The framing of decision situations : automatic goal selection and rational goal pursuit

Orthodox versions of Rational Action Theory postulate the maximization of a situation-invariant utility measure as the mechanism that underlies human behaviour. This postulate - as powerful as it may be for predicting behaviour in a multitude of applications - limits the scope of Rational Action Theory. In this dissertation, it is argued that several phenomena of natural interest to sociologists, like normative behaviour or problems of self-command, cannot be addressed adequately in the orthodox framework because they rely on cognitive automatisms that operate in a non-maximizing manner. As an alternative, a cognitive variant of Rational Action Theory is elaborated: Framing Theory. Here, maximizing behaviour - in the form of rational goal pursuit - occurs only locally, after the content of the goal is identified. This goal identification, also called the framing of the decision situation, is driven by mechanisms like mental inertia and situational triggering that are incompatible with globally maximizing behaviour. Next to the theoretical foundation of Framing Theory in the psychological and cognitive sciences, a family of stochastic models is proposed that can directly express the core notions of the theory. In this two-stage decision model, frames are conceptualized as latent classes. The determinants of framing enter the model in a first stage of class identification, while in the second stage, frame-specific rational behaviour models (e.g., the multinomial logit model) figure as measurement model for the latent classes. In a longitudinal setup, the latent class modelling is extended to a latent Markov process model. Further, the possibilities of statistical estimation of these models are discussed, and a maximum likelihood procedure is proposed for obtaining parameter estimates. The proposed methodology allows for assessing frames quantitatively, and enhances the applicability of Framing Theory in empirical research considerably. In the empirical part, Framing Theory is exemplarily applied to the social dilemma domain. This is a particularly interesting application because notably in social dilemma research, utility modelling has established itself as the standard approach among sociologists. In an experimental public goods game, the effect of a sanctioning system on individual cooperation, and the mediating role of framing, are studied. The results show that individual as well as aggregate behaviour in the experimental game confirm several diagnostic predictions of Framing Theory. Finally, in a series of three small experimental studies, the possibilities and limitations of Framing Theory as a descriptive model of human decision making are assessed. This is done by testing the predictive validity of some cognitive assumptions underlying Framing Theory. Overall, the empirical studies show that the data analytical method, as well as theoretical actor model of Framing Theory, are a fruitful contribution to sociological research and methodology.