Strip Plot Design Strip Plots n Experimental units

Strip Plot Design

Strip Plots n Experimental units for factors are row strips (row factor) and column strips (column factor) n Convenient design for two-factor agricultural experiments n Alternative to split-plot design when splitplots are not practical as experimental units for either factor

Strip Plot Alternatives n The design requires replication of both row and column strips n Just as with split plot design, we can replicate across blocks, or replicate within a single block (i. e. , plot)

Strip Plot Model I n Yandell Model

Strip Plot I Discussion n In Yandell’s model, sk indexes set , which actually functions as a block n It’s surprising that Yandell doesn’t present nested strip plot errors n Strip plot errors are really just block (or set) by treatment interactions

Strip Plot Model II n It may be more proper to consider the model as:

Strip Plot Model III n Another version of the strip plot model (consistent with Yandell’s split plot design):

Strip Plot Design I & II ANOVA

Strip Plot Design I & II ANOVA (cont)

Strip Plot Design III ANOVA

Strip Plot Design III ANOVA (cont)

Pairwise Contrasts n Main effects contrast estimates are straightforward to compute, though their standard errors vary: n E. g.

Strip Plot Design LSMEANS n Use, e. g, . /E=A*S in LSMEANS to assign correct error term in PROC GLM n Standard interaction contrast estimates are straightforward as well n We did not discuss this, but split plot interaction contrasts involving cell means could involve comparisons across subplots, or plots or both.

Strip Plot Design LSMEANS (cont) n This same problem occurs for interaction contrasts in the strip plot design n Care must be taken when estimating these contrasts n Yandell somewhat overstates severity of this problem

Strip Plot Example n Strip plot design with Variety (DPL 454 BR, DPL 555 BR, ST 5599 BR, PHY 375 WRF) as row factor and Fungicide (Ridomil, Terraclor, Quadras) as column factor n Fields are block n Nested row and column strip plots n Response is yield (cotton lint in pounds/acre)