|
Both, opinion dynamics and innovation diffusion, describe processes on
groups of interacting agents, where an agent is represented by its local
state. Opinion dynamics considers this local state to be a mental state,
i.e. the opinion (or better described as attitude) of an agent. An
attitude describes a consistently positive or negative response towards
an object of interest. In the mainstream opinion dynamic models,
attitudes and opinions change only due to communication between agents.
The research on opinion and attitude dynamics is mostly influenced by
social psychology and currently receives little attention in purely
economic literature. Contrary, innovation diffusion receives much
attention by economists as well as by marketing scientists. Many
innovation diffusion models are tested against aggregated and partly
against individual data. Also behavioural assumptions are empirically
tested. Innovation diffusion considers the state of an agent to be a
particular behaviour of this agent. The behaviour changes because an
agent perceived other agents behaviours or because it receives new
information from elsewhere, e.g. by communication with others or by
adopting the behaviour.
Both research areas are very similar in their general structure. They both can be approached at an aggregate level but also at an individual level, and their associated methodologies share many of the instruments. Because of this similarity, one type of model can easily be transferred into the other kind of model, just by reinterpreting the local state. But this does not add any insight into the dynamics of social systems nor does this take into account the different theoretical fundaments. A theory-based comparison and combination of the two approaches seems to be interesting. For this I will focus on individual level models and briefly sketch differences in research question, differences in underlying theories and their behavioural explanations. For instance, innovation diffusion models very often emphasize information diffusion and include a decision rule based on the collected information, while a decision component is not subject to opinion dynamic models. Apart of a decision rule individual-level innovation diffusion models consist of a utility concept or another evaluation mechanisms and an updating process, which is very often a Bayesian updating. Opinion dynamic models consider the opinion or attitude as the evaluative concept and have much more emphasis on different updating mechanism; they do not only consider Bayesian updating that has its roots in information diffusion, but also include social impact theories with their relation to public opinion formation. Here we see that opinion dynamic models can refine innovation diffusion models by incorporating a specific evaluation concept, i.e. attitudes, and different update mechanisms. But there is empirical data that suggests that decision and information processing in different contexts follows different rules, e.g. framing effects may occur in purchase decisions but not in political contexts. Hence, model development for specific applications should be based on empirical data and models should be able to incorporate available data. I will sketch the landscape of opinion dynamic models as well as innovation diffusion models by presenting the basics of different models and emphasizing the original aspects of them. A focus is on models that bridge the gap between innovation diffision and opinion dynamics and on models that demonstrate the specific aspects of both research areas. |