1. Science - Variables that have no basis in the (heuristic model of the) model, used to fit the model's behaviour to whatever it is supposed to model.
Example: If you want to model the fall
time of a
stone in the atmosphere, you start with the law of gravity and find it does not give the right fall
time, so you first add air friction as a variable.
So far so good, but as you will soon find out that for a particular
stone, the friction factor does not only depend on dimensions of the stone. It may also depend on surface roughness, air pressure at the
time,
water content in the air, iron content of the stone etc.
Despairing to model all these, you add an unexplained variable that you can use to adjust the outcome of the modelling to (aka "calibrating" or "
fine-tuning" the model) the observed behaviour.
2. Business - The same as the above, but then applied to economic models and scorecards. Also known in the latter case as 'Management Adjustment'.
1. I
don't like experimental
physics: too many fudge factors. Where is the predictive value in that?
2. We were pretty much on
target last year until the wankers upstairs decided to apply a management adjustment.