R help - Feb 2010 - factorial block design with missing data

Hello!

I have read somewhere (somehow, I can't seem to find it again, it's been
a couple of months) that when analyzing factorial block design, the position
where you put the block factor is important, even more when there are missing
values.

I understand that when using anova.lm, the order is sequential, so that if I
want to check for a treatment effect, I should put my blocking factor before in
order to . It's just that I got confused with all the answers from previous
posts and books, and I don't know if the missing values are being handled
properly.

My code is:
P.biom = lm(biomass ~ Bloc + Trt*Clone, data=mydata)
P.aov = anova.lm(P.biom, test="F")
> anova.lm(P.ar.2, test="F")
Analysis of Variance Table

Response: M_aerien
          Df  Sum Sq Mean Sq  F value    Pr(>F)    
Bloc       2   139.7    69.9   0.4054    0.6710    
Trt        1 31069.5 31069.5 180.2905 6.227e-13 ***
Clone      7  1206.2   172.3   0.9999    0.4544    
Trt:Clone  7   570.3    81.5   0.4728    0.8450    
Residuals 25  4308.2   172.3                       
---
Signif. codes:  0 ?***? 0.001 ?**? 0.01 ?*? 0.05 ?.? 0.1 ? ? 1 



Thank you very much. 

Guillaume Th?roux Rancourt
Ph.D. candidate --- Plant Biology
Universit? Laval, Qu?bec, QC, Canada
guillaume.theroux-rancourt.1 at ulaval.ca

Guillaume Th?roux Rancourt writes:
> 
> Hello!
> 
> I have read somewhere (somehow, I can't seem to find it again,
> it's been a couple of months) that when
> analyzing factorial block design, the position where you put 
> the block factor is important, even more
> when there are missing values.
> 
> I understand that when using anova.lm, the order is sequential, 
> so that if I want to check for a treatment
> effect, I should put my blocking factor before in order to . 
> It's just that I got confused with all the
> answers from previous posts and books, and I don't know if 
> the missing values are being handled properly. 
> 
> My code is:
> P.biom = lm(biomass ~ Bloc + Trt*Clone, data=mydata)
> P.aov = anova.lm(P.biom, test="F")
> 
  [snip]

  Sorry to make things even more complicated, but *if* you
are dealing with a large number of missing values you might
want to consider using the lme() function (in the nlme package),
along with its companion book by Pinheiro and Bates (2000);
the methods used in that package should be more robust to
lack of balance than lm/aov ...


  Ben Bolker