Research Areas

The world is complicated. Most important decisions we face don't have simple straight forward answers. In fact, most decisions have multiple objectives to accomplish and sometimes those objectives clash. In addition, it may not be up to a single decision maker to make those decisions, their points of view and goals most likely will differ.

So how do we define a "rational decision" in this complex environment? What is the "right decision"? How do we incorporate the differing points of view and goals to make progress and make decisions? How can we (or even should we) adapt classical decision methodologies when the underlying assumptions that they rest on no longer exist?

Examining these issues and methods to accommodate them is the focus of this area of research. Specific areas being examined at this time include Subjective Rationality of Groups and Adaptive Belief Management.

Uncertainty Management

Risk and decision making, in general, are based on perceptive beliefs or measures of uncertainty. As situations increase in complexity, the need to manage these rapidly changing and dynamic uncertainties and beliefs becomes more and more critical to key decision makers.

Gone are the days of choosing a static distribution to describe your state of uncertainty; or of a single probability to quantify your belief. New information comes in constantly, and new knowledge is distilled from it at higher and higher levels, resulting in the need for a formalized management of these dynamically changing beliefs and the impacts they have on critical business decisions.

Specific areas being examined at this time include Expert/Consulting Valuation and Adaptive Belief Management.

State Based Reconstructabilty Analysis

Continuing on the work of B.Jones and of Johnson and Zwick in the area of State Based Reconstructabilty Analysis, Ike's current research in this area is looking at the underlying relationships and nature of noise in the signal that lend themselves more viable for this type of analysis. In addition, he is looking at using the technique to mine text for predictive purposes of continuous valued response variables.

Data Envelope Analysis [DEA] is an analytical technique to evaluate peer performance and establish internal benchmarking by looking at efficiency of decision making units [DMU] by examining the efficient utilization of resources.

However, DEA typically compares an inefficient DMU by showing one (or more) efficient ones that the inefficient one should be emulating. The issue in DEA that is the current area of Ike's research is what to do in the "zero-sum" situations where the production outputs or resource inputs are shared among DMUs.

For example if the output is market share, the only way for one DMU to improve is for one or more to degrade. So in these situations what is the true benchmark and does the methodology need to be modified to accommodate for shared resource constraints?

Classic conflict based game theory [Van Neumann, Morgenstern, Nash, etc] requires one to assume the opponent is a rational being with as least as much information and access to same methods of analysis as you do, and that they will chose the rationally best option. However, many times people do not act rationally and some times the irrationality is done for deceptive purposes. Knowing that people (and groups) are not required to act rationally and that one may have a sense of deception in their opponents behavior, how are they to respond? Do they need to (or should they?) adapt their strategies to accommodate for the irrationality, the deception, or both - for either protection or exploitation?