Scaling Rating or Ranking of Alternatives

Scaling, rating, or ranking of each alternative for each decision factor is the second major aspect in using the mul-ticriteria decision-making approach. Rating and ranking concepts are described in the previous subsection on importance weighting. Several techniques can be used for this evaluation of alternatives in a decision. Examples of techniques include the use of the alternative profile concept, a reference alternative, linear scaling based on the maximum change, letter or number assignments designating impact categories, evaluation guidelines, unranked paired comparison techniques, and functional curves.

Bishop et al. (1970) discuss the alternative profile concept for impact scaling. This concept is represented by a graphic presentation of the effects of each alternative relative to each decision factor. Each profile scale is expressed on a percentage basis, ranging from a negative to a positive 100%, with 100% being the maximum absolute value of the impact measure adopted for each decision factor. The impact measure represents the maximum change, either plus or minus, associated with an alternative being evaluated. If the decision factors are displayed along with the impact scale from +100% to —100%, a dotted line can be used to connect the plotted points for each alternative and thus describe its profile. The alternative profile concept is useful for visually displaying the relative impacts of a series of alternatives.

Salomon (1974) describes a scaling technique for evaluating cooling system alternatives for nuclear power given regional stream. Scaling is accomplished by quantifying the impact of each alternative relative to each environmental factor, and if the net change is less than the evaluation guideline, it is insignificant. If the net change is greater and moves the environmental factor toward its highest quality, then it is considered a beneficial impact; the reverse exists for those impacts that move the measure of the environmental factor away from its highest existing quality.

One of the most useful techniques for scaling, rating, or ranking alternatives relative to each decision factor is the unranked paired comparison technique described by Dean and Nishry (1965). This technique was mentioned earlier relative to its use for importance weighting of decision factors. Again, this technique can be used by an individual or group for the scaling, rating, or ranking of alternatives.

Functional curves, also called value functions and parameter function graphs or curves, can also be used in environmental impact studies for scaling, rating, or ranking the impacts of alternatives relative to decision factors. Figure 2.5.1 shows an example of a functional curve for species diversity (Dee et al. 1972). Dee et al. (1972) describe the following seven steps used in developing a functional curve (relationship) for an environmental parameter (decision factor):

Step 1. Obtain scientific information on the relationship between the parameter and the quality of the environment. Also, obtain experts in the field to develop the value functions. Step 2. Order the parameter scale so that the lowest value of the parameter is zero and it increases in the positive direction—no negative values.

No. species/1000 individuals

FIG. 2.5.1 Functional curve for species diversity (Dee et al. 1972).

No. species/1000 individuals

FIG. 2.5.1 Functional curve for species diversity (Dee et al. 1972).

Step 3. Divide the quality scale (0-1) into equal intervals, and express the relationship between an interval and the parameter. Continue this procedure until a curve exists.

Step 4. Average the curves over all experts in the experiment to obtain a group curve. (For parameters based solely on judgment, determine value functions by a representative population cross section.)

Step 5. Indicate to the experts estimating the value function the group curve and expected results of using the curves. Decide if a modification is needed; if needed, go to Step 3; if not, continue.

Step 6. Do Steps 1 through 5 until a curve exists for all parameters.

Step 7. Repeat the experiment with the same group or another group to increase the reliability of the functions.

0 0

Post a comment