On Tue, 23 May 2000 13:49:38 -0700, "G. Anthony Reina" <[EMAIL PROTECTED]>
wrote:
> Rich Ulrich wrote:
me >
> > > Plus, I've run the multiple
> > > regression without the transform and seen only about a 5% difference
> > > (not much).
> >
> > - damned if I know what that sentence means. You mean, like,
> > accounting for 99% of the variance, instead of 94%? -- that means "5%"
> > by two different criteria.
anthony >
> Yes, exactly. If I use a square root transform, the variable accounts for 99%
> of the variance and if I don't (i.e. just use the raw spike rates) it accounts
> for 94% of the variance.
Anthony,
I was not being explicit, but one thing that I had in mind was the
notion that 94% to 99% *might be* a ridiculously big difference
despite the fact that it measures only "5%". And here, in the stated
example, is one way that it *is* that big.
- Since you don't see the grievous problem with that, I will try to
point to it, and recommend that you need more advice than you are apt
to receive over the Net.
When you approach a limit, such as 100%, it is wise to consider what
you have as the DIFFERENCE, or 100%-X. "Consider" it. Sometimes it
may not be important; a lot of times, it is. And when it is in terms
of VARIANCE, it is always important.
For 99%, the Error variance amounts to 1%; for 94%, the Error is 6
times as large at 6%. When you have an F-test, the numerator has
R-squared and DF, the denominator has (1-Rsquared) and its DF. The 1%
or 6% is Denominator -- so for this 5 points, the F-test is 6 times as
big.
Now, another question is, Are your two percentages really comparable?
If you want to consider the "variance of Y", then you need to use some
equations with your transformations, to PREDICT some Y scores.
The residuals of the Y-predictions is always fair to look at, as
someone else suggested lately. Or, look at the correlation of those
Predicted, with Y. But you cannot legitimately compare a correlation
"with Y" to another correlation "with square-root(Y)".
--
Rich Ulrich, [EMAIL PROTECTED]
http://www.pitt.edu/~wpilib/index.html
===========================================================================
This list is open to everyone. Occasionally, less thoughtful
people send inappropriate messages. Please DO NOT COMPLAIN TO
THE POSTMASTER about these messages because the postmaster has no
way of controlling them, and excessive complaints will result in
termination of the list.
For information about this list, including information about the
problem of inappropriate messages and information about how to
unsubscribe, please see the web page at
http://jse.stat.ncsu.edu/
===========================================================================
