[S] residuals from aov on split-plot, and lme for repeated measures

Hobie Perry (hperry@forestry.umn.edu)
Tue, 30 Jun 1998 09:58:47 -0500


Hello,

I have been using the split-plot analysis of variance to test a repeated
measures design (following the suggestions of the S-Plus Guide, sec. 14.5).

Now I have a problem. I want to examine the distribution of the residuals
so I can assess the ANOVA assumptions, but I cannot get any residuals from
the model. I've tried the things that work with normal ANOVA objects, such
as resid, but it doesn't seem to work. The model and output are provided
below. Any suggestions?

Also, some have suggested the use of lme to perform analyses of repeated
measures designs (e.g., V & R, 1997, sec. 10.3). Could folks please
elaborate? I would like to use lme but my data oscillate in time, and I
don't think that fitting a linear model to these data would be very
straightforward. I have experimented with it a little bit, and adding fixed
effect terms to the model actually caused my Loglik get more negative (move
away from zero) when I added went from fit1 to fit2 (following the example
of Pinheiro and Bates (1995)). This appears to be a bad thing. Any
suggestions?

Thank you very much for your time and assistance.
Hobie

---
> anova.01
function()
{
# This function prepares the aov for repeated measures designs
# using conc.01
	temp <- aov(conc ~ trt * date.factor + Error(plot.no), data = conc.01, subset
		 = (lysimeter == "T" & depth == "depth1" & chem.species == "NH3-N"))
	print("Here's the summary")
	print(summary(temp))
	print("And here are the residuals")
	print(resid(temp))
}
> anova.01()
[1] "Here's the summary"
Error: plot.no 
          Df Sum of Sq   Mean Sq  F Value     Pr(F) 
      trt  1  0.352776 0.3527755 1.273139 0.3222712
Residuals  4  1.108365 0.2770911                   

Error: Within Df Sum of Sq Mean Sq F Value Pr(F) date.factor 7 0.1548245 0.02211778 2.927402 0.0197695 trt:date.factor 7 0.0073453 0.00104933 0.138884 0.9941803 Residuals 28 0.2115521 0.00755543

[1] "And here are the residuals" NULL NULL

--
C. Hobart Perry				hperry@forestry.umn.edu
Department of Forest Resources		+1 612 625 5765 (voice)
University of Minnesota, St. Paul	+1 612 625 5212 (fax)
-----------------------------------------------------------------------
This message was distributed by s-news@wubios.wustl.edu.  To unsubscribe
send e-mail to s-news-request@wubios.wustl.edu with the BODY of the
message:  unsubscribe s-news