Almost everything that we do involves a choice. In recent years there has been a growing interest in the development and application of quantitative statistical methods to study choices made by individuals with the purpose of gaining a better understanding of how choices are made and also to predict future choice responses. In many fields, the choices made by individuals will determine the effectiveness of policy. Understanding what drives people's choices and how these choices may change is critical for developing successful policy. Discrete choice modeling provides an analytical framework with which to analyze and predict how people's choices are influenced by their personal characteristics and by the different attributes of the alternatives available to them.

In an ideal situation we would build discrete choice models using information from choices that people are observed to make, i.e., revealed preference (RP) information. From these data we can quantify the influence of particular variables in the real choice context; for example, how important is price in the decision to travel by train? There are, however, potential problems with these data. There might not be enough variation of the explanatory attributes; for example little price variation across alternatives. Furthermore, several attributes might be highly correlated e.g. price and quality. But the most important of all is the fact that it is not possible to observe choices for alternatives that do not yet exist; for example new programs and technologies. In cases where the data limits the information provided by real choices it may be appropriate to collect stated preference (SP) data, which is information on preferences provided from hypothetical choice situations.

This dissertation provides several applications of discrete choice modeling using both raveled preferences and stated preference. Unlike the last two chapters which deal with the revealed preference, the first Chapter, uses stated preference data. This Chapter evaluates the impact of several attributes of monetary incentives on the decision of patients to participate in a new weight loss program. Since this program does not exist yet, revealed preference data were not available and stated preference data were collected. The attributes of interest in this study include magnitude, timing and form of payment. The goal is to see what level and what combination of these attributes provides greater impact on the reach of the program. We also account for preference heterogeneity by using a random parameter framework.

Chapter 2 discusses another application of discrete choice models in event history models (also called survival analysis). In these type of models, the main goal is to use the history of happening an event to learn more about the effect of different factors on the probability of occurrence. The event of interest in our case is the birth. We use the birth history of rural women and try to model their decision to give birth over time. The ultimate goal is to evaluate the effect of health clinics and family planning program on this decision.

The final Chapter considers the application of discrete choice modeling in an electoral framework. The 2005 presidential election in Iran is used to model the decisions of Iranian voters. Using this revealed preference data we try to learn more about the main factors evolved in both participation and in the candidate selection.