I've found that the variance estimates for parameters from MS Linear Regression don't match the standard, textbook variance estimates for linear regression.

A good example of this is the Fuel Consumption data from Weisberg's "Applied Linear Regression" text. MS Linear regression will produce regression parameter estimates that match exactly with the text (or with what you'd get from a standard stats program). But the variances for the individual parameter estimates are just wrong, with no clear pattern that I could find.

I assume you used Fuel, right?
What variance did you get using Analysis Services? you extracted the variance from the model content , right?

thanks
bogdan
[quoted text, click to view]