(Page 9 of 23) - Questions of Party Nationalization authored by Morgenstern, Scott. and Swindle, Stephen.

analysis on two elections. We also display separate results for the proportional representation
(PR) and single-member district elections (SMD) for those countries that employ two-tiered
systems (Japan and Germany).

As the purpose of this table is more illustrative than analytic, we

have included information about different time periods for several cases.

-- Figure 1 --

Using the graph and the appendix data, several facts are immediately noticeable. First,

most cases fit into the nationalized box, though a significant number of cases fall in the other
quadrants as well. Second, all parties from a single country usually fit into the same box, but
there are important exceptions. For example, the U.S. location has migrated from the unbalanced
to the locally focused box over time, the UK has cases in the unbalanced and nationalized boxes,
Japan has parties in the nationalized and locally focused boxes, and there are other examples as
well. Third, by and large parties or coalitions from Europe are more nationalized on both scales
than cases from other regions. That said, the nationalized box does not encompass all or just
European parties. Most notably the UK parties score low on the static scale, as does the
Portuguese CDS/PSD. Further, the Italian Communists and Iceland’s Progressives just miss the
(arbitrary) cutoff. On the dynamic scale, the European cases are all significantly above the mid-
point of the scale (40), though the values do cover a large range (from 56.4 to 75.1). In addition
to the European cases, a number of Latin American cases fit the nationalized box, as do most of
the Japanese cases.

Different observations (decades) for the United States dominate the locally focused box,

but also found there are Canada’s Liberals, Japan’s LDP, and the PAN and PRD from Mexico.
The analysis on the United States was conducted on just those districts where both major parties
competed in each of the decade’s elections. Still, as documented by Mayhew (1974a) in terms of
the declining number of “marginal districts,” one or the other party dominates many districts to
yield the low level of static nationalization. At the same time, the high level of incumbency
advantage has decoupled national and local level politics, thus yielding low levels of dynamic
nationalization. Using a related analysis, Cox and McCubbins argue that this analysis does not
imply the absence of a national tide among the legislative elections, and they use that finding to
justify an argument about why legislators work together towards collective goals. Within this
comparative context, however, that tie among U.S. legislators seems rather weak.

The other locally focused cases are also revealing, even if not particularly surprising.

Competing in single-member districts, the Canadian Liberals’ support has also been highly
variant. The fact that Canada has a strong regional party in Quebec feeds into the analysis, but
even leaving Quebec out of the model still returns low levels of nationalization for the Canadian
Liberals. It is notable that the Conservatives, who collapsed in 1993, appear as a party with
much greater static nationalization than the Liberals. Across the country in 1988, the
Conservatives won an average of 42 percent of the vote with a standard deviation just under 12
points. By contrast, on average the Liberals won 32 percent of the vote with a standard deviation
of about 13.6 (in a range from under 10 percent to 70 percent). The coefficient of variation
(standard deviation divided by the mean) is thus considerably higher for the Liberals in that year.
With the collapse, the Conservatives dropped to an average vote of 16 points and a standard
deviation of about 8, thus increasing the coefficient of variation significantly while that
coefficient hardly moved for the Liberals. Conducting the two-year analysis, however, yields a
static value for the Liberals that is one-fifth the size of the Conservative value. As we argue in
the next section, this comparison shows the value of the multivariate analysis; by itself the high

Mexico also distributes seats through a two-tiered system, there is just one vote. The PR seats are compensatory.