The Research Process - Estimating the Value of Information
The Research Process - Estimating the Value of Information
Before the research approach can be selected, it is necessary to have some estimate of the value of the information, that is, the value of obtaining answers to the research questions. Such an estimate will help determine how much, if anything, should be spent on the research.
The value will depend on the importance of the decision as noted in the research purpose, the uncertainty that surrounds it, and the influence of the research information on the decision. If the decision is highly significant in terms of investment required or in terms of its impact upon the long-run success of the organization, then information may have a high value. However, uncertainty that is meaningful to the decision also must exist for the information to have value. If the outcomes are already known with certainty or if the decision will not be affected by the research information, the information will have no value.
To illustrate and expand these concepts, consider the simplified examples in Figure 2-3. In Case A the decision to introduce a new product is shown as a decision tree. The first two branches represent the decision alternatives—to introduce the product or to decide not to introduce it. The second branch represents the uncertainty. Our descriptive model indicates that if the product is successful, a profit of $4 million will result. The indication is that there is a probability of 0.6 that the product will be successful. However, if the product is not successful, the profit would be only $1 million, an event that will occur with probability 0.4. How much should we be willing to pay for perfect information in this case? If someone could tell us, in advance, and with certainty, whether the product would be successful or not, how much would we pay for that information? The correct answer is nothing! The fact is that regardless of the information our decision would
be the same. We would introduce the product, for even if the product were not well accepted, we would still make $1 million. In this case, not only is the decision insignificant to the organization, it is nonexistent. There is only one viable alternative. Without alternatives there is no decision contest, even if uncertainty exists; therefore, there is no need for additional information.
In the second case, the estimate is that, if the product is not successful, a loss of $2.5 million will occur. Since the expectation of the new product's eventual performance is still, on balance, positive, the product would be introduced. However, in this case, perfect information now would have value. If we knew in advance that the product definitely would not be accepted, we would decide against introducing it and save $2.5 million. Since our best estimate of the probability of the product's not being accepted is 0.4, the value of the information would be 0.4 times $2.5 million, or $1 million. Thus, if this decision contest could be repeated many times, perfect information would save us $2.5 million about 40 percent of the time and would save us nothing (since it would not alter our decision) about 60 percent of the time. On average, it would save us $1 million. By spending money on research, we might improve knowledge as to how the product will be accepted. But market research is unlikely to be as good as perfect information, and therefore its value will be less than $1 million. Obviously, if the cost associated with an unsuccessful product were lower, or if the probability of an unsuccessful product were smaller, the value of information would be less. (The Appendix to this chapter extends this example to include the possibility of using a concept test to predict whether or not the
product will succeed. A method is developed to determine the value of the concept test to the decision maker.)
Before the research approach can be selected, it is necessary to have some estimate of the value of the information, that is, the value of obtaining answers to the research questions. Such an estimate will help determine how much, if anything, should be spent on the research.
The value will depend on the importance of the decision as noted in the research purpose, the uncertainty that surrounds it, and the influence of the research information on the decision. If the decision is highly significant in terms of investment required or in terms of its impact upon the long-run success of the organization, then information may have a high value. However, uncertainty that is meaningful to the decision also must exist for the information to have value. If the outcomes are already known with certainty or if the decision will not be affected by the research information, the information will have no value.
To illustrate and expand these concepts, consider the simplified examples in Figure 2-3. In Case A the decision to introduce a new product is shown as a decision tree. The first two branches represent the decision alternatives—to introduce the product or to decide not to introduce it. The second branch represents the uncertainty. Our descriptive model indicates that if the product is successful, a profit of $4 million will result. The indication is that there is a probability of 0.6 that the product will be successful. However, if the product is not successful, the profit would be only $1 million, an event that will occur with probability 0.4. How much should we be willing to pay for perfect information in this case? If someone could tell us, in advance, and with certainty, whether the product would be successful or not, how much would we pay for that information? The correct answer is nothing! The fact is that regardless of the information our decision would
be the same. We would introduce the product, for even if the product were not well accepted, we would still make $1 million. In this case, not only is the decision insignificant to the organization, it is nonexistent. There is only one viable alternative. Without alternatives there is no decision contest, even if uncertainty exists; therefore, there is no need for additional information.
In the second case, the estimate is that, if the product is not successful, a loss of $2.5 million will occur. Since the expectation of the new product's eventual performance is still, on balance, positive, the product would be introduced. However, in this case, perfect information now would have value. If we knew in advance that the product definitely would not be accepted, we would decide against introducing it and save $2.5 million. Since our best estimate of the probability of the product's not being accepted is 0.4, the value of the information would be 0.4 times $2.5 million, or $1 million. Thus, if this decision contest could be repeated many times, perfect information would save us $2.5 million about 40 percent of the time and would save us nothing (since it would not alter our decision) about 60 percent of the time. On average, it would save us $1 million. By spending money on research, we might improve knowledge as to how the product will be accepted. But market research is unlikely to be as good as perfect information, and therefore its value will be less than $1 million. Obviously, if the cost associated with an unsuccessful product were lower, or if the probability of an unsuccessful product were smaller, the value of information would be less. (The Appendix to this chapter extends this example to include the possibility of using a concept test to predict whether or not the
product will succeed. A method is developed to determine the value of the concept test to the decision maker.)
Comments
Post a Comment