Data Quality in Content Analysis. The Case of the Comparative Manifestos Project

Abstract

While textbooks offer numerous devices for enhancing and testing the data quality of content analysis, all tools must be tailored in line with the contexts of the text and the analytical concepts of research. This is particularly the case in a long-term project such as ours that has continued for three dec ades to code election programs of all significant parliamentary parties in old and new representative democracies since World-War II for the purpose of measuring policy preferences of political parties. This article starts with a dis cussion of the strengths and weaknesses of the two basic types of quantitative approaches - human-based and computer-based content analysis. The basic features of our classical human-based approach for estimating parties' policy preferences are outlined by reference to Krippendorfs (2006) typologies of reliability and validity. The conclusions highlight implications of the contexts of manifestos and the concepts applied to them for providing high quality