STATISTICA 3-way ANOVA nightmare!

#1
Hello everyone and thank you for your time.

First I want to say that maybe my problem is derived from the fact that I'm not an statistics expert and I'm trying to use advanced (at least for me) methods.

So, the design of my experiment is the following: we are studying the microbial stability of a certain liquid. To test that, we filtered several samples with different filters and we bottled half of the samples using vacuum and the other without vacuum. Then we analyzed the microbial content in our lab and in an external lab.

All this being said, now I'm trying to analyze the results using a Repeated measures ANOVA where I have three factors (Lab, Vacuum/Not Vacuum and Filter type (A, B, C)) and when I run the analysis the F and the p-value are not calculated so I can't know if the results are significant or not.

If instead of using the Repeated measures ANOVA I run General Linear Models and I select the interactions I want to see manually it works except if I select the interaction between the three factors (Lab*V/NV*Filter) that has the same result as the one I get using the Repeated measures ANOVA method.

What am I doing wrong? Why does this happen?
 

Miner

TS Contributor
#2
This is a guess, but it sounds like you have a saturated model. When you try to model the 3-way interaction there are no degrees of freedom left to estimate experimental error. The only correction is to either run replicates or to remove the 3-way interaction from the model.