BOOSTING MODEL PERFORMANCE THROUGH INCLUSION OF RELEVANT PREDICTOR: AN ILLUSTRATION OF FOOD SECURITY STATUS MODEL USING DIMENSION OF FOOD SECURITY

Abstract

The study attempt to illustrate the influence the dimension of food security as additional predictors has on explanation of variation in the outcome variable (Food Security Status) as well as the usefulness of the model for predictability. A total of 120 respondents were selected from Kura local government area using a multistage sampling technique for the illustration. and data  collected using a schedule questionnaire. The Reliability of items used for dimension of food security was  done using Cronach’s Alpha Reliability test. Binary Logistics Regression analysis was carried out using socio-demographic variables as predictors in the first model while the second model has the dimensions of food security added to the predictor. Comparism was done using test statistics such as omnibus test of significance, Cox &Snell R-square,Nagelkerke R-square and Hosmer-Lemeshow goodness of fit test. The compared result showed that each of the dimension of food security satisfied reliability test with an overall Alpha value of 0.74.The Cox &Snell R-square and Nagelkerke R-square  for including dimension of food security ranged from 0.174 to 0.588 compared to that of  0.094-0.319 when the conventional socio-demographic variables were used as predictors. The Omnibus Test was also significant at pvalue of 0.03.It is therefore recommended that model should look beyond conventional predictors and include more variables of relevance in explaining variation and increasing power of predictability of outcome ( Dependent) variables.