预期的样本信息——海盗面板.docVIP

Back Expected Value of Sample Information D: In the last lecture, we developed a rough estimate of the value of information (EVPI) to answer our boss’s question as to whether or not we wanted to buy some market research data. We used a quick calculation because s/he was on the phone and wanted an immediate response. After hanging up, you went on to other projects, abandoning the Expand/Build/Subcontract question until the new information came in. Then your phone rings again. It’s your boss, saying that s/he got cutoff from the market research company (they had been on hold) and before calling them back, wanted to know if you were sure you wanted the information. Since you now seem to have a little more time, you ask your boss to give you 30 minutes, and you will have a more definite answer. Q: How will we get a better answer to whether or not we should buy the information? A: By using the Expected Value of Sample Information. D: We begin by revising our decision tree. Q: Why do we have to revise the decision tree? A: The situation the tree describes has changed. D: When we drew the tree, we started by making lists of the decision-alternative group and the state-of-nature group. We then decided which came first and then which came next for each branch. We now have more groups to consider, as shown in Figure 1: Figure 1: Decision Alternative and State-of-Nature Groups The first two, of course, are just the ones we had to begin with. The new decision is whether or not we should buy the data, and the new states-of-nature are the two possible survey results. Remember that when buying information, you do not what it says, Also, the data is second-hand, so it won’t tell us specifically what our sales will be, only if the future market is “Favorable” or “Unfavorable.” So, once again we are asking: Q: What comes first? A: The decision on whether or not to buy the market research data. D: This is a change from last time, but we