How would you go about estimating the different probabilities of three alternatives occurring if you have little empirical data on which to base your estimates? Using either expected value rule, mean-variance analysis, or coefficient of variation analysis? or all?