Soft Computing Evaluation Logic

1.3Drawbacks of Simple Additive and Multiplicative Scoring and Utility Models

Everything should be made as simple as possible, but not simpler.

—A. Einstein

Simplicity is a good thing only for those who know and never cross the border line between simplicity and oversimplification. Unfortunately, it is in human nature to trade accuracy for simplicity. Simplicity at any cost is incompatible with professional problem solving. Simple scoring techniques are the area of decision making where accuracy is traded for simplicity. More often than not, the simple scoring models yield oversimplifications and questionable results. This is the reason why simple scoring and utility models are generally not acceptable in professional evaluation. In this chapter, our goal is to explain and prove this claim.

The simple scoring methods are used to score individual components of complex systems, and then to evaluate and compare competitive systems using a weighted mean of component scores [KLE78, KLE80]. The most important such approach is the simple additive scoring (SAS, also called simple additive weighting, SAW [TZE11], or weighted linear combination, WLC [MAL99]) that has been used for many years. SAS models for evaluation (e.g., [SCH69, SCH70, MIL66, MIL70, YOO95, TRI00, PAL02]) assume that the quality of an object is a weighted sum of the quality indicators of its components/attributes, as follows:

In a similar way, simple multiplicative scoring (SMS) (e.g., [WHI63, CHE92, CHA01,