I conducted an experiment with three different treatments and want to see if there are significant differences in the survival of tadpoles in each treatment and if there were significant block effects. I performed an Anova using the package 'car' and find that there were no block effects, however there were significant differences in survival between treatments. When I attempt to perform the Ryan-Einot-Gabriel-Welsch Post Hoc test i get the following error code:

Error in if (Tprob[ntr - 3] > Tprob[ntr - 2]) Tprob[ntr - 2] <- Tprob[ntr - : argument is of length zero In addition: Warning message: In qtukey(p = (1 - alpha)^(i/ntr), i, df = DFerror) : NaNs produced

I get the same error if I drop the block effects or make a new data file with only the treatment and tadpole survival percentages.

setwd("C:/Users/garig/Desktop/TubDataAnalysis")
HylaSur<-read.csv(file="Tadpole_Survival_Nocray.csv", row.names='Tank_ID', sep=",") #data file
mod2<-lm(Mean_Hyla_Survival ~  Block + Treatment, data=HylaSur, na.action = na.omit)
Results_mod2<-Anova(mod2, type=3)
Results_mod2
shapiro.test(resid(mod2))
Posthoc<-REGW.test(mod2, "Treatment" ) #Ryan-Einot-Gabriel-Welsch

Error in if (Tprob[ntr - 3] > Tprob[ntr - 2]) Tprob[ntr - 2] <- Tprob[ntr -  : 
  argument is of length zero
In addition: Warning message:
In qtukey(p = (1 - alpha)^(i/ntr), i, df = DFerror) : NaNs produced

0 Answers