Book contents
- Frontmatter
- Contents
- Acknowledgments
- I Introduction
- II Formal Models in Political Science
- III Empirical Evaluation of Formal Models
- 4 Fundamentals of Empirical Evaluation
- 5 Evaluating Assumptions
- 6 Evaluating Predictions: Equilibria, Disequilibria, and Multiequilibria
- 7 Evaluating Relationship Predictions
- 8 Evaluating Alternative Models
- IV A Second Revolution
- References
- Name Index
- Subject Index
6 - Evaluating Predictions: Equilibria, Disequilibria, and Multiequilibria
Published online by Cambridge University Press: 10 December 2009
- Frontmatter
- Contents
- Acknowledgments
- I Introduction
- II Formal Models in Political Science
- III Empirical Evaluation of Formal Models
- 4 Fundamentals of Empirical Evaluation
- 5 Evaluating Assumptions
- 6 Evaluating Predictions: Equilibria, Disequilibria, and Multiequilibria
- 7 Evaluating Relationship Predictions
- 8 Evaluating Alternative Models
- IV A Second Revolution
- References
- Name Index
- Subject Index
Summary
In solving formal models, we seek to find equilibrium outcomes. As mentioned in Chapter 4, the equilibrium concept a formal modeler uses often depends on the formal modeling technique. In general, equilibria are outcomes that are stable points. Depending on the modeling technique and the assumptions used, models can have unique equilibrium predictions or multiple equilibria predictions; sometimes, no equilibria are predicted. The type of equilibrium prediction a model makes is also a function of whether the model is deterministic or deterministic and stochastic. Each situation provides different opportunities and problems for the empirical assessment of the models.
Evaluating Equilibrium Point Predictions
The Paradox of Point Predictions
Point Predictions Are an Easy Empirical Target. Point predictions are predictions of a unique particular outcome, like the policy position convergence prediction of the traditional Hotelling–Downsian two-party or twocandidate spatial model of electoral competition. A point prediction is the equilibrium outcome of a model, which by definition must ignore details and have false assumptions. Of course, in the real world the ignored details exist and not all the assumptions hold. More significant for equilibrium point predictions is that we rarely are able to measure the real world at the state of “rest” that equilibrium analysis implies. This means that point predictions are less likely to be observed in less controlled empirical analyses than in controlled empirical tests. Most formal modelers expect that outcomes will likely diverge from point predictions in empirical analysis. When a formal model with a point prediction is viewed as a Complete DGP, an empirical assessment of the point prediction is highly unlikely to be supported.
- Type
- Chapter
- Information
- Methods and ModelsA Guide to the Empirical Analysis of Formal Models in Political Science, pp. 164 - 208Publisher: Cambridge University PressPrint publication year: 1999