R：2向Anova与ez软件包不匹配

Question

I'm using the ez package however I realized that the results sometimes does not match the anova function. I'm wondering what I'm doing wrong.

options(contrasts=c("contr.sum","contr.poly")) 
datafilename="https://docs.google.com/spreadsheets/d/1PNHSBMUdPnB_mnP_WrcxYYiwCwN8KHL2Q57KoAoWdAY/pub?gid=0&single=true&output=csv"
data.ex2=read.csv(datafilename,header=1,stringsAsFactors=FALSE) 
data.ex2                                     
aov.ex2 = aov(value~Treat*Gender,data=data.ex2  )         
summary(aov.ex2)  

### testing the ez package now 

library(ez) 

A = ezANOVA(
  data = data.ex2
  , dv = value
  , wid = ID
  , between = .(Treat,Gender)
  ,return_aov=TRUE
  , type = 3
)

print (A$ANOVA)

• when I use the example given on another website the results matches perfectly so I'm wondering if there is something I'm doing wrong with my table. The example that matches if from this dataset here.

  datafilename="http://personality-project.org/r/datasets/R.appendix2.data"

thanks!

Answer 1

You have multiple observations of ID per each grouping. This is mentioned in the warning when running your code, and it says that the values have been collapsed to their mean. So, under the hood it would be doing the following,

library(plyr)
collapsed <- ddply(data.ex2, .(ID, Treat, Gender), function(x)
    mean(x[, "value"]))

## the mean value is now in column `V1`
aov(V1 ~ Treat*Gender, data=collapsed)

Answer 2

Ok I finally figured it out. Needed one last command.

drop1(aov.ex2, .~., test="F")

now it works, for reference I got this info from here:

http://goanna.cs.rmit.edu.au/~fscholer/anova.php