Decision trees are attractive. They offer a straight forward way of writing down the various available alternatives, and choosing among them. But here are some questions that always need answering.
1. Can the decision really be reduced to a set of discrete alternatives? Are there some factors that can't be listed and quantified, such as beauty and morality?
2. Where do the alternatives come from? Are they exhaustive; that is, have we discovered and listed all of them? Do they reflect reality, or only the biases and preconceptions of the decision maker?
3. How confident are we in the outcome values, aka the payoffs? Are they time-sensitive?
4. How confident are we in the probabilities attached to the outcomes? Where on earth do they come from?
5. In short: Is a decision tree really useful in this situation, or is it just a way of camouflaging a wild guess, and making it look "rational?"
Expectations
Choose an organization with which you are familiar. Describe an important decision the organization has made, or a type of decision it routinely makes. Can a decision tree be used to make that decision? If so, how? If not, why not?