A Reassessment of Presidential Campaign Strategy Formation and Candidate Resource Allocation
This paper finds that, contrary to Shaw (1999), the factors thought to predict presidential campaign strategy and its effect on resource allocation are not significant when appropriate statistical methods are used.
Why It Matters
The publication begins with a motivating question: Do the factors identified by Shaw (1999) predict presidential campaign strategy formation and independently affect candidate resource allocation when analyzed with the appropriate statistical methods?
Its central contribution is to show that this paper finds that, contrary to Shaw (1999), the factors thought to predict presidential campaign strategy and its effect on resource allocation are not significant when appropriate statistical methods are used.
It matters because the findings connect institutional choices to the way authority, public responsibility, and political behavior are experienced in practice.
Key Findings
- Shaw's (1999) analyses used ordinary least squares (LS) rather than the claimed ordered probit and 2SLS methods.
- When the correct ordered probit and 2SLS methods are applied, the factors (competitiveness, electoral votes, TV ad cost) do not significantly predict campaign strategy.
- There is no evidence that campaign strategy, as measured, independently affects candidate resource allocation using the methods and data described.
- The results from the correct methods are characterized by a high degree of uncertainty.
Research Design
- Design
- Article
- Data
- Data provided by Daron Shaw (1999), including campaign and journalistic sources on state-level campaign strategy, candidate appearances, and television advertisement buys.
- Geography
- United States (presidential elections)
- Time Period
- 1988-1996
- Unit of Analysis
- State-level (U.S. states in presidential elections)
- Methods
- Replication of Shaw (1999)'s analyses using LS, ordered probit, and 2SLS regressions.; Comparison of published results to results from correct statistical methods.; Simulation and graphical analysis (ternary plots) to illustrate findings.
Full Abstract
This paper reassesses Daron Shaw's (1999) claims regarding the formation of presidential campaign strategies and their effect on candidate resource allocation. The authors demonstrate that Shaw's analyses, which were claimed to use ordered probit and two-stage least squares (2SLS) regressions, actually used ordinary least squares (LS) regressions. When the correct methods are implemented, the key substantive conclusions of Shaw (1999) do not hold: the factors attributed to the formation of electoral college strategy are insignificant, and the independent effect of these strategies on campaign resource allocation cannot be ascertained from the methods and data used.
Citation
Typescript.
- Venue
- Typescript