This paper challenges the notion that voting games with purely instrumental players cannot account for high turnout (the ‘turnout paradox'). Although it has been known for over 25 years that such games can generate high-turnout equilibria, the said equilibria have been rejected on the grounds that they are fragile. This paper shows that this claim is incorrect because it is based on a computation of pivot probabilities that is not consistent with equilibrium analysis. Once the relevant computations are corrected, it becomes possible to obtain upper bounds on the cost of voting compatible with high-turnout equilibria showing that these equilibria are indeed quite robust. This paper concludes by extending the model to include continuous outcome functions, showing that while they preserve the high-turnout equilibria obtained before, they allow for better characterizations of the results.
