[R] Split plot analysis problems

Jean-Paul Maalouf jean-paul.maalouf at u-bordeaux1.fr
Thu Jul 23 16:50:10 CEST 2009


Thanks Mark and Richard for your propositions on how to extract residuals.

However, I've tried both solutions on my model, and I got different  
residuals :
If we consider the within residuals :

Mark's solution gave me a vector of 24 residuals :

> as.vector(residuals(aov(PH~Community*Mowing*Water + Error(Block))$Within))
  [1] -1.107277057 -0.977277057 -0.007277057 -0.719777057 -1.021627882  
  0.215872118 -1.621627882  0.508372118  1.241311954  1.498811954  
-0.616188046
[12]  0.483811954 -0.477827381 -1.660327381  2.684672619  1.924672619  
-0.465198010 -1.735198010  1.502301990  0.839801990 -0.428515358  
-0.751015358
[23]  0.836484642  1.181484642

and Richard's solution gave me a vector 32 residuals :

do.call(data.frame,proj(aov(PH~Community*Water*Mowing + Error(Block))))->proj
proj$Within.Residuals
  [1] -0.216875  1.745625 -1.174375 -0.354375  0.800625  0.460625  
-0.799375 -0.461875 -0.404375 -0.274375  0.695625 -0.016875 -0.541875   
0.695625 -1.141875
[16]  0.988125  0.589375  0.846875 -1.268125 -0.168125 -1.095625  
-2.278125  2.066875  1.306875 -0.500625 -1.770625  1.466875  0.804375  
-0.638125 -0.960625
[31]  0.626875  0.971875

What could be the difference between the two?

Regards

JPM


Quoting "Richard M. Heiberger" <rmh at temple.edu>:

> Jean-Paul Maalouf wrote:
>> Do you have any idea on how can I verify preliminary assumptions in  
>>  this model (normality of the residuals and variance homogeneity),   
>> since R is not able to extract residuals?
>
> Of course, R extracts residuals.  Use the proj() function.  See ?proj
> for the example
> to get the projection of an aovlist object onto the terms of a linear model
>
> proj(npk.aovE)
>
>
> To get the projections into a simple data.frame, use
> tmpE <- proj(npk.aovE)
> tmpE.df <- do.call("data.frame", tmpE)
> tmpE.df
>
> Mark Difford's solution effectively retrieved columns 3 and 10 from tmpE.df
>
> Rich



-- 
Jean-Paul Maalouf
UMR 1202 BIOGECO
Inra - Université Bordeaux 1
Bâtiment B8, Avenue des Facultés
33405 Talence, France
Tel : 05 40008772




More information about the R-help mailing list