The road to partisan independence: An extension and empirical test of the "running tally" approach in Latin America

Abstract

This article presents an analytical extension to Fiorina's "running tally" approach to party identification through empirical testing using Latin American data, in order to include partisan independence as a possible outcome of negative retrospective evaluations of governmental performance. Voters who evaluate government performance more negatively have a higher propensity to not identify with any political party. An instrumental probit model shows that, when controlling for the possible inverse causality between partisan independence and a negative assessment of government performance, the latter variable provides the strongest prediction of the lack of partisan identification in comparison with sociological, cultural, modernization and political-institutional variables. A theoretical explanation is offered based on the "principal-agent" model.