Systematic, large-N, research on the requisites of democracy has been pursued for about 50
years. Over these years several things have been learnt. The fist major finding, made by
Lipset (1959), and confirmed by many later researchers, is that democratic advancement is
tied to economic and social development. Eventually, the list of established explanatory
factors have been extended to include ethnic fragmentation (having negative import), religion
(Muslim countries under-perform), diffusion (democracy level in neighboring counties
matters), and the incidence of oil (oil-exporting counties under-perform).
analysis, including the largest possible sample of states, these explanatory factors can explain
some 10 percent of the variance in yearly democratic changes around the world. In a longer
time perspective, these factors can explain some 60 percent of the variance (Teorell and
Hadenius 2005).
Yet, the great majority of studies have one feature
and possible deficiency
in common. It
is assumed that the all non-democratic countries are affected the same way by the economic
and social factors, which have proved to drive democratic advancement. But perhaps that is
not the case. Some forms of authoritarian government may if reality be harder to change than
others. That, in short, is the hypothesis to be tested in this paper.
This brings us to the question of how to distinguish between different forms of authoritarian
regimes. There is a growing literature on how to classify autocracies, as well as the “hybrid”
regimes located in the gray zone in between democracy and autocracy (Diamond 2002;
Levitsky and Way 2002; Schedler 2002, 2004; Linz and Stepan 1996). The authors of these
regime typologies of course claim their distinctions are important for understanding patterns
of regime change and stability. Thus, Diamond (2002, 33) argues, “As democracies differ
among themselves in significant ways and degrees, so do contemporary authoritarian regimes,
and if we are to understand the contemporary dynamics, causes, limits, and possibilities of
1
For recent broad-based testings, see Ross 2001 and Teorell and Hadenius 2005.
3
Convention