0

I'm trying to conduct a 3 way Anova in R, using the WRS2 package. My data is heteroskedastic so i need to do a robust version e.g. trimmed means. I have my data arranged in long form (csv with 4 columns - 3 factors and 1 numeral). My input looks like this: 

t3way(happiness ~ money*job*relationship, data = Dataset)

I get the following error: "Incomplete design! It needs to be full factorial!"

Thank you in advance!

Kevin Sohn
  • 1
  • 1
  • 2
    Probably one of your variables is not factorial. Provide an example of the data. – Al14 Apr 20 '18 at 19:44
  • i tried a standard aov with the same input and it computes. Data looks like this: `happiness: continuous values from 0 to 3 money: 8 categories 1 to 8 job: 5 categories relationship 4 categories: 0,1,2,3` – Kevin Sohn Apr 20 '18 at 19:58
  • a small sample can be found here: [link](https://1drv.ms/x/s!AiejnoNjAUvPhVgdwr9-wXxhecqY) – Kevin Sohn Apr 20 '18 at 20:20
  • ok i guess not every factor combination is avaliable in my data. so what are my options for a robust 3 way anova? – Kevin Sohn Apr 20 '18 at 22:24

1 Answers1

2

I solved the same problem by re-factoring the variable with empty levels. Dataset$myvar <- factor(Dataset$myvar) Thus only the levels having data remain. Then the t3way worked fine.

E Raufaste
  • 21
  • 2