Puzzling Out Genuine Political Party Support in Sub-Saharan Africa

• In Event: Contrasting Expected and Actual Political Behavior in Africa and the Middle East

Abstract

As countries develop into full-fledged democracies, analysts become increasingly confident in observing the accurate levels of public support for political parties among the adult population. Inferences are made through national election results and representative public opinion surveys. In countries with hybrid or authoritarian regimes, however, both sets of indicators are biased in complex ways through the restrictions on free and fair competition and electoral manipulation as well as self-monitoring and preference falsification. This paper seeks to establish a more accurate depiction of political party popularity through adjusting and comparing these two data sources. The analysis is presented in two steps. First, Afrobarometer party support figures are readjusted based on a battery of self-censorship variables using a multinomial model. This results in moderately higher support for opposition parties compared to the raw data, though a significant number of non-responses, including “refused” answers, remain. Second, these readjusted polling figures are then compared to actual returns from the closest election for the relevant parties. Critically, how the readjusted Afrobarometer polling figures relate to election results depends on the status of the party (incumbent, main opposition, or smaller opposition) interacted with the integrity of the election as measured through V-Dem’s Clean Elections Index. After controlling for democratic indicators, non-responses are compared to the electoral returns and reallocated to generate final figures for each party covered in the Afrobarometer surveys.

Author

• Paul Friesen, University of Notre Dame