Problems of Subjectivity and Quantification
Decision analysis provides a structured and analytical way of making decisions, but does little to eliminate subjectivity in the process. Selecting the appropriate model, such as a decision tree, determining the available options, assigning values to outcomes, and determining the probability of occurrence for each option all depend on the subjective judgment of the decision maker. False assumptions, reliance on expectations rather than facts or what has already happened, errors in forecasts, inability to measure utility, or failure to factor in all probabilities all distorts numerical values greatly, with even minor changes in number-values multiplying to major changes in results. This makes high stake decisions, where both the scope for distortions and the implications of such distortions remain huge, as situations when not to use decision analysis.
The process requires quantification of various alternatives or options, for analysis, to make judgments. Not all alternatives or options however remain easily quantifiable, and as such, the application of decision analysis suits only normative cases. For instance, purely mathematical optimal order scheduling in a manufacturing facility or optimal hedging strategies remain easily quantifiable, and the results provable. The same is however not true of, say, providing moral education in schools, where it is not possible to measure in numeric terms the qualitative data that shows the value of moral education. In such situations, assigning values to options makes the process as subjective as taking a decision right away on one's assumptions.