In Search of Appropriate Models of Communication- Behavior Using Multiple Input Variables for Micro-level Study

Studies on Communication Behavior indicate that modeling communication behavior of farmers will require a microlevel study after stratifying them into homogeneous categories in terms of a suitable criterion. The principal objective of social science research is to understand human behavior based on empirical studies. However, empirical prediction of behavior is not meaningful unless it has a strong theoretical basis and it is logically consistent. The efficient combination of inputs and output in this case of modeling communication behavior, it is tempting to use the functional form to be of Cobb-Douglas type. This is because after taking logarithms the right hand size becomes a linear function and linear functions are easily dealt with by multiple regression techniques. In theory, the output function is supposed to involve variables representing individual input variables measurable in physical terms. But input variables involved in communication behavior are mostly aggregation of several factors. For an output function fit to be considerable “good” in the meaningful way it is necessary that R² should not merely be significant but also should take a high value. But even a large R² is not sufficient for considering an output function to be a “good” fit. It is quite possible for two different functional forms involving different sets of input variables each give fits with high R², but they may carry totally different interpretations.