Last Computer Problem

.

4/26/00 cp4-00

Last Computer Problem

The data is below. You need to set up Columns for REV(revenue of the Coca-Cola company 1986 -1996),YEAR(year with 1986 as zero), YEARSQ (YEAR squared), GDP(Real GDP) and RMINW(Real Minimum Wage) as well as resid and pred.

Row REV YEAR GDP RMINW
1 7.0 0 5.5 3.06
2 7.7 1 5.6 2.95
3 8.3 2 5.9 2.83
4 9.0 3 6.1 2.70
5 10.2 4 6.1 2.91
6 11.6 5 6.1 3.12

7 13.0 6 6.2 3.03

8 14.0 7 6.4 2.94

9 16.2 8 6.6 2.87

10 18.0 9 6.7 2.79

11 18.5 10 6.9 2.94

To set up YEARSQ use

LET 'YEARSQ' = 'YEAR'*'YEAR'

To do the problem with all independent variables, you need to do:

Brief K = 3

Regress 'REV' on 4 'YEAR’ ’YEARSQ’ ’GDP’ ’RMINW’ ‘resid’ ’pred’

Plot ‘REV’ * ‘pred’

Explanation:

The 4 in here tells it that there are 4 explanatory (independent) variables. This number must be correct - for example if it were 3, ‘RMINW’ would be zapped.

Brief K=3 needs to be set only once.

The plot shows how well you did - a perfect prediction gives a 45 degree line.

Check the significance tests on the coefficients of your independent variables. Which are not significant? Which have the wrong sign? Can you suggest why?

The rest of the assignment:

First: Try replacing the Regress in the instruction above with Stepwise Regression as is done on page 844 of the text.

Second: Do more regressions using Regress and Plot with appropriate changes to the number in the Regress instruction. First regress ‘REV’ against YEAR alone, then against YEAR and YEARSQ, then YEAR, YEARSQ and GDP. then try REV against GDP alone. Create a new variable equal to GDP squared. Do GDP and GDP squared do as well as YEAR and YEARSQ in predicting REV? What about GDP, GDP squared and YEAR? Try the stepwise regression command with GDP squared added to your independent variables.