A word more about the political side.  When the formula is chose, 
since many formulas will be about equally good at forecasting, it 
would be well to note whether the chosen one has any bad quirks that
will lead to hilarious newpaper columns or to the appearance of 
injustice for reasons that might be anticipated from its form or 
values.  One must assume that the formula will become public 
property and will be criticized, at best, by neutral and, more 
likely, by unsympathetic observers.  Consequently, if it has 
features such as that the outcome on one true-false spelling item 
is worth more than a B in two years' study of the calculus, it is 
going to be hard to explain that it wouldn't matter much which 
formula was used, however true that might be.   Consequently, among 
the many formulas that might be used, it would be well to choose 
one that is reasonably presentable and defensible if other people 
choose to make interpretations.

F. Mosteller & J. W. Tukey, 1977, pp 270-271.
<p><p><p>Quote brought to you by <a href=/~behrens> Methodological Studies Program 
</a> ASU College of Education</a>