An Appropriate Non Linear Regression Model for Assessing Community Policing and Violent Crimes Reduction

An appropriate non-linear regression of the relationship between a response variable and the predictor variables was considered using Probit, Logit and Poisson Log-linear regression. The study focused on analysis of crime rate before and after engagement of Onelga Security Peace Advisory Committee (OSPAC) in Ogba, Ndoni, Egbema Local Government Area (ONELGA) of Rivers State using 220 questionnaires administered to households in twenty-five communities in the LGA by convenience sampling method. Descriptive statistics, ranks, percentage analysis and non-linear regression techniques were methods of analysis used. Microsoft Excel, SPSS 23 and Minitab 18 statistical Software were used. The Akaike Information Criterion (AIC) was used to compare the models. The results showed that Probit and Logit regression models identified the covariates of killing and rape cases as the major crime before engagement of OSPAC since both coefficients have significant effect at 5%. No independent variables have significant effects on response variables after the engagement of OSPAC, except the constant coefficient [ Bo] which implies violent crime reduction in the community. It was concluded that the Logit regression model is more suitable for modelling response variable on the covariates and community policing intervention has an impact on violent crime reduction.