Notes and additional parts on #33:

There are four departments. Create dummy variables, only three of which are needed. I chose departments 1, 2, 3.

Include all variables in a multiple regression and state which ones are significant. (Normally, in model building we would delete the variables that were not significant (one at a time).)

(f) In answering this part use StatTools to compute the exact prediction interval for the person with the given profile.

Additional parts:

• The confidence intervals for all coefficients are computed by StatTools. Using the value of the coefficient and standard error for the Years Employed variable, demonstrate how to compute the upper and lower confidence interval for that coefficient in the population relationship. Show any Excel formula used, with numerical values as arguments.
• Add an interaction variable between gender and years of education and re-run the model. Write down the full regression equation including all variables. Is there a significant interaction effect? Give your reason.
• Leaving all variables in the model, write down the two prediction equations for men and for women. Interpret the two intercept and slope coefficients with respect to years of education in the two models (not in general; referring to the estimated values specifically).