Re: Gnumeric regression is wrong?

["Andreas J. Guelzow" <aguelzow math concordia ab ca>]

For multiple regression there is defintiely much more wrong. Attached is 
a gnumeric file in which the data analysis regression tool was used to 
perform a regression of the dependent systolic blood pressure with 
respect to the independent variables age and weight. Clearly the r 
squared should not be -1.1 times 10 to the 12th... I would have liked to 
see 0.8629.

This is current CVS.

Andreas

Andreas J. Guelzow wrote:

> There is definitely something wrong with the regression calculations 
> in  the data analysis tools. It seems that the standrard error for 
> the  parameter estimates are switched (and then all the calculations 
> based on  them seem to be wrong).
>
> I'll try to locate the exact problem.
>
> Andreas


--
Prof. Dr. Andreas J. Guelzow                    
Assoc. Prof of Mathematics
Concordia University College of Alberta
http://www.math.concordia.ab.ca/aguelzow

Attachment: regression.gnumeric
Description: Binary data