ABSTRACT
At times, the constancy of β0 and β1 at successive data points can legitimately be questioned. For example, if β1 depicts the response in regional sales of a certain product to regional advertising expenditure, it is evident that this sales response is infl uenced by, say, other factors such as regional economic conditions and the activity of competitors (if any). If these other factors can be held constant, then β1 might be expected to be (approximately) constant as well. Furthermore, if these supplemental factors are observable, then they can be explicitly treated as explanatory variables in the model (we obviously have the case of multiple regression) so that β1 can again be viewed as roughly constant because these other factors are being controlled for. However, if these additional factors are unobservable, then one can essentially view β1 as the average of a random response rate in sales per unit change in advertising expenditure.
