R help - Sep 2007 - Cannot get contrasts to work with aov.

I have been trying for hours now to perform an orthogonal contrast through an
ANOVA in R.   

I have done a two-factor factorial experiment, each factor having three
levels.  I converted this dataset to a dataframe with one factor with nine
treatments, as I couldn't work out what else to do.   
I have set up a matrix with the eight orthogonal contrasts that I wish to
perform, but despite having searched this mailing list's archives
thoroughly, I cannot get it to work. 

I am looking to get output along the lines of 
            Df  Sum Sq Mean Sq F value    Pr(>F)     
trt1         1    X         X              X           X 
trt2         1    X         X              X           X 
 ............... 
trt9         1    X         X              X           X 
Residuals   27   495.1   18.3 

However, for everything I've tried, I've simply gotten 

            Df  Sum Sq Mean Sq F value    Pr(>F)     
trt          8 17919.7  2240.0  122.16 < 2.2e-16 *** 
Residuals   27   495.1    18.3         

I have done 
trt <- as.factor(trt) 
contrasts(trt) <- cntrs 
when I then type contrasts(trt), it returns what I expect. 

However, whether I try 
test <- aov(time2 ~ trt) 
test <- aov(time2 ~ trt, contrasts=contrasts(trt)) 
test <- aov(time2 ~ C(trt, cntrs, how.many=8)) 
or anything else I can think of, I only get what I posted above.  I'm
assuming I'm doing something fundamentally wrong, but for the life of me I
can't work out what, and it's killing me at present. 

Any responses would be greatly appreciated 
Marc Burgess
-- 
View this message in context:
http://www.nabble.com/Cannot-get-contrasts-to-work-with-aov.-tf4449485.html#a12695182
Sent from the R help mailing list archive at Nabble.com.

ThatDeadDude wrote:
> 
> I have been trying for hours now to perform an orthogonal contrast through
> an ANOVA in R.   
> 
> I have done a two-factor factorial experiment, each factor having three
> levels.  I converted this dataset to a dataframe with one factor with nine
> treatments, as I couldn't work out what else to do.   
> I have set up a matrix with the eight orthogonal contrasts that I wish to
> perform, but despite having searched this mailing list's archives
> thoroughly, I cannot get it to work. 
> 
> I am looking to get output along the lines of 
>             Df  Sum Sq Mean Sq F value    Pr(>F)     
> trt1         1    X         X              X           X 
> trt2         1    X         X              X           X 
>  ............... 
> trt9         1    X         X              X           X 
> Residuals   27   495.1   18.3 
> 
>   [snippage]
> 
> Any responses would be greatly appreciated 
> Marc Burgess
> 
   I think instead of aov(model) you should try lm(model) and/or
summary(lm(model));
aov() is a special-purpose command that prints only the ANOVA table of a
linear model.
The other commands will print the parameter estimates, which is what you
want.

  I also don't see why you need to convert to a one-way layout: why not

lm(y~treat1*treat2)   ... ?

  good luck
    Ben Bolker
-- 
View this message in context:
http://www.nabble.com/Cannot-get-contrasts-to-work-with-aov.-tf4449485.html#a12745636
Sent from the R help mailing list archive at Nabble.com.