Theoretical Sampling and Qualitative Empirical Model Validation

Abstract

Systems dynamics models of social processes generally have a formal core of quantitative equations. The temporal dynamics of their output may however be categorized according to qualitative typologies: depending on the model parameters, output variables may fluctuate, steadily increase or decrease, etc. Hence the present paper suggests to explore by computer simulation the relation between parameter values and the associated type of model dynamics. The result may be mapped in a multi-dimensional parameter space. Such maps of model dynamics are useful tools for systematic empirical tests, which are often more rigorous and complete than the usual checks with particular data sets. Like in theoretical sampling in the tradition of qualitative social studies, they guide the investigator to empirical observations, which are of special interest for model validation. Thus, if the mentioned observations and the related qualitative model predictions systematically coincide, the tested model has a high degree of empirical validity. The use of quantitative computer models as pilots through the space of qualitative social dynamics is illustrated by a simulation model of the mobilization for political protest. Depending on the contagiousness of the conflict and the levels of frustration and repression, the model has three possible qualitative outcomes regarding the dynamics of protest mobilization: extinction, stable positive equilibria, and regular/irregular oscillations. The analysis of the model by computer simulation allows to construct samples of particular parameter configurations, which lead to the mentioned mobilization dynamics and thus may serve for empirical model validation.