Dear R-listers I have a split-plot experiment with three grazing regimes at the main plot level (with four replicate blocks, i.e. a total of twelve main plots), and within each main plot, I have four disturbance treatments at the split-plot level (i.e there are a total of 48 split-plots).
I have analysed plant performance by means of split-plot ANOVA using aov: e.g. model1<-aov(BE<-Grazing*Disturbance+Error(Block/Grazing) My understanding is that in particular at the split-plot level multiple comparisons a la TukeyHSD tend to be a bit more tricky, and with the above error structure aov returns an object of the class 'aovlist' rather than an object of the class 'aov', but is there any way (e.g. using package multcomp) in which I can obtain +/- accurate multiple comparisons both at the main plot and sub-plot levels? In hardly any of my models do grazing regime and disturbance treatment interact significantly. Would it be ok under such circumstances to ignore the split-plot aspect of the design and to approximate by using a model such as model2<-aov(BE<-Grazing+Disturbance+Block) (which returns an object of class 'aov') and then to use multcomp to do the following: summary(glht(model2, linfct = mcp (Grazing = "Tukey"))) and also summary(glht(model2, linfct = mcp (Disturbance = "Tukey"))) or is there a better/more accurate way to do this? Any comment on the matter would be gratefully appreciated! All the best Markus Dr. Markus Wagner Plant Ecologist NERC Centre for Ecology & Hydrology Maclean Building Crowmarsh Gifford, Wallingford Oxon OX10 8BB United Kingdom Tel. 01491 692418 E-Mail: mwag...@ceh.ac.uk<mailto:mwag...@ceh.ac.uk> -- This message (and any attachments) is for the recipient ...{{dropped:8}} _______________________________________________ R-sig-ecology mailing list R-sig-ecology@r-project.org https://stat.ethz.ch/mailman/listinfo/r-sig-ecology