Models, Measures and Mechanisms: An Agenda for Progress in Cleavage Research

Abstract

Cleavage research has been characterised by insufficient attention to (i) the institutional structuring of political cleavages, (ii) the operationalisation of key social and political constructs via the construction of validated measurement instruments, and (iii) the specification and testing of mechanisms through which linkages between social structure and political change can be interpreted. The contributions in this special issue are evaluated in the light of these considerations. Specific research foci and data construction/analytic strategies facilitating progress on these questions are then advocated.

Notes

Though for various reasons post-communist societies deviated from this pattern in the period following the end of communism (Need and Evans 2001).

It is more common for instruments to be accepted on the basis of ‘predictive validity’– i.e. when an instrument predicts an outcome such as vote. However, such tests are dependent upon robust theories about the relations between the construct being tested, the variables used to assess its validity, and the conditions under which a relationship is predicted – a tough test for instruments to pass in most areas of political science.

In part this commentary is engaging with a moving target as some contributors have revised their papers in response to my observations. Hopefully, I have modified my commentary sufficiently to take these into account, but there might still be discrepancies.

The comparison of pseudo R²s for non-nested models is not generally considered good practice, which might caution interpretations of Knutsen's (and Stubager's) analysis, especially given the small magnitudes and changes in magnitude being compared.