As a preliminary step in the selection of variables to use in a statistical-forecasting model, the management accountant has calculated the coefficient of correlation between the firm's sales and three economic indexes. The results were as follows: Which of the following statements indicates the best course of action for the auditor to take in the development of a forecasting model?
A)Drop all three indexes from further consideration because a coefficient of correlation of + 1.0 is necessary for a statistically significant relationship.
B)Include only indexes B and C in the model because they have the only negative coefficients of correlation.
C)Include only index C in the model because its coefficient of correlation is relatively high and therefore probably statistically significant, while the coefficients of indexes A and B are likely to be insignificant.
D)Include only index A in the model because it has the only positive coefficient of correlation.

Question

As a preliminary step in the selection of variables to use in a statistical-forecasting model, the management accountant has calculated the coefficient of correlation between the firm's sales and three economic indexes. The results were as follows:  Which of the following statements indicates the best course of action for the auditor to take in the development of a forecasting model?
A)Drop all three indexes from further consideration because a coefficient of correlation of + 1.0 is necessary for a statistically significant relationship. 
B)Include only indexes B and C in the model because they have the only negative coefficients of correlation. 
C)Include only index C in the model because its coefficient of correlation is relatively high and therefore probably statistically significant, while the coefficients of indexes A and B are likely to be insignificant. 
D)Include only index A in the model because it has the only positive coefficient of correlation.

Quizplus · Accepted Answer

A coefficient of correlation of +1.0 is not necessary for a statistically significant relationship. Rather, a coefficient of correlation that is significantly different from zero is required for a statistically significant relationship. The coefficient of correlation for index C is relatively high and thus may be statistically significant. Furthermore, the coefficients of indexes A and B are likely to be insignificant, so it is best to exclude them from the model. Therefore, the best choice is to include only index C in the model.

As a Preliminary Step in the Selection of Variables to Use