The comparative study of party politics dedicates a large space party positions analyses. But with the exception of Meguid (2005, 2008) and few other pioneers, party positions have seldom been used for the understanding of party competition. The position of a party cannot be apprehended independently from the positions of the other actors within the same party system. This paper therefore intends demonstrating that political parties compete on certain public policy issues (and not on others) and that the arrival or disappearance of new competitors (new parties, party splits, etc.) together with the arising of new issues lead to a redefinition of the party competition. Based on the CAP (Comparative Agenda Project) data for Belgian parties (Flemish and French-speaking), this paper will identify different patterns of competition depending on the issue, the time frame, and the party system environment. The paper''s dataset consists in fine-grade content analyses of party manifestos of all parties in the Belgian Parliament between 1977 and 2007. First, these parties will be positioned on a large set of public policy issues and niche parties will be indentified. Second, these positions will be compared over time between parties and between linguistic communities. In this perspective, a specific attention will be put on the issues of environment, law and order, and decentralisation for mainstream parties'' manifestos. For achieving this goal, the proposed paper will use three different methods: frequentist, Bayesian, and fsQCA. The pros and cons of each of them will be assessed, and a specific attention will be put on the comparison between Bayesian “priors”, and fsQCA “directional expectations”.
