naive MANOVA question

[Russell Wyeth-2]

Folks,

I am just starting to use General Linear Model:Repeated Measures...
Why does the output not include an F-statistic for a Corrrected Model?
In ANOVAs, I know one shouldn't consider any of the individual treatment
or interaction effects without a significant p-value for the full
model.  Is such a step is not necessary here? Or is it just in my
specific analysis that no statistic is calculated?  I am interested in a
specific interaction effect (which is significant) and I wanted to be
sure I was not making a mistake by considering it on its own.

Thanks in advance,
Russell

The details:
I'm using this to do a similar test as the ANCOVA method to test for
differences in regression slopes.
Anaesthetic is applied to a neuron, and response height is measured,
then anaesthetic is washed off, and response height measured again.
This method is required given that I am trying to show that the
anaesthetic is reversible in its effects.
The anaesthetic is applied at 7 different concentrations, with 3 or more
replicates at each concentration.

The anaesthetic's effect increases with increasing concentration.  The
response is uniform after wash - there is no effect of concentration.

So, in my model, I have one repeated measure effect: testvswash with 2
levels, and one covariate: concentration.
The test I am interested in is the interaction testvswash*concentration,
to examine whether the slopes of the relationships between concentration
and response are different, depending on whether anaesthetic is present
or has been washed away.  The testvswash p-value is not significant, but
the interaction testvswash*concentration is highly significant.

Am I safe to conclude the anaesthetic had a strong effect even without a
Corrected Model statistic and p-value, and thus can continue on with
individual regressions to determine intercept estimates?




--
Russell Wyeth
Dept of Physiology and Biophysics
Dalhousie University
Halifax, NS B3H 1X5 Canada
Ph: 9024946417 Fx: 9024941685
[hidden email]

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

[Swank, Paul R]

Because there is no statistically defensible way to combine the within
subject and between subjects variability into a single F ratio.


Paul R. Swank, Ph.D.
Professor, Developmental Pediatrics
Director of Research,


University of Texas Health Science Center at Houston

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
Russell Wyeth
Sent: Monday, November 12, 2007 2:57 PM
To: [hidden email]
Subject: naive MANOVA question

Folks,

I am just starting to use General Linear Model:Repeated Measures...
Why does the output not include an F-statistic for a Corrrected Model?
In ANOVAs, I know one shouldn't consider any of the individual treatment
or interaction effects without a significant p-value for the full model.
Is such a step is not necessary here? Or is it just in my specific
analysis that no statistic is calculated?  I am interested in a specific
interaction effect (which is significant) and I wanted to be sure I was
not making a mistake by considering it on its own.

Thanks in advance,
Russell

The details:
I'm using this to do a similar test as the ANCOVA method to test for
differences in regression slopes.
Anaesthetic is applied to a neuron, and response height is measured,
then anaesthetic is washed off, and response height measured again.
This method is required given that I am trying to show that the
anaesthetic is reversible in its effects.
The anaesthetic is applied at 7 different concentrations, with 3 or more
replicates at each concentration.

The anaesthetic's effect increases with increasing concentration.  The
response is uniform after wash - there is no effect of concentration.

So, in my model, I have one repeated measure effect: testvswash with 2
levels, and one covariate: concentration.
The test I am interested in is the interaction testvswash*concentration,
to examine whether the slopes of the relationships between concentration
and response are different, depending on whether anaesthetic is present
or has been washed away.  The testvswash p-value is not significant, but
the interaction testvswash*concentration is highly significant.

Am I safe to conclude the anaesthetic had a strong effect even without a
Corrected Model statistic and p-value, and thus can continue on with
individual regressions to determine intercept estimates?




--
Russell Wyeth
Dept of Physiology and Biophysics
Dalhousie University
Halifax, NS B3H 1X5 Canada
Ph: 9024946417 Fx: 9024941685
[hidden email]

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command SIGNOFF SPSSX-L For a list
of commands to manage subscriptions, send the command INFO REFCARD

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD