Khorramshahgol et al.
(1988) proposed an improved methodology for goal programming in project evaluation and selection. In their view, a project is
a risky investment, and decision-makers can limit risk by setting
an upper bound for the project's estimated cost coefficient of
variation σ(Ci)/E(Ci), where E(Ci) and σ(Ci) are the i-th project's
expected cost and its cost standard deviation, respectively. Thus,
the coefficient of variation is widely used as a quantitative
measure for risk in financial investments (Levy and Sarnat (1995)
present a good example).