19
complicated by a greater number of cases – 17, as opposed to 11 in Germany – and a
lower number of surveys conducted in each region – on average four; the number being
over 6 in German L¨ander. This increases cross-sectional variation, but reduces the scope
for comparing different cabinets in the same region.
3.1
Cabinet polarization, performance, and satisfaction
The previous section demonstrated greater citizen satisfaction with one-party govern-
ments in the German case, and to a lesser extent in Spain. In addition, for the L¨ander,
cabinets perceived as “tighter” received higher satisfaction scores – both absolute and
residual – than cabinets in which the constituent parties seemed further removed from
each other. This section elaborates the German finding graphically and both findings
using linear regression analysis.
Figure 1 shows a scatterplot of satisfaction with regional government performance
over the maximal ideological span of the cabinet. Given the similarity between the
two ways of representing satisfaction levels, as demonstrated above in Table 1, it is
safe to let the simple averages represent both in the graph. Placement on the vertical
axis corresponds to the average satisfaction scores for each cabinet type from Table
1; the horizontal scale represents cabinet polarization. Further, the size of each circle
corresponds to the frequency of each cabinet type.
As hypothesized, the plot shows a negative relationship between citizen satisfaction
and cabinet polarization. This is particularly pronounced in the three cabinet types that
occur only once each. The only exception to this rule is found in the grand coalitions,
Convention