Statistical Decision Theory

For many purposes, simply laying out all the alternatives and their consequences is sufficient to arrive at a reasonable solution. The preponderance of effects in one direction may be so overwhelming that you don't need to do anything more. In other cases, it may be difficult to decide on a course of action based on the tree skeleton and consequences alone.

In such cases it may be useful to do the analysis quantitatively by assigning probabilities to various outcomes along the decision tree. Numerical values may also be assigned to the various outcomes, and a mathematical criterion may be chosen to identify the ``best'' decision.

One of the most difficult parts of doing a formal, quantitative analysis of the decision tree is arriving at probability estimates for the various outcomes. After all, the whole reason we're even exploring this approach is that we don't have as much information as we'd like.

• This is one place where you can use some interesting techniques to elicit subjective probabilities from ``gut'' reactions. In that way, you take advantage of the intuitive ``feel'' many will have for the situation and explore their consequences. In the absence of solid data, we can even use guesses. This approach guarantees that we are at least consistent in how we use those subjective impressions.

Of course, if you have a formal population viability analysis with persistence probabilities specified over some time period, so much the better. This approach allows them to be incorporated directly.