Statistical Modeling of Modes of Waste Disposal Practices by Inhabitants of Bolgatanga Municipality, Ghana: An Application of Polytomous Logistic Regression

摘要

The purpose of this study was to employ Polytomous Logistic Regression to predict inhabitants’ modes of waste disposal practices by examining several indicator variables. Data was collected from 419 successful respondents out of a total sample size of 500 through the administration of questionnaires. A justification towards the application of Polytomous Logistic Regression is made evident in this study. Several tests such as the goodness of fit (Pearson and Deviance), Stepwise Method of Forward Entry based on the Likelihood Ratio and Model Fitting Information were assessed using SPSS version 21. The STATA version 14.2 was utilized to test the assumption of Independence of Irrelevant Alternatives (IIA) before generating Average Marginal Effects via post estimation in STATA based on the parameter estimates (coefficients) of Polytomous Logistic Regression. The feasibility for the application of Stepwise Method of Forward Entry to the data suggested that the inclusion of the independent variables (Average Monthly Income, Family Size, Sex, Type of Residence and Age) are significant contributors to the explanatory power or prediction of the dependent variable (modes of waste disposal by inhabitants). Further analysis of the data revealed that, females are more inclined to practice Burning but less likely to practice Open dumping. The age category (36 and above) years are less likely to opt for Burning. Family size comprising (6-10) members are less likely to resort to using Nearby Container as a waste disposal method. Non-residential dwellers have a high chance of being associated to Open dumping but less probable to practice Burning. Inhabitants with Average Monthly Income (above Gh ￠300) have a high chance of opting for Nearby Container but with a less likelihood to opt for Burning.