You should have a very good reason for doing something like this. Without knowing anything about the data, I might speculate that the items being rated are likely to affect the responses of the raters. For instance, if raters are asked about protecting endangered species, and your panel had a few environmentalists on it, we would expect those raters to differ from the rest.

Dear Jim, my data is just short stories (avg. word count 60). Each rater rated same stories on multi-criteria scale (two 7 point scales)

While it is not clear what the objective of the analysis is, in general, it is never a good idea to delete valid data. Just because ratings differ significantly is not a valid reason to discard them.

Deleting data is not recommended. Deleting data - and especially in such a small sample more than likely will bias the results. Insignificant findings are indeed a finding.