Modeling multinomial logistic regression on characteristics of smokers after the smoke-free campaign in the area of Melaka

Abstract

Multinomial logistic regression is used to model the outcome of a polytomous variable with categorical more than two categories and the predictors are nominal, ordinal, interval and ratio. This model is used to predict the probabilities of the different possible outcomes of a categorically distributed dependent variable, given a set of independent variables. Multinomial logistic regression model also estimates a separate binary logistic regression model for each indicator variable. The result is j-1 binary logistic regression models. Each one tells the effect of the predictors on the probability of success in that category, in comparison to the reference category. The model is validated by selection of predictor variables, test of regression coefficients, a significance test of the overall model, goodness-of-fit measures and validation of predicted probabilities using odds ratio. This study used the multinomial logistic regression techniques to examine smokers’ status that were affected by the Smoke-Free Melaka Campaign (SFMC) in the area of Melaka based on the demographic profile, awareness of the campaign, smoking behavior and discussion, perception and opinion of the campaign. The results indicated that the characteristics of smokers affected by the campaign were related to race, marital status, occupation, awareness of advertisements of SFMC, opinion about smoking ban in all public places in Melaka, smoking behavior in the presence of non-smoking family, smoking behavior in the presence of other smokers and also interaction between Malay and smoking behavior in the presence of non-smoking family