In Chapter 1, I suggested that the findings of the KABL model may be understood in terms of an approach that emphasizes the causal role of policy or ideological differences within governments. From this perspective, a large extremist presence in a parliament, as measured by the polarization variable, affects government survival primarily because it raises the proportion of prosystem seats that would have to be represented in any majority government – and hence, on average, the government's ideological diversity. By the same token, the number of failed formation attempts, although barely significant in the KABL model, could be seen as an indirect – and therefore weak – indicator of the existence of serious ideological divisions among potential coalition partners. Even the role of party-system size (frationalization) may be mediated by ideological diversity: in a fragmented party system, the number of member parties required for a government to command a majority tends to be greater, and with greater numbers of government members may come greater ideological diversity.

The model developed in the preceding chapter moves the analysis a couple of steps further in this direction by eliminating the role of formation attempts and, more important, by replacing effective party-system size with effective government size, a change that is directly supportive of the interpretation. Nevertheless, the interpretation's viability cannot be established convincingly unless the ideological diversity within governments is introduced directly into the analysis.