**Quasi F-ratio**:
For some mixed and random models there is no exact *F*-ratio denominator
for certain tests. In such cases it may be possible to add and subtract mean
squares to construct an error mean square with the appropriate estimated
components of variation. This error mean square is then used to obtain a quasi *F*-ratio.

Doncaster, C. P. & Davey, A. J. H. (2007) *Analysis of Variance and Covariance: How to
Choose and Construct Models for the Life Sciences*. Cambridge: Cambridge
University Press.

http://www.southampton.ac.uk/~cpd/anovas/datasets/