A company has recorded its overhead costs, machine hours, and labor hours for the past 60 months. The data are in the file P11_65.xlsx. The company decides to use regression to explain its overhead hours linearly as a function of machine hours and labor hours.
However, recognizing good statistical practice, it decides to estimate a regression equation for the first 36 months and then validate this regression with the data from the last 24 months. That is, it will substitute the values of machine and labor hours from the last 24 months into the regression equation that is based on the first 36 months and see how well it does.
a. Run the regression for the first 36 months. Explain briefly why the coefficient of labor hours is not significant.
b. For this part, use the regression equation from part a with both variables still in the equation. Fill in the fitted and residual columns for months 37 through 60. Then do relevant calculations to see whether the R2 (or multiple R) and the standard error of estimate se are as good for these 24 months as they are for the first 36 months. Explain your results briefly.