Regression can help the auditor develop fairly precise expectations for financial reporting. However, auditors should use regression carefully. For example, consider the following scenario:

The relationship depicted in the model is not valid; rather, the relationship in the model exists because a third variable not depicted in the model is driving the regression coefficient. What impact would this result have on the auditor's use of regression in developing financial statement expectations? How can auditors reduce the likelihood that this scenario is present?