user3185925 user3185925 - 3 months ago 16
R Question

aov won't return pvalues in R

I have a strange problem with anova summary results

summary(aov)
.

So here is the problem. I have a dataset with 6 columns. Here is the dataset sample:

Panelist Prod.ID Overall Appearance Flavor Texture
1 196 9 9 9 9
1 239 7 9 6 7
1 354 9 8 8 7
1 427 3 8 2 3
1 577 8 9 7 9
1 638 7 9 7 8
1 772 6 4 3 3
1 852 9 8 9 8
2 196 8 8 7 8
2 239 7 7 7 7
2 354 6 5 6 4
2 427 6 7 4 6
2 577 3 6 3 5
2 638 4 4 5 2
2 772 6 2 6 7
2 852 7 6 7 6
3 196 7 9 7 8
3 239 8 9 8 8
3 354 8 8 7 8
3 427 7 8 6 8
3 577 8 9 8 8
3 638 8 9 8 7
3 772 5 8 8 8
3 852 8 9 8 8


Anyway the first two columns are the factors and the rest are the response variables. The Panelist and the Prod.ID are considered by the
summary()
as continuous variables, so I converted them to be a factors with
as.factor()
.

After that conversion I ran the anova-test with following model
Overall ~ Panelist * Prod.ID
, but as summary results I got only this:

> summary(aov(Overall ~ Prod.ID * Panelist, data = paneElements))

Df Sum Sq Mean Sq
Prod.ID 7 189.6 27.085
Panelist 160 1252.9 7.830
Prod.ID:Panelist 1116 3116.1 2.792


I can't find any cause that makes the F-test values and P-values disappear.

Any help will be very appreciated.

Thanks a lot.

Answer

You have only one observation for each combination of Prod.ID and Panelist (at least in your sample data), so the number of groups is equal to the number of observations. This would cause a divide-by-zero in the F-Test, which may be the reason for the lack of reported F-Test and p-values.

For example, when I add an extra observation for Prod.ID 196 for just one level of Panelist, I get F and p values reported in the output.

Comments