between the different F-values will exist. In such cases it
is incorrect to use the point of significance given in the Ftables.
If we, let us say, find one F to be significant at the
5% level, the possibility for the other Fs to be higher
than the tabulated value, are greater and sometimes far
greater than 5%. Ottestad (1960) has calculated that for
two Fs related in that way, under the assumption that
the degrees of freedom for the denominator is 4, for the
nominator is 10, and the largest F is significant, the possibility
for the second F to be higher than the F for the
5% level in the table is 22.5% (and not 5% as it should
be). This is an error which is commonly ignored, but will
be of importance when the degrees of freedom for the
common error are small.