In article <[EMAIL PROTECTED]>,
EAKIN MARK E <[EMAIL PROTECTED]> wrote:

>I started a multiple regression using a dependent variable whose mean was
>zero and four independent variables. I created four more dependent variables 
>by adding 10, 100, 1000, and 10000 to the first dependent. I expected the
>r-square of the no-intercept to always increase since the model is
>explaining why y differs from zero but after initially increasing, the
>r-square started to decrease again.

I assume that you meant to say "I created more independent variables
by adding 10, 100, 1000 and 10000 to the first independent [variable]".  
Creating more dependent (ie, response) variables makes no sense.
Creating more independent variables in this way also makes no sense if
the model has an intercept term, as you maybe realize.  If there is no
intercept term, adding one more independent variable equal to another
independent variable plus a constant allows an intercept term to be
simulated by a combination of these two independent variables, which
will likely increase R-squared.  Adding more such independent
variables allows nothing new, so R-squared should stay constant
(though "adjusted R-squared" will decrease).

Your whole procedure makes no sense, really.  Either use a model with
an intercept term or a model without an intercept term.  That's all
you have to decide.  Using a model without an intercept term is
generally a bad idea, unless you have a very good understanding of
what's going on that would indicate that leaving out the intercept is
appropriate.  The fact that you're fiddling with these extra
independent variables is a good indication that you don't have such an
understanding.  So just use a model with an intercept term and get on
with whatever your trying to accomplish by fitting this model.

   Radford Neal

----------------------------------------------------------------------------
Radford M. Neal                                       [EMAIL PROTECTED]
Dept. of Statistics and Dept. of Computer Science [EMAIL PROTECTED]
University of Toronto                     http://www.cs.utoronto.ca/~radford
----------------------------------------------------------------------------
.
.
=================================================================
Instructions for joining and leaving this list, remarks about the
problem of INAPPROPRIATE MESSAGES, and archives are available at:
.                  http://jse.stat.ncsu.edu/                    .
=================================================================

Reply via email to