Question2 How do I randomise for a split plot

Q When should I use a Split plot design ?

The Split plot is more complex to interpret then a randomised and blocked  factorial design, which may be an alternative choice. In general it is advisable to keep the designs as simple as possible as interpretation is the most important outcome.

In a situation when you have two or more factors of interest, and one of the factors is hard to apply at the plot level, then having one factor applied to larger Main-plots can occur.

For example in a tillage and fertilizer trial it may be practically easier to apply the tillage to large Plots and the second factor (say fertilizer) can be applied to randomised plots within the Main-plot. In this case there are two stages of randomisation First randomise the treatments to main-plots within the Blocks, and then randomise the sub plots with the second factor.

It is possible to have a two factor design within the main-plots; for example  Factor 1 Tillage, (2 levels, T1, T2)

Factor 2 Nitrogen  at 3 levels (N0, N1, N2)

Factor 3   Phosphorous at 2 levels (P1, P2)

An example of R code for a split plot is given here  (2 Tillage and 3 Nitrogen ) Split Plot ramdomisation

Terms to be familiar with:    Split Plot design, Main-plot, sub-plot, two stage randomisation, practical considerations, Module 2