Improper Linear Models are Better than Intuition

I was always was fascinated by people who don't care if their judgment is true. To maintain control over this process, you have to ban evaluation, and explain away errors by shifting goals at hindsight. Thats the "intuition" approach. "ci arrangiam", how Italians say and do. For the rest of us, there is the possibility to construct models and use them as predictive tools.

Improper linear models are those in which the weights of the predictor variables are obtained by some nonoptimal method; for example, they may be obtained on the basis of intuition, derived from simulating a clinical judge's predictions, or set to be equal. This article presents evidence that even such improper linear models are superior to clinical intuition when predicting a numerical criterion from numerical predictors. 

http://heatherlench.com/wp-content/uploads/2008/07/dawes2.pdf