An alternate approach to calculating standard errors and confidence intervals for regression is to apply them to the calculated bQ and x coefficient(s), instead of to the y estimate. For example, this is part of Excel's regression output for comparing the weights and lengths of bears.
This shows that the bQ term, which the regression formula would estimate as -352, could, in fact, range from -663 to -40 in the 95% confidence interval. Similarly, the x coefficient, which is estimated as 9.66, could range from 4.9 to 14.4.
Refer to the data in Exercise 11 in Section 11-3. Let x represent the number of years of a person's education, and let y represent his or her salary (in thousands of dollars). Use Excel to find the confidence limits for the intercept and x coefficient for the regression between education and salary.