# UAV3.r, UAV experiment, Tables 19.26-27, pp751-2 # Table 19.26, p751 uav3.data = read.table("data/uav3.txt", header=T) uav3.data = within(uav3.data, {fA = factor(A); fW = factor(W); fB = factor(B) }) head(uav3.data, 3) # ANOVA: full model # Set contrast options for correct lsmeans and contrasts options(contrasts = c("contr.sum", "contr.poly")) model1 = aov(Time ~ fA + fB + fA:fB + Error(fA:fW), data=uav3.data) summary(model1) # Multiple comparisons library(lsmeans) lsmA.aov = lsmeans(model1, ~ fA) summary(contrast(lsmA.aov, method="pairwise"), infer=c(T,T), side="two-sided") lsmAB.aov = lsmeans(model1, ~ fA:fB) summary(contrast(lsmAB.aov, method="pairwise", adjust="tukey"), infer=c(T,T), side="two-sided") # Table 19.27, p752 # ReML: full model library(lmerTest) model2 = lmer(Time ~ fA + fB + fA:fB + (1|fA:fW), data=uav3.data) anova(model2) summary(model2) # Gives fA:fW variance component estimate approx 0 # Multiple comparisons # Detach then reload lsmeans, to avoid issues from its masking by lmerTest detach("package:lsmeans", unload=TRUE) library(lsmeans) lsmA2 = lsmeans(model2, ~ fA) summary(contrast(lsmA2, method="pairwise"), infer=c(T,T), side="two-sided") lsmAB.reml = lsmeans(model2, ~ fA:fB) summary(contrast(lsmAB.reml, method="pairwise", adjust="tukey"), infer=c(T,T), side="two-sided") # ANOVA: reduced model--without fW model3 = aov(Time ~ fA + fB + fA:fB, data=uav3.data) summary(model3) # Multiple comparisons lsmA.aov2 = lsmeans(model3, ~ fA) summary(contrast(lsmA.aov2, method="pairwise"), infer=c(T,T), side="two-sided") lsmAB.aov2 = lsmeans(model3, ~ fA:fB) summary(contrast(lsmAB.aov2, method="pairwise", adjust="tukey"), infer=c(T,T), side="two-sided")