Professor Seldon Wright has decided to forecast daily sales at the Carrollton Wal-Mart. As part of his data collection process, he gets the bright idea of putting a student in the parking lot to count the number of cars in the parking lot for each day. He also uses another student to count the number of people entering the store for each day. Dr. Wright calculated the correlation coefficient between cars and sales (.81) and between people and sales (.75). Because both correlation calculations were large and positive, Dr. Wright expected both the number of cars in the parking lot and the number of people in the store to have a big effect on sales. However, his regression results for those two variables were not significant, even though the R-Squared was 78%. What might have happened?
HINT: THE ANSWER ISN'T THAT PEOPLE PARKING IN THE PARKING LOT AREN'T BUYING STUFF!!! THE CORRELATION COEFFICIENTS ARE BIG!!!
a) The model likely has a multicollinearity problem, and Dr. Wright should consider dropping either the number of cars OR the people entering from the model.
b) There is no solution to Dr. Wright's problem.
c) Dr. Wright needs to add a third variable that counts the number of people who get in line at the cash register.
d) The two variables are obviously not very good predictors. Dr. Wright needs to drop both and look for other determinants of sales.