Candidate Behavior in Presidential Nomination Campaigns: A Dynamic Model

Substantial gaps exist in our understanding of nomination politics in part because existing models do not fully reflect the dynamics of the processes involved. In this research, I develop and evaluate a model that simulates candidate behavior over time. Differences in candidates' resource allocations depend on the relative weights assigned to strategic goals and the rate at which candidates discount the future. Candidates allocate resources in proportion to each state's potential contribution to their strategic goals. The model presented here allows candidates' goals and their discount rates to vary as circumstances vary. To evaluate the explanatory power of the simulation model, allocations predicted by this model, and those predicted by statistical models were compared to observed campaign spending data. The simulation model, without estimating parameters, performs as well as the statistical models do; when a key variable is allowed to vary, the simulation predictions fit the observed data slightly better than the statistical predictions do.