Karar teorisi üzerine makaleler

This thesis consists of three different papers on the subject of decision theory. Economic theory tends to attract attention on binary relations existing between the objects of choice. Any rationalization is based upon a so called "Rationale" that solely depends on the binary relation over objects; however the unobservable concepts that are related to the choice set themselves may affect the choice process. In the first paper, we develop a model, where the decision maker has a concept set associated with each element in the choice set. Accordingly, she has a hierarchy of the concepts in her mind and orders the elements of the choice set with respect to each concept. Then she chooses the top element with respect to the most important concept. We will derive the empirical characterization and explain the scope of the procedure. The second paper suggests a novel way to characterize a certain class of choice procedures which are defined by various conditions put on the consideration parameter. By showing the equality between the consideration parameter and the well-known hazard rate h, we will show that the machinery provided through the hazard rate h will characterize some of the famous stochastic choice procedures based on the consideration parameter directly. In addition, a new stochastic choice procedure based on a similarity relation is introduced and characterized. The last paper connects two empirically observed facts through a new theory. The randomness of the choices people make and their lack of considering the full set of available alternatives are both well-known and empirically supported facts. We present a "theory of selves" approach to build a firm connection between these two observations by assuming that each decision maker (DM) has multiple deterministic consideration filters where she maximizes her well-defined preferences on the consideration sets resulting from these filters. Each consideration filter corresponds to a different self that arises with some probability and thus the choice becomes random. We characterize this "Random Filtering" procedure using some of the influential consideration filter forms from the literature. Several well-known choice anomalies and context effects can be captured by the model.