17
competitive, with a strong foundation in regional differences. Jamaica, however, has a
highly polarized two-party system based on deep ideological differences whereas, in
Trinidad and Tobago, the main cleavage is ethnic. It is worth noting, however, that if
change of more than half a government is calculated, there are a higher proportion of
countries with PR in this category than those with SMP systems. And SMP governments
are nearly three times more likely to experience no change at all than those under PR. Of
course, this data indicates the extent of change: it tells us nothing about the extent to which
it was driven by voters.
TABLE 2 ABOUT HERE
TABLES 3 AND 4 ABOUT HERE
Table 3 summarises the data in more detail and depth. Measuring change on a 0-100
index shows that, on average, it is slightly higher in SMP countries. Levels of government
support are slightly higher under PR, and shifts to and from governments are also lower.
The lower reaches of both change columns in table 3 are instructive: three low PR dismissal
countries have very low levels of voter movement for or against governments (Germany,
Italy, and Austria). These low rates of dismissal appear more closely related to voter
preferences, not electoral systems.
Table 4 takes the process further, displaying a correlation matrix. It indicates that while
PR or SMP systems are not significantly correlated with change across the cabinet change
index, both dichotomized versions of the variable do indicate the expected negative effect
of PR on government change, with full dismissal versus partial or no dismissal as having
Convention