What are you trying to do? Your example is not what is commonly
called ANOVA (some call it ANCOVA) and more often lm() is used.
I suspect that you intended 'population' to be a factor, and it is
not. So population:condition is not an interaction but different
slopes for population by levels of condition.
population*condition expands to
population + condition + population:condition
and this is a larger model with different intercepts by levels of
condition.
I suggest you need study the primary reference (Chambers & Hastie
1992) or at least Bill Venables' exposition in MASS (the book, any
edition). And note that you cannot test interactions in a two-way
layout without replication, so perhaps you also need to talk to a
statistician about ANOVA.
BTW: I think you have messed up your first example: perhaps you meant
studentDataSource="http://files.davidderiso.com/r/studentData.data";
There are no P values in that example because there is no residual
variation: the model fits exactly.
On Sat, 9 Jan 2010, Dave Deriso wrote:
Hello,
I have a simple question about using the aov function syntax (ie. * + or :)
for the interaction of 2 factors. I have read the help files, and researched
other sites, and have included my source files. My goal is to measure the
signifigance of the interaction between population and condition (aka.
population:condition). I can't seem to figure it out.
1. In the first example the significance of population:condition works with
the "allData" but not with the "studentData." Can you please explain why it
fails and how I can fix it?
2. In the second example I can get the measure the significance of
population:condition with 2 different methods, but I get 2 different results
(using the "allData" source). Can you please explain why these Pr(>F) values
are different?
Thank you so much for your help!!
Sincerely,
Dave Deriso
UCSD Psychiatry
#Example 1 ---------------------------COPY & PASTE THE FOLLOWING
#import the data
allDataSource="http://files.davidderiso.com/r/allData.data";
allData.import=read.table(allDataSource,header=T)
studentDataSource="http://files.davidderiso.com/r/allData.data";
studentData.import=read.table(studentDataSource,header=T)
#aov for allData WORKS
allData.integral.aov = aov(integral~population*condition,
data=allData.import)
summary(allData.integral.aov)
#aov for studentData DOES NOT GIVE Pr(>F) of population:condition
studentData.integral.aov = aov(integral~population*condition,
data=studentData.import)
summary(studentData.integral.aov)
#Example 2 ---------------------------COPY & PASTE THE FOLLOWING
#population:condition has a Pr(>F) of 0.96372
allData.integral.aov = aov(integral~population*condition,
data=allData.import)
summary(allData.integral.aov)
#population:condition has a Pr(>F) of 1.070e-06 ***
allData.integral.aov = aov(integral~population:condition,
data=allData.import)
summary(allData.integral.aov)
[[alternative HTML version deleted]]
______________________________________________
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
--
Brian D. Ripley, rip...@stats.ox.ac.uk
Professor of Applied Statistics, http://www.stats.ox.ac.uk/~ripley/
University of Oxford, Tel: +44 1865 272861 (self)
1 South Parks Road, +44 1865 272866 (PA)
Oxford OX1 3TG, UK Fax: +44 1865 272595
______________________________________________
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
