Next, for each criterion, they rated the remaining alternatives between 0 and 100. Remember that a utility is an assessment of how much a certain alternative on one criterion is "worth" to the agency. It is very important to understand that this ranking is often not linear, this can be true even if the underlying measurements are linear. E.g. the criteria "increased revenue generation" is measured by "3 x agency spending" at the modest investment, "4 x agency spending" at the moderate level of investment, and "5 x agency spending" at the elaborate level. This might be interpreted as a linear variable, however, the funding sources for the agency might have set a minimum goal of 4 times agency spending. This means that achieving the modest goal of "3 x agency spending" is of little or no value. Whereas the difference between four times and five times is negligible. This might lead the weighting of the "increased revenue generation" criteria to be: 0, 30, 95, and 100. The procedure for finding the weights is explained on page 21.

Figure 17. Complete MAU model, OBT (numbers rounded)

			Alternatives
Rank	Weight	Criteria	No Investment	Modest	Moderate	Elaborate
2	.27	Cost of developing WWW-service	100	90	70	0
			27	24	19	0
4	.06	Expanded comm. Capabilities	0	33	67	100
			0	2	4	6
1	.52	Increased revenue generation	0	30	95	100
			0	16	49	52
3	.15	Better comp. w/ other states	0	33	67	100
			0	5	10	15
		Total Utility:	27	47	82	73

Notice that the second criterion in the "Performance variables, measures, and targets" worksheet was dropped from this analysis. This is because it did not change over the three investment alternatives, and thus would not have changed the model. With the MAU model almost complete, the group from OBT had to compare the importance of the different criteria. This was done by comparing every criterion to the others to find the most important ones. Increased revenue generation, was given a 1 in the "rank" column, and then compared to every other remaining criteria. This led to the "cost of developing WWW-service" getting the second highest ranking. This criterion was then compared with the two remaining criteria, resulting in the "better competition with other states" ending up as rank number three, and the last criterion getting the last rank.

With the ranking done, weights needed to be assigned to the criteria. Knowing the rank made it a little easier for the OBT group, because they knew that the criterion ranked as number one had to have a higher weight than the criterion ranked as number two, and the criterion ranked as number two had to have a higher weight than the third ranked criterion, etc. When there are relatively few performance criteria (i.e., 3 - 5), weights may be determined relatively easily by an expert. When the number of performance criterion is larger, it may be very difficult indeed to determine how to weight the cost attribute relative to all the other performance criteria so that the trade-off between cost and benefits is appropriate. An expert in MAU modeling can assist in this process.

Rather than assigning weights, it may be simpler and more efficient to just rank the criteria in order of importance. This approach has been developed in F. Hutton Barron and Bruce E. Barrett, "Decision Quality Using Ranked Attribute Weights," Management Science, 1996, 42(11), 1515-1523. In their process, each of the performance criteria are ranked from most important to least important. Once that ranking has been established, weights are assigned according to the following table:

Figure 18. MAU-Model Weighting