Multiple Linear Regression
- A statistical technique based on an assumed linear relationship (i.e. of the form y = a + bx + c z +. .. + e) between a dependent variable and a variety of explanatory or independent variables. e (epsilon) is an error term generated by the fact that the independent variables are unlikely to account for all the changes in the dependent variable. The technique involves finding the line that best fits the data to the hypothetical linear structure. The least squares method does this by minimizing the sum of squares of the vertical distances of each actual observation from the fitted line (assuming the dependent variable to be on the vertical axis). The coefficients give a quantitative account of the relationship between y and x. Thus, if b = 7.4, then this means that a one-unit increase in x (any other variables constant) is associated with a 7.4 increase in the predicted value of y. The method is widely used in health economics as in most other branches of economics.