Partisanship and Question-Wording Effects: Experimental Evidence from Latin America

Abstract

The existing literature suggests that partisanship in Latin America is relatively weak. However, these findings have been based largely on a survey methodology that systematically underestimates partisanship. This study provides caution about measuring party identification when it is framed in a short-term time horizon and includes a filter question—as most comparative surveys do—since this tends to result in an underestimation of levels of partisanship in the region. In turn, surveys that rely on a question wording that is more consistent with early theories of voting behavior show that the proportion of voters who self-identify with a political party is larger than the literature assumed. The findings of this paper have broader implications for studies in comparative politics. They suggest that the existing literature on parties and party systems has underestimated the ability of political parties in Latin America to recruit and create strong linkages with the electorate.