Overview: Following the theoretical literature, conditional preferences lead to significantly different predictions of the political behavior and the political outcome. The present paper discusses the merits of empirically modeling conditional preferences Abstract: The political conflict over legislative proposal is often characterized by multiple issues. During the last decade empirical testing of spatial models has flourished within the subfield of legislative studies. However, this literature assumes that actors'' preferences over two (or more) issues are unconditional (i.e. separable). The problem is that following the theoretical literature conditional preferences lead to significantly different predictions of the political behavior and the expected political outcome. While this literature seems to highlight the effect for decision over non-separable issues being separated, the conditionality in actors preferences alter the entire utility function. The present paper discusses the merits of empirically modeling conditional preferences. Specifically, we reanalyzed the 52 multidimensional EU legislative proposals contained in the DEU data set. Deploying qualitative document analysis we find conditional preference in 18 cases. Next, we categorize our cases and derive theoretical expectations over the size and the direction of the conditional preferences for each member state. Finally, we test these expectations empirically using a standard bargaining and an agenda setter model. Our findings present a strong case for adding complexity to the spatial models of legislative politics.
