Re: interpretting F & t-vals from regression
- From: "Reef Fish" <Large_Nassau_Grouper@xxxxxxxxx>
- Date: 23 Mar 2006 09:21:22 -0800
Rob Campbell wrote:
<snip>
I'm not sure I entirely follow you. I think it's because I should have sent
more information. Here it is.
Brett Magill gave you completely inappropriate explanations given your
unmistable question (which was very clear, in the light of your lenghy
verbal statements).
The answer is very simple.
You didn't know HOW to interpret ANY of your multiple regression
coefficients!
Go to sci.stat.math and use google to find any number of threads
that dealt with the MEANING of a multiple regression coefficient, and
your mystery would be solved immediately.
What you experienced is a case involving "multicollinearity condition"
among your independent variables X's.
See e.g. http://tinyurl.com/nt8go
and other posts within that thread. The coefficient in any multiple
regression is the effect of THAT X, IN THE PRESENCE of all other Xs
in the model!
A significant F, together with insignificant Ts for ALL regression
coefficients simply mean that each one is "not useful" in the
presence of the others -- or, you have too many REDUNDANT
variables!
Read other threads about Multicollinearities and the MEANING of
regression coefficients.
and summary shows:
Value Std.Error DF t-value p-value
duration40:intensity 0.078111 0.0533651 165 1.463718 0.1452
duration100:intensity 0.086904 0.0533651 165 1.628480 0.1053
duration200:intensity 0.029069 0.0533651 165 0.544725 0.5867
duration500:intensity 0.008720 0.0533651 165 0.163403 0.8704
duration1000:intensity 0.006706 0.0533651 165 0.125671 0.9001
That indicates that "duration 1000" is the "most useless" in the
presence of other variables in the model.
REMOVE IT from your model, and see how the rest of them change.
The coefficients from summary.lme are the slopes and they are not with
reference to anything so I guessed that the t-tests must be for a
significant difference from zero slope. As you can see, none are
significantly different from zero.
That's as sure a sign of a "multicollinearity condition" as
"bears do it <sic> in the woods".
So this high F-value is confusing me.
It says you have a good fitting model, but you used TOO MANY
variables in your model.
So I'm safe in reporting the result as "none of the 5 durations tested had
intensity slopes that were significantly different from zero" ?
IN THE PRESENCE of the others, yes!
I ignore the F-value because it's not the hypothesis of interest?
That is the ONLY piece of useful information you have, about the
OVERALL effect of your regression model.
Just GET RID of one (any one) of the existing variables, and ALL
the Fs and Ts will change, possibly quite drastically.
If you had been reading the threads I cited, and understood them,
you should have no problem dealing with whatever new results
you get in the reduced multiple regression model(s).
-- Bob.
.
- Follow-Ups:
- Re: interpretting F & t-vals from regression
- From: Rob Campbell
- Re: interpretting F & t-vals from regression
- References:
- interpretting F & t-vals from regression
- From: Rob Campbell
- Re: interpretting F & t-vals from regression
- From: Brett Magill
- Re: interpretting F & t-vals from regression
- From: Rob Campbell
- interpretting F & t-vals from regression
- Prev by Date: Re: interpretting F & t-vals from regression
- Next by Date: Re: interpretting F & t-vals from regression
- Previous by thread: Re: interpretting F & t-vals from regression
- Next by thread: Re: interpretting F & t-vals from regression
- Index(es):
Relevant Pages
|
