakeholders.