Derivation and Discussion of a Regression Model for Estimating the Fuel Economy of Automobiles

This paper presents a discussion on the functional form (based on a theoretical analysis) of a regression model for estimating the fuel economy of motor vehicles. The statistical analysis tests the data for heteroskedasticity, investigates the use of different combinations of explanatory variables, discusses multi-collinearity, and determines the elasticities of fuel economy with respect to the various explanatory variables. Also, the analysis shows that the regression model can be treated as a case of restricted coefficients.

The results of the statistical analysis--mainly the high value of the corrected coefficient of determination, and the very high values of the test statistics of all the coefficients of the regression--indicate that the selected model is satisfactory for the population represented by the given data.