LAS INQUIETUDES DE LA JUSTICIA
1.5.2. Actividad jurisprudencial y derechos humanos
Decision theory teaches how to break complex problems into man-ageable parts. Without a framework to attack difficult situations, such cases quickly become unmanageable. For example, QA can be used to help a wildcatter decide whether to drill for oil. The inherent risks of oil exploration, however, cannot be eliminated. A decision tree diagram can organize the problem’s alternatives, risks, and uncertainty.
Decision tree analysis consists of the following five steps:
1. Determine all the possible alternatives and risks associated with the situation.
2. Calculate the monetary consequences of each of the alternatives.
3. Determine the uncertainty associated with each alternative.
4. Combine the first three steps into a tree diagram.
5. Determine the best alternative and consider the nonmone-tary aspects of the problem.
Decision tree diagrams include activity forks and event forks at the junctures where alternatives are possible. For example, the deci-sion whether to drill for oil represents an activity fork in the tree for an oil wildcatter. It is symbolized on a decision tree by a square. If the different alternatives are subject to uncertainty, that is an event fork. The uncertain outcome of a well producing oil would be con-sidered an event. It is symbolized on a decision tree by a circle.
DECISION TREE EXAMPLE
As an illustration of a situation where the decision tree could be
helpful, consider Mr. Sam Houston of Texas. Mr. Houston is about to exercise his option to drill for oil on a promising parcel. Should he drill? If he hits a gusher, there is an estimated $1,000,000 to be gained. When he investigated all of the alternatives, Mr. Houston made the following list:
1. Sam paid $20,000 for the drilling option.
2. Sam could lower his risks if he hired a geologist to perform seismic testing ($50,000). That would give him a better in-dication of success and lower his risk of wasting drilling costs.
3. Should he roll the dice and incur $200,000 in drilling costs without a seismic evaluation to guide him?
4. Sam consulted with oil experts. They believe Sam’s parcel has a 60 percent chance of having oil without the benefit of any tests.
5. It has also been the experts’ experience that if seismic tests are positive for the oil, there is a 90 percent chance there is
“some” oil. And conversely, there is a 10 percent chance of failure.
6. If the seismic tests are negative, Sam could still drill but with a 10 percent chance of success and a 90 percent chance of failure.
7. Sam could decide not to drill at all.
Each piece of information above is incorporated into a tree diagram. A tree diagram graphically organizes Mr. Houston’s alternatives.
Before you get too enthusiastic over the drawing of trees, you must determine what information is irrelevant. In this case, the
$20,000 Sam paid for his drilling option is extraneous; it is a sunk cost. The money is out the door, sunk down a well. It isn’t coming back no matter what Sam decides. Sunk costs are therefore excluded from decision trees.
DRAWING A DECISION TREE
The first step to drawing the tree is to determine the first decision (or fork of the tree) that needs to be made. Should Sam choose to test first? If seismic testing is chosen, it would precede all of the other ac-tivities that follow. It is reflected in the tree as a square at the first fork.
If Sam tests, it could result in a positive event (60 percent chance) or negative event (40 percent). If there are no tests, he can still choose to drill or not (square). Regardless of the results of the seis-mic report, Sam can still “choose” to drill or not. But once the oil rig
ACTIVITY FORKS EVENT FORKS
is drilling, the existence of oil is an uncontrollable event. Either there will be a lucrative oil event or not.
The next step is to add the monetary consequences — they are like the “leaves to a tree.” If there is oil, there would be a
$1,000,000 payday. Drilling costs are $200,000 per well. Testing costs are $50,000 per well.
To know the potential financial outcomes of each decision, mul-tiply the possible dollar outcomes by their probabilities at forks where there is an “event circle.” ([$1,000,000 payday × .90 proba-bility] + [$0 payday × .10 probability] = $900,000.) This gives you the expected monetary value (EMV) of the event, although the ac-tual individual outcomes can be a range of values. At any circle, the
probabilities must add up to 100 percent (.90 + .10 = 1.00) to denote that all possibilities are accounted for. Each fork is mutually exclu-sive of other alternatives, and within that alternative the probability is 100 percent or collectively exhaustive.
At activity squares the decision maker has the ability to choose the best outcome. To determine the best alternative, subtract the ap-plicable cost from the payoff of the alternative. You calculate the monetary consequences by beginning at the far right and working your way to the left. This process is said to be “folding back” or
“pruning” the tree to arrive at your best action plan decision. At square forks you should choose the highest dollar alternative. At the circle multiply the possible payoffs by their probabilities.
The decision dictated by the tree is to throw caution to the wind and forgo the seismic tests. The expected monetary value of going ahead with testing is $370,000 (420–50), while the EMV of going ahead without tests is $400,000. You choose the highest expected monetary value (EMV). This relatively simple conceptual frame-work can be applied to new product development, real estate devel-opment, and store inventory level decisions. Whatever the decision to be considered, a decision tree structure forces the decision maker to take a comprehensive view of all the alternatives, to make an evaluation of the uncertainty (you often have to make your best guess about probabilities), and to explicitly calculate the dollar out-comes possible. The tree forces decision makers to state their as-sumptions explicitly. In this case, you may consider that the
probability of oil when there is a test result to be misstated. In that case, using a different assumption, you may get the opposite final answer. Others looking at the same situation could see it otherwise.
By comparing trees, analysts can debate specific assumptions in an organized way.
“Draw a tree and get a B” was the saying on exams involving de-cision trees. The complexity of seemingly simple problems can be seen using decision trees. Therefore, just creating an accurate tree framework was a challenge during a four-hour exam; it takes a lot of practice to become proficient.