Re: interpretting F & t-vals from regression
- From: Brett Magill <magillb@*nomail*.sbcglobal.net>
- Date: Thu, 23 Mar 2006 19:25:26 GMT
Rob Campbell wrote:
I've skimmed the posts and I know a small amount about these issues already
but I'm sure I should learn more: I'm rather new to this game. Your information made me realise that I was
approaching the model building incorrectly rather than misinterpreting
coefficients. For anyone who's interested, here's my take on the situation
now.
You're right that the model was over-specified but I don't think removing
intensity1000 is possible. This is a designed experiment and in the context
of this analysis, that variable is needed to explain what subjects did at
sound durations of 1000 ms.
That is probably a better approach in any case. Though my initial
comments were off target as to exactly what your question was, I think that the over-specification of the model is still the result of improper coding of the factors. You have five levels (k) of duration as an ordered factor. It takes k-1 levels to fully represent that factor in the model. All five levels appeared in your model. As I understand it,
this is the original model you wanted:
lme(outcome ~ intensity + duration + intensity *
duration,data=X, random=~duration|subject/duration )
Your ANOVA table would contain entries for:
(Intercept)
intensity
duration
intensity*duration
and your vector of coefficients would contain:
(Intercept)
intensity
duration100
duration200
duration500
duration1000
intensity:duration100
intensity:duration200
intensity:duration500
intensity:duration1000
The reference category for the ordered factor, duration, is 40. All of
the other coefficients for duration would be interpreted relative to that.
.
- 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
- Re: interpretting F & t-vals from regression
- From: Reef Fish
- Re: interpretting F & t-vals from regression
- From: Rob Campbell
- interpretting F & t-vals from regression
- Prev by Date: Re: nonparametric partial correlation test
- 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
|