[Joe Ross]

Since both the weights and the scores are subjective, it's important to look more closely if you have two or more choices that are close after the initial scoring. One way to do this is by performing a sensitivity analysis.

This is performed by changing the weights and scores by small amounts individually, and seeing how that affects the rank order. For example, if there's one criteria with a high weight and one alternative with a high score in that criteria, that might be the sole driver why that alternative is higher then another. If you made a mistake in either the weight or the score, you might be selecting a sub-optimal solution. There's a lot of fancy things you can do analytically (monte-carlo, treat each weight or score as a variable and see what value makes the result change, etc).

There is a quick and dirty way to do this also. I'm going to use BJC's spreadsheet example. Individually, set each weight to 0, and look at how things are ordered. For example, if you set the weight of cost to 0, you'll see that the order of the results do not change at all. This means that in this study, cost isn't a discriminator. If you do the same thing with the maneuverability weight, you'll find that 6wd wins big. Thus, maneuverability is important to consider further. The same thing happens with the traction weights, with those set to 0, Mecanum wins.

Another thing to look for is whether any scores are the same across the board. That would be an indication that a certain criteria isn't a discriminator and could be removed to simplify the study without much consequence. In this case, virtually the same scores are given to both the hard to be pushed and the pushing others criteria. They could be combined to a traction criteria with a weight of 15. This makes it easier to decide if traction is as important maneuverability. In some cases, this double representation might cause something to be over-weighted compared to the others.

Thanks for posting the spreadsheet.