LOGISTIC REGRESSION MODELING OF REDUCTANT HERBICIDE IN PAGARALAM COFFEE FARMING

Abstract

The presence of weeds can affect the productivity of coffee plants. The use of herbicides that are not wise in controlling weeds can have a negative impact on the quality of coffee production and land. This study aims to obtain a binary logistic regression model of the use of reductant herbicides by coffee farmers in Pagaralam South Sumatera. This research involved 165 coffee farmers, consisting of 81 farmers who used reductants and 85 farmers who did not use reductants. In the results of bivariate analysis, variables that have a significant effect on the status of using reductant herbicides, do not necessarily have a significant effect on the logistic regression model. Overall prediction accuracy of the model results of the enter method and backward method are respectively 78.2% and 76.4%. The two best models obtained show that farmer age, number of trees, number of family workers, and land productivity can reduce the probability value of farmers using reductant herbicide. On the other hand, variables that can increase the opportunity value of using reductants, starting with the greatest effect, are net income, length of harvest, frequency of herbicide use, frequency of use of organic fertilizers, and age of trees. Based on the factors that affect the use of reductants, coffee farmers should set aside costs for land maintenance, including costs for environmentally friendly weed control, so that they can support the coffee plants to continue producing optimally.