18
It is further necessary to evaluate these results in light of the other control
variables. Several of the results for the controls performed as expected. In terms of
candidate characteristics, incumbent status for the LDP positively shaped their vote
shares. However, the previous number of terms LDP politicians have won office was
only positively correlated with vote shares in Model I. Also, it appears that the number of
support groups established or maintained by LDP politicians has no direct bearing on
their share of district votes.
For the district characteristics, the measure for the percentage of district that is
densely inhabited has a negative and statistically significant impact on LDP vote shares.
18
In terms of the level of competition in the district, the larger the distance between the
LDP candidate and the second-place finisher indicates less competition, which may have
allowed LDP candidates to capture larger vote share. Finally, the number of candidates
competing in the district has a negative and statistically significant effect in all of the
models. In districts with more candidates, the LDP was not able to secure as large of a
vote share in contrast to districts with fewer candidates, which were also likely to be less
competitive.
Having detailed the results for the LDP, I turn to the results for the DPJ in Table 3.
The results of Model I show the effects of own and rival campaign-period spending on
DPJ vote shares. Own campaign period expenditures has a small and statistically effect of
8% of the district vote share for an increase in spending of ¥100 ($.94) per elector.
Interestingly, the effects of own campaign-period spending for the DPJ is parallel to the
18
This suggests there are fewer votes to capture in more metropolitan districts, an effect that is likely
correlated with the number of candidates competing in the district and the level of competition.
Convention