Dear R Help -
I am analyzing data from an ecological experiment and am having problems
with the ANOVA functions I've tried thus far. The experiment consists of a
blocked/split-plot design, with plant biomass as the response. The following
is an overview of the treatments applied (nitrogen addition, phosphorus
addition, and seeded/not seeded) and at what level (block, main-plot, and
sub-plot):
- 6 experimental blocks divided into 2 main-plots
- each main-plot divided into 8 sub-plots, for a total of 96 sub-plots (6
blocks * 2 main-plots * 8 sub-plots)
- 16 experimental treatment conditions in a 4 x 2 x 2 factorial design:
- N at 4 levels for each sub-plot
- P at 2 levels for each sub-plot
- Seed at two levels for each block (one level per main-plot)
- response variable = biomass
Block
Main Plot 0 : No Seed
Sub-Plot 0.1 : N0 P0
Sub-Plot 0.2 : N0 P1
Sub-Plot 0.3 : N1 P0
Sub-Plot 0.4 : N1 P1
Sub-Plot 0.5 : N2 P0
Sub-Plot 0.6 : N2 P1
Sub-Plot 0.7 : N3 P0
Sub-Plot 0.8 : N3 P1
Main Plot 1 : Seed
Sub-Plot 1.1 : N0 P0
Sub-Plot 1.2 : N0 P1
Sub-Plot 1.3 : N1 P0
Sub-Plot 1.4 : N1 P1
Sub-Plot 1.5 : N2 P0
Sub-Plot 1.6 : N2 P1
Sub-Plot 1.7 : N3 P0
Sub-Plot 1.8 : N3 P1
I've tried several different approaches to run an ANOVA (lmer, aov, lme) on
this data, trying to use type III SSs and include a random factor, but am
having problems. Any suggestions?
Thank you for your help!
-Mari Pesek
--
Dept of Ecology and Evolutionary Biology
University of Kansas
Haworth Hall, 1200 Sunnyside Ave
Lawrence, KS 66045
[[alternative HTML version deleted]]
RICHARD M. HEIBERGER
2010-Nov-23 05:59 UTC
[R] factorial ANOVA for block/split-plot design
Please see the maiz example at the end of ?HH::MMC
You will need to all the way to the end of the example.
If you don't yet have HH, you can get it with
install.packages("HH")
If you need to write back to the list, please include your attempts.
Rich
[[alternative HTML version deleted]]
Hello,
To add more detail to my last post in which I asked for help regarding
a complex ANOVA, I've attached my data and have included my most
recent attempts at analysis. Although I tried to refer to the maiz
example as Rich suggested (and thank you, Rich, for suggesting it), I
didn't understand how to apply its methods to my data. If someone
could explain what's going on in that example to some degree, I would
appreciate it.
Anyway, my recent attempts are R code include the following:
m = aov(biomass ~ (N * P * seeds) + Error(1/N * P) + (seeds),
data=data), which gives me incorrect SSs but gives multiple error
terms (which I want).
m = aov(biomass ~ (N * P * seeds) + Error(1/sub.plot) + (seeds),
data=data), which gives correct SSs, but only one error term, thus
giving me incorrect F and p values.
("incorrect" is based on an analysis run in SPSS that I am trying to
replicate in R).
Thank you for your help!
-Mari
---
Dept of Ecology and Evolutionary Biology
University of Kansas
Haworth Hall, 1200 Sunnyside Ave
Lawrence, KS 66045