Security Studies Program Seminar
Are Democracies as Smart as We Think They Are? Reassessing the Selection Effects Theory of Democratic Victory in War
Alexander Downes
Assistant Professor of Political Science
Duke University
November 5, 2008
- Q: Are democracies more likely to win wars, especially wars that they start?
- Follows up and evaluates Dan Reiter and Allen Stam’s work Democracies at War
- Main contributions:
- Quantitative: Reiter and Stam’s results are not robust to changes in the model or modeling approach
- Qualitative:
- Democratic leaders ought to expect they should win the wars they start for the theory to be true
- Does not hold up on further inquiry
- Why do some think democracies are more likely to win?
- Reiter and Stam (RS) have a two part theory
- Selection effects: Democracies pick wars they are more likely to win
- Electoral accountability
- Leaders need a policy success to win (Bueno de Mesquita et al.)
- Marketplace of ideas
- Democracies have higher quality of decisions from their traditions of open debates
- War-fighting: Democracies fight better in the wars they are in
- Democratic norms and culture
- Individualism+ popular government = better soldiers
- RS’ Evidence Quantitative
- Interstate war outcomes
- Variables
- Regime type
- Wins losses
- Control for capabilities, etc.
- Initiators and targets do better if they’re democracies
- Use a probit model
- Find key variables significant in predicted direction
- Existing critiques of RS
- Focus on mismatch of concepts, the theory, and the testing
- Tactical
- Democracy ≠ liberalism
- Variables don’t measure concepts
- Democracy ≠ human capital
- Strategic
- Marketplace of ideas is fundamentally flawed
- Caution is not unique to democracies
- Process tracing not supportive of Selection effects for war initiation
- Another critic that challenges the evidence is necessary
- Desch’s approach to challenge the regression
- Consider only fair fights
- Eliminate 54 of 75 wars, 83% of cases.
- Findings still weakly supportive. Probit not significant.
- However, Desch misunderstands the selection argument
- Democracies select unfair fights
- Quantitative re-analysis
- Add non-initiator or target states
- States entering war after it began are neither targets nor initiators
- Call them joiners, instead. N=30
- On initiator side but do not initiate.
- On target’s side but not attacked.
- Similar selection process
- RS omit draws
- Say democracies more likely to have a draw
- However, democracies suffer electoral consequences for protracted draws
- Truman, LBJ, Anthony Eden etc.
- Add draws, N = 233
- With draws, we have a trichotomous DV; requires new methods
- Ordered probit
- Multinomial logit
- More interaction terms from joiners
- Coefficients are no longer significant terms for regime time although the direction remains correct.
- Lack of robustness occurs in different models
- Qualitative arguments
- Should see selection criteria in democratic decision-making before war
- Optimism on success
- Factors that lead to victory in evidence
- Selecting stalemate in Vietnam
- War of choice
- US had the choice not to fight.
- US is a war joiner
- Decision 1: Bomb in Fall 1964?
- Options debated by NSC
- Status quo
- Escalate
- Graduate escalation
- Not big support for these options among the staff.
- No confidence in the SVN gov’t
- Fighting and losing better viewed that withdrawing and losing.
- Decision 2: Use of American ground troops
- Decision not based on optimism about victory.
- So, why did the US escalate?
- Dominoes
- China
- Credibility
- Good doctor theory
- Winning not necessary for the US: just the act of fighting
- Domestic politics
- Withdrawal weakens the LBJ’s Great Society.
- Summary of above argument
Notes by Andrew Radin
back to Wednesday Seminar Series, Fall 2008